Although many factors may be involved in how monarch butterflies choose an aggregation site, a simple test of monarch habitat preference is to observe where they aggregate and note the tree composition of those sites. The trees present at a chosen aggregation site inform what works at aggregation sites. Additionally, if native tree habitats are present in the vicinity of aggregation sites, we could assess whether those native tree habitats are used preferentially to non -native tree Meade, Griffiths, and van der Heide
Monarch Butterfly Habitat on the Razor’s Edge:
Native and Non-native Vegetation in California and Occam’s Razor; February 2023 4 habitats. Our immediate goal must be to recover and enhance prot ection for monarchs at overwintering sites to the best of our ability given what we have learned about the characteristics of overwintering sites during the peak of the western monarch population.
2 SANTA BARBARA COUNTY HABITATS
2.1 Background
In Santa Barbara County there are 114 locations named as monarch butterfly aggregation sites by the Xerces Society (Xerces Society 2023). Another source, the 2018 report Monarch Butterfly Overwintering Sites, Santa Barbara County, California (Meade, Griffiths, van der Heide, and Villablanca 2018) documented 117 monarch butterfly aggregation sites, and inspected a total of 130 potential or known aggregation sites in Santa Barbara County. This study relied on previous work (Meade 1999, Calvert 1991) and the Xerces Society Monarch Overwintering Sites database for site locations along with additional potential sites added by the study investigators. These 130 sites are grouped into 5 categories based on what tree species the monarchs roost on at each site , listed in Table 1 Of the 130 sites investigated, 96 are dominated by eucalyptus (2 of which are eucalyptus and non -natives), 10 are mixed eucalyptus and California native trees, 9 sites are eucalyptus with local California native trees, 9 sites are California native trees, and 6 sites are local California native trees (only trees native to Santa Barbara County). Eucalyptus trees are either the primary vegetation, or a significant contributor to the creation of monarch butterfly aggregation sites in Santa Barbara County in 88% percent of known or potential aggregation sites.
TABLE 1. NUMBER OF AGGREGATION SITES BASED ON TREE SPECIES-COMPOSITION
CATEGORIES
Category
Number
Tree Composition Label Tree Species Count
1 Eucalyptus and other non-native trees Blue gum, red iron bark, silver wattle, evergreen ash
97 2 Eucalyptus and California native trees Monterey cypress, Monterey pine, coast redwood
9 3 Eucalyptus and Santa Barbara native
trees
Eucalyptus trees and coast live oak,
western sycamore, arroyo willow,
California bay laurel
9 4 California native trees Monterey cypress, Monterey pine, coast redwood
9 5 Santa Barbara native trees Coast live oak, western sycamore, arroyo willow, California bay laurel
6 Total number of aggregation sites 130
Meade, Griffiths, and van der Heide
Monarch Butterfly Habitat on the Razor’s Edge:
Native and Non-native Vegetation in California and Occam’s Razor; February 2023 5
2.2 Data Analysis
Analysis of data from the Santa Barbara County study of monarch abundance across these 5 tree categories by month shows that 92 -99% of the monarch population in Santa Barbara County aggregated at sites with eucalyptus compared to sites with only native vegetation (Figure 1). About 8% of the monarch population roosted at sites with native trees (combining the California and local native categories) in the early overwintering season and this number decreased to less than 1% in the later months of the overwintering s eason. Sites with only Santa Barbara County local native trees were utilized by 1% of the monarch population at peak use (data from Meade et al. 2018).
This result raises the question of whether monarchs have a choice between eucalyptus and native trees f or overwintering aggregations. If eucalyptus and native trees are equally available, monarchs may be preferentially choosing to aggregate in eucalyptus. However, if eucalyptus is more abundant and therefore more available than suitable native tree habita t, the monarchs may be aggregating in eucalyptus because they do not have a choice.
This data on a logarithmic scale presented in Error! Reference source not found. shows more detail into the behavior of the monarchs through the season. Like the stacked bar graph in Figure 1, the logarithm scaled data in Error! Reference source not found. shows the monarchs use sites in the eucalyptus only group (Category 1) more than the other groups. The logarithm scale graph shows the monarchs used the categories 3 and 4 almost identically and they were presen t at sites in category 2 slightly more than categories 3 and 4 in the winter months of December and January.
Sites with only species native to Santa Barbara County were utilized by the fewest monarchs in the fall months with a steep decline of utilization in January when the monarchs abandoned those sites for the season.
Meade, Griffiths, and van der Heide
Monarch Butterfly Habitat on the Razor’s Edge:
Native and Non-native Vegetation in California and Occam’s Razor; February 2023 6 FIGURE 1. MONARCH BUTTERFLY USE OF SITES IN 5 TREE COMPOSITION CATEGORIES IN SANTA
BARBARA COUNTY BY MONTH IN BAR GRAPH
Meade, Griffiths, and van der Heide
Monarch Butterfly Habitat on the Razor’s Edge:
Native and Non-native Vegetation in California and Occam’s Razor; February 2023 7 FIGURE 2. MONARCH BUTTERFLY USE OF SITES IN FIVE TREE COMPOSITION CATEGORIES IN
SANTA BARBARA COUNTY BY MONTH IN LOGARITHMS
2.3 Tree and Habitat Choices
To explore this idea of tree choice by monarch butterflies, we examined locations with native trees along the south coast of Santa Barbara County. Criteria for consideration include well-established native trees, locations near the coast, and sites of similar elevation within one mile of known aggregations. At least 17 locations of these native riparian habitats are in proximity to known eucalyptus dominated mon arch aggregation sites, yet no records of aggregating monarch butterflies occur for these native areas (Table 2). These native riparian habitats are within ½ mile or less of known monarch aggregation sites with a range of nearly adjacent (e.g. Rincon Creek) to a distance of less than ½ mile (e.g. Garrapata Creek). Choice is available for monarch butterflies between eucalyptus tree stands and native riparian vegetation habitats.
Meade, Griffiths, and van der Heide
Monarch Butterfly Habitat on the Razor’s Edge:
Native and Non-native Vegetation in California and Occam’s Razor; February 2023 8 TABLE 2. SANTA BARBARA COUNTY SOUTH COAST LOCATIONS WITH ABUNDANT NATIVE
VEGETATION, NO ROOSTING MONARCHS, AND NEAREST KNOWN MONARCH BUTTERFLY
AGGREGATIONS
Location Native vegetation Nearest monarch aggregation (Name and Site Number)**
Distance to
native tree
habitat (ft)
Rincon Creek willow, cottonwood,
sycamore, oak Rincon Canyon, Site 118 200
Rincon Creek willow, cottonwood,
sycamore, oak 6765 Rincon Road, Site 102 1,200
Carpinteria Creek sycamore, willow,
cottonwood Carpinteria Creek*, Site 99 400
Arroyo Paredon sycamore, willow, oak Padaro Lane sites 900 Garapata Creek sycamore, coast live oak
riparian forest Lambert Road, Site 95 2,100
Toro Canyon Creek oak, sycamore Lambert Road, Site 95 1,100 Romero Creek sycamore, oak, willow Ortega Hill, Site 92 650 San Ysidro Creek sycamore, willow, oak Crane School, Site 90 600 Arroyo Burro Creek willow, sycamore Arroyo Burro Canyon, Site 82 900 Atascadero Creek sycamore, willow Atascadero Creek, Site 78 (historic location) 250
San Jose Creek willow, cottonwood,
sycamore Elks Grove*, Site 76 1,500
Bell Canyon sycamore, willow,
cottonwood, oak Bell Canyon, Site 58 1,000
Tecolote Canyon willow, oak Tocolote Canyon, Site 57 1,400 Gato Canyon sycamore, oak Arroyo west of Gato Canyon, Site 53 1,600
El Capitan Canyon sycamore, oak, pine,
cottonwood El Capitan State Park, Site 52 750
Honda Canyon sycamore, willow,
cottonwood Arroyo Hondo, Site 110 100
Gaviota Creek willow, cottonwood Gaviota Beach Road, Site 109 250
oak = coast live oak (Quercus agrifolia)
willow = arroyo willow (Salix lasiolepsis)
cottonwood = Fremont cottonwood (Populus fremontii)
sycamore = western sycamore (Platanus racemosa)
*Some native trees occur at the aggregation site and
support clusters, however the site is protected by
large blue gum eucalyptus.
**Site names and numbers correspond with Meade
et al 2018.
Meade, Griffiths, and van der Heide
Monarch Butterfly Habitat on the Razor’s Edge:
Native and Non-native Vegetation in California and Occam’s Razor; February 2023 9 There are nine south coast sites where monarch butterfly aggregations were observed clustering on native trees. These sites include: Canada Refugio (Site 49), El Rancho Refugio (Site 101), Corral Canyon (Site 51), Elks Grove (Site 76), Los Carneros Creek (Site 71), Carpinteria Creek (Site 99), Maria Ygnacio Creek at La Ramada Road (Site 115), Padaro Lane (Site 119), and Stow Grove County Park (Site 75).
Five of these sites are closely associated with eucalyptus. Carpinteria Creek and Elks Grove have large blue gum eucalyptus trees that provide shade and weather protection for the site. At those two sites, monarch butterflies cluster on both eucalyptus and native trees that overhang the creek.
At Rincon Creek, native sycamore, willow, and oak trees line the drainage adjacent to eucalyptus, and yet monarch butterflies chose to aggregate in the one stand of eucalyptus trees along the creek channel. At one time monarchs also aggregated at a eucalyptus row on Rincon Road (since removed). At Los Carneros Creek, monarchs have been observed clustering on oak branches over the creek and on eucalyptus trees that protect the site and line the creek bank.
The remaining four of these sites are not associated with eucalyptus trees: Canada Refugio, El Rancho Refugio, Corral Cany on, and Stow Grove County Park. Canada Refugio and El Rancho Refugio, which each held about 3,000 butterflies in 1998, had a high count of 7 and 75 respectively in October 2016. El Rancho Refugio site is adjacent to a lemon orchard where up to 12,000 butterflies were observed in native willow and cottonwood vegetation on January 6, 1990 (Calvert 1991). Calvert (1991) considered the location to be a nectaring bivouac for lemon blossoms. This site is in a relatively protected canyon, and native vegetation i s still present. Corral Canyon held an autumnal aggregation of 637 monarchs in 2016 and 1,425 butterflies in November 2017. These sites were then abandoned during the winter. The Stow Grove site is one of only two locations in the County where monarchs aggregated on redwood trees, holding 100 butterflies in 1999.
Drought severely damaged the redwoods, and only 3 butterflies were observed there in fall 2017.
These site-specific observations together with the monthly analysis shown in Figure 1 show us that butterflies can use south coast native stands of vegetation as aggregation sites; however, they support low numbers of monarchs in the fall and not through the overwintering period. The use of habitat dominated by eucalyptus is likely a matter of preference as well as availability.
2.4 Tree Characteristics
To evaluate candidates for restoration planting, we compiled a list of characteristics of trees native to California and species known at monarch aggregation sites. We used records in the Xerces Society’s database and site descriptions (Lane 1993, Meade 1999, Meade et al. 2018, Xerces Society 2023). The list of species and the compilation of characteristics are shown in Attachment A. We used tree character information from the Urban Forest Ecosystems Tree Detail list (Urban Forest Ecosystems Institute 2023), Calscape (California Native Plant Society 2023), and Cal-IPC Inventory (Cal-IPC 2023) databases, along with our observations and information from an analysis of aggregation habitats (Meade et al. 2018). Ratings are assigned based on characteristics that are directly related to forming su itable aggregations sites for monarch butterflies along south coast sites in Santa Barbara County . A higher rating score indicates a species better suited for use in monarch butterfly site restoration. Characters are summed for each species providing a comparative rating among species, shown in Table 3. The highest rated species are three eucalyptus trees commonly found at aggregation sites: red iron bark gum, blue gum, and river red gum. The next six species are trees that create suitable aggregation habitat for the du ration of the
Meade, Griffiths, and van der Heide
Monarch Butterfly Habitat on the Razor’s Edge:
Native and Non-native Vegetation in California and Occam’s Razor; February 2023 10 overwintering season: lemon scented gum, red flowering gum, toyon, karri, Monterey cypress, and sugar gum. All the trees with lower ratings would provide additional habitat benefits by increasing diversity and wind shelter to an aggregation grove in associate with the nine higher ranked species:
Bishop pine, silver wattle, Canary Island pine, holly-leaf cherry, Monterey pine, Torrey pine, coast redwood, evergreen ash, cork oak, blackwood, California bay laurel, western sycamore, Fremont cottonwood, coast live oak, grand fir, arroyo willow, Douglas fir, monkey-puzzle, dawn redwood, and black cottonwood.
TABLE 3. TREE SPECIES BY OVERALL MATRIX SCORE
Species Common Name Overall Score
Eucalyptus sideroxylon Red iron bark gum 62.0
Eucalyptus globulus Blue gum 61.0
Eucalyptus camaldulensis River red gum 59.0
Corymbia citriodora Lemon scented gum 56.0
Corymbia ficifolia Red flowering gum 56.0
Heteromeles arbutifolia Toyon 56.0
Eucalyptus diversicolor karri 55.0
Hesperocyparis macrocarpa Monterey cypress 54.0
Eucalyptus cladocalyx Sugar gum 53.0
Pinus muricata Bishop pine 51.0
Acacia dealbata silver wattle 50.0
Pinus canariensis Canary Island pine 50.0
Prunus illicifolia Holly-leaf cherry 50.0
Pinus radiata Monterey pine 49.0
Pinus torreyana Torrey pine 49.0
Sequoia sempervirens Coast redwood 48.0
Fraxinus uhdei Evergreen ash 47.0
Quercus suber Cork oak 47.0
Acacia melanoxylon Blackwood 46.0
Umbellularia californica California bay laurel 45.0
Platanus racemosa Western sycamore 44.0
Populus fremontii Fremont cottonwood 44.0
Quercus agrifolia Coast live oak 44.0
Abies grandis Grand fir 43.0
Salix lasiolepsis Arroyo willow 43.0
Pseudotsuga menziesii Douglas fir 42.0
Meade, Griffiths, and van der Heide
Monarch Butterfly Habitat on the Razor’s Edge:
Native and Non-native Vegetation in California and Occam’s Razor; February 2023 11 Species Common Name Overall Score
Araucaria araucana Monkey-puzzle 40.0
Metasequoia glyptostoboides Dawn Redwood 34.0
Populus trichocarpa Black cottonwood 34.0
3 CONCLUSION
Why are so few monarch butterfly aggregations in south Santa Barbara County in native vegetation, even though native vegetation is present near or adjacent to eucalyptus dominated aggregation sites? The simplest conclusion based on observation is that in this area of California, eucalyptus trees provide better shelter for monarch butterfly overwintering aggregations than do native plant habitats. The butterflies have shown us their preference when a choice is available and when we consider growth forms an d attributes of tree choices available. Tree species with a structure most likely to provide protection added with a history of monarch’s behavior explains what might continue to work. Eucalyptus grove structures often provide better protection from wind and rain than sycamore, oak, cottonwood, and willow habitats.
The recovery of the western monarch butterfly population will depend on action at every stage of their migration and life cycle. Restoration and enhancement of monarch overwintering sites is an important part of that recovery (Xerces Society 2019). When creating a protective aggregation habitat is the goal, eucalyptus trees remain a reliable choice for restoration. Monarchs have shown their preference for sites with eucalyptus trees in Sant a Barbara County. Using the species most likely to provide the best protection for aggregating monarchs seems the most reasonable approach to the pressing conservation question of what species to plant in Santa Barbara County south coast monarch restoration sites.
Meade, Griffiths, and van der Heide
Monarch Butterfly Habitat on the Razor’s Edge:
Native and Non-native Vegetation in California and Occam’s Razor; February 2023 12 4 REFERENCES
Bell, Brower, Calvert, Dayton, Frey, Leong, Murphy, Pyle, Sakai, Snow and Weiss. 1993. The Monarch Project’s Conservation and Management Guidelines. The Xerces Society. Pg. 21.
California Invasive Plant Council (Cal -IPC). 2023. The C al-IPC Inventory. https://www.cal - ipc.org
California Native Plant Society. 2023. Calscape Plant Description, Natural Setting, and Landscaping Information Guide. https://calscape.org/
Calvert, William H. 1991. Monarch butterfly overwintering site locations in Santa Barbara County, California. Center for Conservation Biology, Dept. of Bio. Sci., Stanford University, Stanford, CA.
Farmer, Jared. 2013. Trees in Paradise: The Botanical Conque st of California. W.W. Norton & Company. Ney York.
Griffiths, Jessica and Francis Villablanca. 2015. Managing monarch butterfly overwintering groves: Making room among the eucalyptus. California Fish and Game 101(1):40 -50; 2015.
Lane, John. 1993. Overwintering Monarch Butterflies in California: Past and Present. In Biology and Conservation of the Monarch Butterfly. Edited by Stephen B. Malcom and Myron P.
Zalucki. No. 38 Science Series, Natural History Museum of Los Angeles County. pp. 335- 344.
Leong, K. 1 990. Microenvironmental factors associated with the winter habitat of the monarch butterfly (Lepidoptera: Danaidae) in central California. Annals of the Entomological Society of America, 83: 906–910.
Leong, K. 2016. Evaluation and Management of California monarch Winter Sites. Western Wildlife 3:19-26.
Longcore, Travis, Catherine Rich, and Stuart B. Weiss. 2020. Nearly all California monarch overwintering groves require non -native trees. California Fish and Wildlife 106(3):220 - 225; 2020.
Marris, Emma. 2016. The Great Eucalyptus Debate. The Atlantic. Nov. 30, 2016. From The Last Word on Nothing” website.
Meade, Daniel E. 1999. Monarch Butterfly Overwintering Sites in Santa Barbara County. Report for the County of Santa Barbara. Althouse and Meade, Inc. 114 pp.
Meade, Daniel E., Jessica Griffiths, Charis van der Heide, and Francis Villablanca. 2018. Monarch Butterfly Overwintering Sites, Santa Barbara County. Althouse and Meade, Inc. Paso Robles, California.
Santos, R. L. 1997. The Eucalypt us of California: Seeds of Good, or Seeds of Evil? Denair, CA:
Ally-Cass Publications. 165 pp.
St. George, Zach. 2016. The Burning Question in the East Bay Hills: Eucalyptus is Flammable Compared to What? Bay Nature. October – December 2016.
Meade, Griffiths, and van der Heide
Monarch Butterfly Habitat on the Razor’s Edge:
Native and Non-native Vegetation in California and Occam’s Razor; February 2023 13 Urban Fores t Ecosystems Institute. 2023. Select Tree: Tree Detail. https://selectree.calpoly. edu/tree-detail
Xerces Society for Invertebrate Conservation. 2023. Wester Monarch Thanksgiving Count Database. https://www.westernmonarchcount.org/data/
_____. 2019. Western Monarch Call to Action. https://xerces.org/save-western-monarchs/ Weiss, S.B., P.M. Rich, D.D. Murphy, W.H. Calvert, P.R. Ehrlich. 1991. Forest canopy structure at overwintering monarch butterfly sites: measurements with hemispherical photography.
Conservation Biology 5:165–175.
Wenner, Adrian. 1987. Monarch Butterflies: General Considerations. Le tter to Pacific Grove Planning Commission. October 1987.
Meade, Griffiths, and van der Heide
Habitat Analysis for the Ellwood Mesa, City of Goleta 14 February 2023
ATTACHMENT A. TREE COMPARISON FOR CHARACTERISTICS IMPORTANT TO MONARCH BUTTERFLY AGGREGATION SITES
Species Common Name Nativeness
Percentage
of Monarch
Use in SB Co
Used by Roosting
Monarchs in California Nectar value Growth Form Height (m) Canopy/wind
protection
factor Ease of care
Water
require-
ment
(Drought
Tolerance)
Branch
Strength
Life
Span Growth Rate
Climate
change
Resilience
Potential for
success at
Ellwood
North site
Cal-IPC
invasive
level Overall Score
Abies grandis Grand fir 3 0 1 0 5 5 5 4 1 3 5 4 1 1 5 43.0 Acacia dealbata silver wattle 0 1 4 3 5 4 5 5 4 1 3 5 4 4 2 50.0 Acacia melanoxylon Blackwood 1 0 3 2 4 3 3 5 5 1 3 5 4 4 3 46.0 Araucaria araucana Monkey-puzzle 1 0 1 0 5 5 2 4 1 3 5 3 2 3 5 40.0 Corymbia citriodora Lemon scented gum 1 1 3 5 5 5 3 4 5 3 3 5 4 4 5 56.0 Corymbia ficifolia Red flowering gum 1 2 4 5 5 3 2 5 5 3 3 3 5 5 5 56.0 Eucalyptus camaldulensis River red gum 0 4 5 5 5 5 3 5 5 4 3 5 3 4 3 59.0 Eucalyptus cladocalyx sugargum 0 1 4 3 4 4 3 5 4 3 5 5 4 4 4 53.0 Eucalyptus diversicolor karri 0 0 1 4 5 5 5 5 4 3 5 5 4 4 5 55.0 Eucalyptus globulus Blue gum 0 5 5 5 5 5 5 5 4 2 5 5 3 4 3 61.0 Eucalyptus sideroxylon Red iron bark gum 0 3 5 5 5 4 4 5 5 3 3 5 5 5 5 62.0 Fraxinus uhdei evergreen ash 1 1 5 1 5 4 5 5 1 2 3 5 3 1 5 47.0 Hesperocyparis macrocarpa Monterey cypress 3 2 4 0 5 4 5 4 3 4 3 5 3 4 5 54.0 Heteromeles arbutifolia Toyon 5 2 5 2 3 1 5 5 5 3 3 3 4 5 5 56.0 Metasequoia glyptostoboides Dawn Redwood 2 0 1 0 1 4 2 3 1 3 5 5 1 1 5 34.0 Pinus canariensis Canary Island pine 2 1 3 0 5 4 5 4 5 3 3 5 4 1 5 50.0 Pinus muricata Bishop pine 4 1 4 0 5 4 5 4 4 3 3 5 1 3 5 51.0 Pinus radiata Monterey pine 3 2 3 0 5 5 5 4 1 2 3 5 3 3 5 49.0 Pinus torreyana Torrey pine 3 0 1 0 4 5 5 4 4 3 5 5 3 2 5 49.0 Platanus racemosa Western sycamore 5 1 4 0 1 4 3 3 1 3 5 5 2 2 5 44.0 Populus fremontii Fremont cottonwood 5 1 4 2 1 4 4 5 1 1 3 5 2 1 5 44.0 Populus trichocarpa Black cottonwood 5 1 1 0 1 4 3 5 2 1 3 3 2 1 5 37.0 Prunus illicifolia Hollyleaf cherry 5 0 1 5 5 1 2 5 4 3 3 4 3 4 5 50.0 Pseudotsuga menziesii Douglas fir 3 0 1 0 4 5 5 5 1 3 5 3 1 1 5 42.0 Quercus agrifolia Coast live oak 5 1 2 0 4 2 2 5 3 2 5 2 3 3 5 44.0 Quercus suber Cork oak 0 0 1 0 4 4 5 4 5 4 5 3 4 3 5 47.0 Salix lasiolepsis Arroyo willow 5 1 4 3 1 2 5 5 1 1 1 5 2 2 5 43.0 Sequoia sempervirens Coast redwood 3 1 5 0 4 5 3 4 1 5 5 5 1 1 5 48.0 Umbellularia californica California bay laurel 5 1 2 2 2 1 2 5 4 5 5 2 2 2 5 45.0 Ratings Broken Down for each column Rating Scale Nativeness Percentage
of Monarch
Use in SB Co
Used by Roosting
Monarchs in California Nectar value Growth Form Height (m) Canopy/wind
protection
factor Ease of care
Water
require-
ment
(Drought
Tolerance)
Branch
Strength
Life
Span Growth Rate
Climate
change
Resilience
Potential for
success at
Ellwood
North site
Cal-IPC
invasive
level
5 Goleta 75-100% Goleta
Nectar
available over
winter
Evergreen,
spreading 35 m and up Full canopy Easy
Extremely
Low Strong
Greater
than
150
years
Fast/ 36 in.
per year High High Not rated
4 Santa
Barbara
County 50-74% Santa Barbara Co
Some nectar
available
overwinter
Evergreen
columnar 15-35 m Top only
Moderately
Easy Very Low
Strong-
medium n/a Good Good watch/alert
3 South and
Central
California 25-49% Monterey Co south
Limited
nectar
available Evergreen, shrub 10-15 m
Top only
sparce Moderate Moderate Medium
50 to
150
years.
Moderate, 24
in. per year Medium Medium limited
2 West Coast 5-24% California
Nectar out of
season Sub-tree 5-10 m Sparce canopy
Moderately
High High
Medium-
weak n/a
Question-
able Question-able moderate
1 North
America 1-5% Limited records Poor nectar
Winter
deciduous 0-5 m No canopy High Very High weak
Less
than 50
years
less than 12
inches per
year n/a Low high
0 Another
continent 0% No Records No Nectar n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a Althouse and Meade, Inc.
Habitat Analysis and Recommendations for Ellwood Mesa, City of Goleta E-1 April 2023
APPENDIX E. ELLWOOD WIND DATA
Source: https://weatherspark.com/y/1430/Average-Weather-in-Goleta-California-United-States-Year-Round
Althouse and Meade, Inc.
Habitat Analysis and Recommendations for Ellwood Mesa, City of Goleta E-2 April 2023
November 2021 – January 2022 – Daily Wind
FIGURE 1. NOVEMBER 2021
Nov 9th – W 12-16 mph (3:30PM-5PM)
Nov 25th – W/SW 12-13 mph (12PM-3PM) / Gust of 25 mph (4am), variable direction
FIGURE 2. DECEMBER 2021
Dec 9th – W/NW - 12-20 mph (12pm-4pm) / Gusts 18-25 mph (12pm-4pm) Dec 13th – E/SE - 13-22 mph (9am-11pm) / Gusts – 18-23 mph early in the day, 23- 29 mph (9pm-11pm)
Dec 14th – E/SE - 13-20 mph (12am-3am) / Gusts 18-28 mph (12am-3am) – W/NW – 12-17 mph (8am-9pm) / Gusts 20-24 mph (8am-9pm) Dec 23-31 – Average wind – 6 mph variable direction / Gusts 22-26 mph (Large gusts mostly
W/NW/SW)
FIGURE 3. JANUARY 2022
Jan 21st – NE – 18 mph (11pm)
Althouse and Meade, Inc.
Habitat Analysis and Recommendations for Ellwood Mesa, City of Goleta E-3 April 2023
TABLE 1. JANUARY 21ST – JAN 22ND 2022- NE WIND AT MIDNIGHT
November 2022 – January 2023 – Daily Wind
FIGURE 4. NOVEMBER 2022
Nov 2nd – W/NW - 13-21 mph (6am -10pm) / Gusts 21- 31 mph (*especially strong @ 10am-4pm)
Nov 3rd – W/NW/NE - 10-18 mph (2am-7pm) / Gusts 22-28 mph (8am-2pm) Nov 7th – E/SE - 10-15 mph (9am-2pm)
Nov 8th – E/SE - 12-21 mph (7am-1pm) / Gusts 22-29 mph (10am-12pm) Nov 9th - W/SW 13-15 mph (1pm-4pm) / Gust 23 mph (1pm)
Althouse and Meade, Inc.
Habitat Analysis and Recommendations for Ellwood Mesa, City of Goleta E-4 April 2023
FIGURE 5. DECEMBER 2022
Dec 10th – E/SE - 10-18 mph (1pm-11pm) / Gusts 22-26 ((9pm-11pm) Dec 11th – S/SW/W/NW – 10-18 mph (12am -4pm) / Gusts - S/SW 20-28 mph(12am-2am) – N/NW 21-25 mph (12pm-4pm)
FIGURE 6. JANUARY 2023
Jan 1st - W/NW 13-24 mph (12am-7pm) / Gusts 17 -33 (12am-7pm, *especially strong from 8am – 7pm)
Jan 4th - E/SE 12 -25mph (12am -11pm) / Gusts 22- 38 mph (5am -11pm, *especially strong from 6pm-11pm)
Jan 5th - E/SE 12 -24 mph (12am -3pm) / Gusts 22- 33 mph (12am -3pm, *especially strong from 12am-5am)
Jan 16 - NE/E/SE/SW/W/NW 12 -22mph (12am -8pm) / Gusts 16 -28 (Sporadic throughout the day)
Jan 22 -W/SW/NW/W/NE 12-20 (4pm-11pm) / Gust 20 (3pm) & 29 mph (7pm)
Althouse and Meade, Inc.
Habitat Analysis and Recommendations for Ellwood Mesa, City of Goleta E-5 April 2023
TABLE 2. JAN 15TH – 16TH – E/NE WINDS THROUGHOUT THE NIGHT
TABLE 3. JANUARY 22ND – 23RD – N/NE WINDS THROUGHOUT THE NIGHT
Althouse and Meade, Inc.
Habitat Analysis and Recommendations for Ellwood Mesa, City of Goleta E-6 April 2023
Example Santa Ana wind events – October 2007, December 2017 TABLE 4. OCTOBER 21ST 2007
Althouse and Meade, Inc.
Habitat Analysis and Recommendations for Ellwood Mesa, City of Goleta E-7 April 2023
TABLE 5. DEC 16TH 2017 – SANTA ANA WIND DRIVEN FIRE EVENT
Althouse and Meade, Inc.
Habitat Analysis and Recommendations for Ellwood Mesa, City of Goleta E-8 April 2023
TABLE 6. DEC 20TH 2017 – POSSIBLE SANTA ANA WIND EVENT
Althouse and Meade, Inc.
Habitat Analysis and Recommendations for Ellwood Mesa, City of Goleta F-1 April 2023
APPENDIX F. DEAD WOOD REMOVAL AND MONARCH BUTTERFLY HABITAT EFFECTS
Dead wood removal and monarch butterfly habitat effects
March 9, 2022
TO: George Thomson
Parks & Open Space Manager
City of Goleta - Public Works Department
FROM: Dr. Daniel Meade, Principal Scientist, Althouse and Meade, Inc.
Dr. Stu Weiss, Chief Scientist, Creekside Science
Removal of dead downed wood and dead standing trees benefits monarch butterfly habitat and forest health. Forest habitat is best for monarch butterflies when it provides a balance between cover and room for flight. Wind shelter by a dense outer screen of trees is critical. Inside that screen, dappled light is key to monarch aggregation locations. Very dense canopies from overcrowded trees or open canopies from death of trees are not suitable as monarch aggregation habitat.
The Xerces Society guidelines for managing monarch groves (The Xerces Society 2017) suggests that even very important trees (VITs), “…that are also hazard trees should be replaced.” (pg. 14). “Completely dead trees and branches generally do not contribute to monarch habitat and are a potential liability. Felled debris from diseased and infected trees should be removed from the habitat to eliminate host material.” (pg. 19). “Standing dead trees generally do not contribute to monarch habitat and are a hazard to people and other trees.” (pg. 25).
The 1993 Monarch conservation and management guidelines (Bell et al. 1993) state, “The authors recognize the importance of removing hazardous trees from forests.” (Recommendation 18). And “Understory levels should be managed to prevent a localized buildup of dead plant materials.” (Recommendation 30).
Monarch Habitat
In our experience of visiting scores of monarch butterfly aggregation sites during more than 30 years of observation, monarch butterflies prefer to aggregate in locations with a wind protection buffer surrounding a more open center that allows access and patrolling space for flying butterflies.
Monarchs leave roosts subject to disruptive wind (Leong 2016) for protected locations that provide dappled light and ameliorate temperature and humidity extremes.
Monarch aggregation habitats often have low to moderate densities of ground vegetation with an open mid story within the wind protected area. Several famous monarch butterfly aggregation sites, such as Pismo Beach and Pacific Grove have highly managed understories free of downed wood.
The relatively clear floors in those groves include wide walking paths that are not detrimental to monarch aggregations that form in the surrounding trees. 56 In Ellwood, monarchs have sought out wind-sheltered gaps in the forest to form their aggregations.
Ellwood Main prior to the recent drought die -off was the exemplar of an aggregation site with a “Cathedral” structure consisting of an open understory, minimal ground cover, and a high vertical open space encompassed by protective trees. Surrounding slopes contributed to wind shelter and tall canopy trees formerly provided ideal dappled light and room to fly. The northern entrance to the aggregation site remained open and thicker vegetation around the edges of the site provided excellent wind buffering. The drought die-off of trees changed the site and where protective trees stood west and east of the grove, tangles of dead fallen wood and standing dead stems occupy the space and allow wind to penetrate far deeper into the grove than previously. Within the aggregation site, standing dead trees place other trees at risk, and downed trees and debris interfere with recruitment of replacement trees. Ellwood Main no longer provides the excellent aggregation habitat as in past years.
Monarchs are often observed flying through stands of trees investigating habitat. This behavior can lead them to aggregations. Groves where trees are too dense, or where debris and tangles of dead wood block patrolling areas, reduce access and increase effort to locate suitable roosting locations.
Historically, Ellwood forests were managed to remove dead wood and maintain healthy stands of trees through normal forestry practices (Santos 1997, Meade pers onal observation). Failure to manage a grove results in over recruitment of trees during periods of higher rainfall and then death of trees during times of water stress (droughts). Appropriately managing tree density based on water demand creates a more stable grove and is a fundamental tenet of forestry management, even for groves managed for wildlife (Grebner et al. 2021, Taylor 2018).
Eucalyptus does not break down as quickly as native vegetation and wood and bark debris that accumulates on the ground retains nutrients and can inhibit new growth. Fallen blue gum eucalyptus trees decompose very slowly (DiTomaso et al. 2013), especially in dry climates. The high level of phenolics in eucalyptus inhibits fungi from breaking down the wood and allows accumulation of fire fuel (Reid et al.2005). Chipping will aid the release of nutrients from woody debris by converting coarse wood to fine pieces (Russell et al. 2015). Mulching around new forest plantings can increase planting survival and growth rate (Cahill et al. 2005, Chalker-Scott 2007).
Wildfire Risk
Accumulation of bark and dead wood on the ground and leaning into the tree canopy creates a major fire hazard to monarch habitats (Xerces 2017). Ignition from camp fires and other human sources can feed on ground debris and quickly escalate into a catastrophic fire event. This risk can be managed by reducing fuel load through removal of dead branches, trunks and highly flammable vegetation that can create fire ladders into the forest crown. The Ellwood groves, once managed for wood production, now present a tangle of downed wood, dead leaning trunks and over abundant understory shrubs. The City of Goleta Community Wildfire Protection Plan (2012) provides management recommendations to reduce, “potential fire intensity, rate of spread, and severity of effects.” (pg 58). Management prescriptions in the CWPP when implemented will help to protect life, property, and natural resources including monarch butterfly habitat.
57 The presence of large quantity of dead wood at Ellwood results in:
Increased wind penetration and speed at aggregation locations due to downed trees. Higher risk to remaining live trees
Reduction in shade at aggregation sites
Loss of cover for butterflies from predators
Poor recruitment and growth of replacement trees
Increased wildfire risk - intensification of fires and a fire ladder to canopy Public safety hazard from dead standing and leaning trees and ground debris
Removal of dead wood from the Ellwood forests and monarch groves will have positive effects for monarch butterflies.
Removal and replacement of dead trees will increase humidity and shade and boost protection of monarchs.
Removal of downed trees and branches eliminates debris tangles that interfere with recruitment of saplings and growth of trees with good form. Space for planting trees will be opened by removing dead wood. Removal of leaning dead trees and branch and debris tangles will improve patrolling and clustering habitat for monarch butterflies.
Reduction of hazards for visitors to the grove
Preserve habitat by reducing intensity and rate of spread of fires. 58 Photo 1 shows the dead stand along the trail that runs west of the gully at Ellwood Main. Large tangles of fallen trees lie among the standing dead, and the canopy is very open.
The hemispherical photographs (hemiphotos) below show some of the impacts of tree mortality on the effectiveness of windbreaks. Photo 2 was taken near the same site as Photo 1 above. The hemiphoto is taken straight up - the horizon is at the circular edge, and the zenith is the center of the circle. North is at the top of the photograph, and east and west are reversed from a map view because the photograph is looking upward. The opening in the orange circle in the SW direction is where many trees died (the site in Photo1), allowing wind to penetrate well into the aggregation site. The dead standing trees block only a tiny fraction of the sky compared with the live trees in other directions. Photo 3 was taken 10 meters south of Photo 2, showing the same wind shelter gap, which affects a large area downwind. In this case, the loss of shelter against prevailing SW winds has degraded the quality of Ellwood Main.
59 Photo 2
Photo 3
The map below shows a measure of canopy greenness (Normalized Difference Vegetation Index, NDVI) where the blue-green shows healthy foliage and the brown shows dead canopy (and brown grass outside of the forest). Note how much of the canopy around Ellwood Main has died, leaving dead standing trees, and large piles of downed wood.
The orange circle is the gap in Photos 1-3., which leaves the aggregation area to the east of the photograph sites (424 and 425) vulnerable to prevailing SW winds. Note also the numerous dead tree crowns throughout Ellwood Main, and the particularly large dead area at the western edge of the map, which is shown from the ground in Photo 4.
60 Normalized DIfference Vegetation Index
(NDVI) map of Ellwood Main. Blue-green is
live foliage (high NDVI) and brown is dead
trunks, branches, and grass.
Map 1
Photo 4
Below is a profile of forest, with the same NDVI color scale as Map 1, taken along the west edge of Ellwood Main (red rectangle). The small dip at the left is the road from the mesa going down to Deveraux Creek. The large masses of brown dead trees stand out, with a particularly dense mass of dead trees at the center -right. The oblique view gives some sense of the depth of the dead zone. All of the dead standing material will eventually fall to the ground, further opening up the canopy to wind.
61 Profile view west side Ellwood Main
Oblique view west side Ellwood Main
62 Maps of wind exposure
The map below shows “wind roses” of the exposure from eight cardinal directions, with size of the arrow being proportional to the fraction of open sky in that direction. The arrows are color coded where blue is well -sheltered (<0.20), yellow is marginal (0.20 -0.30) and red is high exposure (>0.30). The same colors are used for the interpolated surfaces of SW wind (left) and W wind (right).
The increased wind penetration into the western edge, caused by the tree deaths, is apparent in the red and yellow intrusions. Sealing up this edge will greatly increase habitat suitability within Ellwood Main.
63 References
Bell, Elizabeth, et al. 1993. Conservation and Management Guidelines for Preserving the Monarch Butterfly Migration and Monarch Overwintering Habitat in California. The Monarch Project. .
Cahil, A., L. Chalker-Scott and K Ewing. 2005. Wood-chip mulch improves plant survival and establishment at no-maintenance restoration site (Washington). Ecological Restoration , Vol. 23. 212-213.
Chalker-Scott, L. 2007. Impact of mulches on landscape plants and the environment – a review.
Journal of Environmental Horticulture 25(4): 239-249.
DiTomaso, J.M., G.B. Kyser et al. 2013. Weed Control in Natural Areas in the Western United States. Weed Research and Information Center, University of California. 544 pp.
Grebner, Donald, Pete Bettinger, Jacek Siry, and Kevin Boston. 2021. Introduction to Forestry and Natural Resources. 2nd Edition, Academic Press.
Leong, Kingston L.H. 2016. Evaluation and Management of California Monarch Winter Sites.
Western WIldlife 3:19-26.
Reid, J.B., B.M .Potts, R.S. Hill, M.J. Brown and M. J. Hovenden. 2005. Eucalypt Biology, Vegetation of Tasmania, Australian Biological Resources Study, Canberra. ISBN 064644512X.
Russell MB, Fraver S, Aakala T, Gove JH, Woodall CW, D’Amato AW, Ducey MJ 2015:
Quantifying carbon stores and decomposition in dead wood: a review. For. Ecol. Manage., 350, 107–128.
Taylor, Mac. 2018. Improving California’s Forest and Watershed Management. Legislative Analyst’s Office, State of California.
The Xerces Society. 2017. Protecting California’s Butterfly Groves.: Management Guidelines for Monarch Butterfly Overwintering Habitat. 32+vi pp. Portland, OR: The Xerces Society for Invertebrate Conservation.
Santos, R. L. 1997. The Eucalyptus of California. Alley-Cass Publications. 165 pp. 64 Althouse and Meade, Inc.
Habitat Analysis for the Ellwood Mesa, City of Goleta G-1 April 2023
APPENDIX G. ELLWOOD MONARCH BUTTERFLY HABITAT CHARACTERIZATION REPORT
Prepared by NV5 dated October 24, 2022
1100 NE Circle Blvd
Suite 126
Corvallis, OR 97330
Ellwood Mesa Monarch Butterfly Habitat Characterization
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION October 24, 2022
Prepared For:
Althouse and Meade, Inc.
Dan Meade
1602 Spring St
Paso Robles, CA 93446
[email protected]
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | i
TABLE OF CONTENTS
1.0 Introduction.............................................................................................................................. 1
2.0 Acquisition ............................................................................................................................... 3
2.1 Planning ............................................................................................................................................. 3
2.2 Airborne Survey ................................................................................................................................ 3
2.2.1 Lidar ........................................................................................................................................... 3
2.2.2 Digital Imagery ......................................................................................................................... 5
2.3 Ground Survey ................................................................................................................................... 6
2.3.1 Base Stations ............................................................................................................................ 6
2.3.2 Ground Survey Points (GSPs) ................................................................................................. 7
2.3.3 Air Targets ................................................................................................................................. 7
3.0 Lidar and Imagery Processing .............................................................................................. 8
3.1 Lidar Data .......................................................................................................................................... 8
3.2 Digital Imagery ................................................................................................................................. 8
4.0 Acquisition Results and Accuracy Assessment ................................................................. 9
4.1 Lidar Density ..................................................................................................................................... 9
4.2 Lidar Accuracy Assessments ........................................................................................................ 10
4.2.1 Lidar Non-Vegetated Vertical Accuracy .............................................................................. 10
4.2.2 Lidar Relative Vertical Accuracy .......................................................................................... 11
4.2.3 Lidar Horizontal Accuracy ..................................................................................................... 12
4.3 Orthophoto Accuracy ..................................................................................................................... 13
4.3.1 Ground Control ....................................................................................................................... 13
4.3.2 Anomalies and Misfits ........................................................................................................... 14
4.4 Acquisition Accuracy Controls ...................................................................................................... 14
5.0 Analytical Methodology ........................................................................................................ 16
5.1 Lidar Point Density Profiles .......................................................................................................... 16
5.1.1 Products .................................................................................................................................. 16
5.2 Standing Dead and Downed Woody Debris Classification ...................................................... 20
5.2.1 Woody Material Identification .............................................................................................. 20
5.2.2 Standing Dead ........................................................................................................................ 20
5.2.3 Downed Woody Debris .......................................................................................................... 22
5.2.4 Dead Material Volume Estimates ........................................................................................ 22
5.2.5 Products .................................................................................................................................. 23
5.3 Fisheye Photo Simulation from the Lidar Point Cloud ............................................................. 26
5.3.1 Plot Adjustments .................................................................................................................... 27
5.3.2 Photo Simulation ................................................................................................................... 28
5.3.3 Products .................................................................................................................................. 30
6.0 Discussion .............................................................................................................................. 31
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | ii
7.0 Fisheye PLot Photo Comparison ........................................................................................ 33
8.0 Glossary .................................................................................................................................. 43
LIST OF TABLES
Table 1: Acquisition dates, acreage, and data types collected on the Ellwood site ................ 1 Table 2: Deliverable product projection information. ................................................................... 1 Table 3: Lidar and imagery products delivered for the Ellwood site. ......................................... 2 Table 4: Analytical products delivered for the Ellwood site. ........................................................ 2 Table 5: Lidar specifications and survey settings ......................................................................... 4 Table 6: Camera manufacturer’s specifications. .......................................................................... 5 Table 7: Project-specific orthophoto specifications. ..................................................................... 6 Table 8: Base station position for the Ellwood acquisition. Coordinates are on the NAD83 datum. .................................................................................................................................................. 6 Table 9: NV5 Geospatial ground survey equipment identification. ............................................ 7 Table 10: ASPRS LAS classification standards applied to the Ellwood dataset. ..................... 8 Table 11: Average lidar point densities. ......................................................................................... 9 Table 12: Absolute accuracy results. ............................................................................................ 10 Table 13: Relative accuracy results .............................................................................................. 12 Table 14: Lidar horizontal accuracy. ............................................................................................. 12 Table 15: Orthophotography accuracy statistics for Ellwood Mesa. ........................................ 13 Table 16: Location of air target check points withheld from aerial triangulation adjustment. 14 Table 17: Lidar accuracy error sources and solutions. ............................................................... 15 Table 18: Las classification scheme. ............................................................................................ 26 Table 19: Plot adjustment change log. Bearings are recorded in degrees from North and distances are recorded in feet to match the project projection units. ..................................... 27 Table 20: Field photo and simulated fisheye photo comparison. Simulated fisheyes are generated from the lidar point cloud at 1.75-meter height above ground. The asterisk (*) on a plot number indicates that a confident plot geolocation adjustment could not be made and that there is not confidence in the simulated fisheye results. ............................... 33
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | iii
LIST OF FIGURES
Figure 1: Location map of the Ellwood 2021 site in California. ................................................. 3 Figure 2: NV5 Geospatial’s DJI M600 with Riegl VUX-1LR LiDAR Scanner............................... 4 Figure 3: Ground survey location map. ........................................................................................... 7 Figure 4: Cross section of the lidar point cloud in Ellwood. ......................................................... 8 Figure 5: Frequency distribution of first return point density values per 1 x 1 meter cell. ..... 9 Figure 6: Frequency histogram for lidar classified LAS deviation from ground check point values (NVA). ..................................................................................................................................... 11 Figure 7: Frequency histogram for the lidar bare earth DEM surface deviation from ground check point values (NVA) ................................................................................................................ 11 Figure 8: Frequency plot for relative vertical accuracy between flight lines ........................... 12 Figure 9: Map of orthophoto accuracy assessments. ................................................................. 13 Figure 10: [Left] Demonstration of lidar point density percentage from 1-5 meters above ground. Houses, shrubs, leaning eucalyptus stems, and the lower levels of the tree canopy are visible in this map. [Right] Ladder fuels AOI within the Ellwood project area. ................ 17 Figure 11: Summarized ladder fuels AOI showing average lidar point density for the 1-2 meter profile. .................................................................................................................................... 17 Figure 12: Map displaying lidar point density profiles for an example region of Ellwood. Top to bottom: Canopy Height Model shows a surface model of the highest returns captured in the height normalized point cloud. The density profiles from 0-1 meters above ground, 1-2 meters above ground, and 2-3 meters above ground, respectively. ........................................ 18 Figure 14: Example of dead trees identified in the orthoimagery. ........................................... 21 Figure 15: Mapped results of the identified standing dead trees within the Ellwood area of interest. .............................................................................................................................................. 21 Figure 17: Voxel representations of classified downed woody debris for a lidar tile. A six- inch voxel size is shown on the left and a 12-inch voxel size is shown on the right. Voxel visualizations are colored by elevation above ground. Cool colors represent features closer to ground level, while warmer colors represent features farther from the ground. .............. 23 Figure 21: Downed woody debris volume raster for an example region of Ellwood. The raster visualization is colored by the number of voxels representing standing dead material within a given pixel footprint. ......................................................................................................... 25 Figure 22: Example image of the classified lidar point cloud showing the identified standing dead material and downed woody debris for a particular unhealthy region of the Ellwood project area. ....................................................................................................................... 26 Figure 23: An example of a point cloud projected to a sphere. The yellow sphere located in the center of the unmodified point cloud (left) represents the center of the sphere and the point of view for the simulated photo. .......................................................................................... 29
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | iv Figure 24: (A) Projected Point Cloud: An example of a point cloud projected to a sphere.
The small yellow sphere located in the center of the projected point cloud represents the center of the projection sphere and the point of view for the simulated photo. (B) Cropped Projected Point Cloud: The projected point cloud cropped at the height of the point of view. (C) Point of View Enforced Simulated Photo: The cropped point cloud is viewed from below and the orientation of the photo is enforced to ensure that North is on top. ......................... 29 Figure 25: Image matrix showing various sphere radius and voxel size pairs. The selected parameters with the best look and feel compared to field photographs was a 400 foot radius and 2 foot voxel size (shown in the bottom right). .......................................................... 30 Figure 26: Image matrix showing how the input data tile size changes the resulting simulated image. A tile size of 400 feet was selected to balance the inclusion of features represented in the image and memory limitations. ................................................................... 30 Figure 27: Examples of simulated fisheye photos centered on a historic monarch cluster site and moving upwards through the canopy. Left to right images show the camera point of view situated 20 feet, 30 feet, 40 feet, and 50ft above the ground. .................................. 31
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 1
1.0 INTRODUCTION
In June 2020, NV5 Geospatial (NV5G) was contracted by Althouse and Meade, Inc. to collect high density Light Detection and Ranging (lidar) and multispectral orthoimagery data for the fall of 2021 across the Ellwood Mesa site in Goleta, California. Site characterization products were then developed to support planning, management, and future analysis of the Ellwood Mesa site using the lidar, imagery, and field data provided by Stu Weiss of Creekside Science.
The Ellwood Mesa Monarch Butterfly Habitat Characterization Project was undertaken using an experimental approach that would allow for the design of custom deliverables to meet the needs of various stakeholders involved in the project. By working with Stu Weiss of Creekside Science and Althouse and Meade, Inc. a suite of analytical datasets was outlined to support the characterization of monarch butterfly habitat and management of the Ellwood Mesa project area. An important outcome for this project was to create analytical products that would support the rehabilitation of the eucalyptus grove and site management to mitigate wildfire risk for nearby structures. To support these efforts, deliverable datasets were designed to leverage the three-dimensional nature of a lidar point cloud to identify the presence of vegetation and dead material in different height strata throughout the project area. The three primary deliverables designed for this project are lidar point density profiles for ladder fuels assessment, standing dead and downed woody debris classification, and simulated fisheye photos from the lidar point cloud. These datasets were identified as being more valuable to the characterization of monarch butterfly habitat and site management than the tree and canopy metric deliverables originally outlined in the project proposal.
This report accompanies the delivered lidar data, imagery, and analytical products. It documents contract specifications, data acquisition procedures, processing methods, and analysis of the final dataset including lidar accuracy and density. Acquisition dates and acreage are shown in Table 1, the projection information for the project is shown in Table 2, a complete list of contracted deliverables provided to Althouse and Meade, Inc. and Creekside Science is shown in Table 3 and Table 4, and the project area of interest (AOI) is shown in Figure 1.
Table 1: Acquisition dates, acreage, and data types collected on the Ellwood site Project Site Contracted
Acres
Buffered
Acres Acquisition Dates Data Type
Ellwood 230 270 09/27/2021 Lidar
Ellwood 230 270 10/29/2021
5 band (Red, Green, Blue, Near
Infrared, Red Edge) Digital
Imagery
Table 2: Deliverable product projection information.
Projections EPSG Horizontal Datum Vertical Datum Units
California zone 5 6424 NAD83(2011) NAVD88 (GEOID18) US Survey Feet
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 2 Table 3: Lidar and imagery products delivered for the Ellwood site.
Product Type File Type Product Details
Points LAS v 1.4 (*.las, *.laz) All Classified Returns
Vectors Shapefiles (*.shp)
• Project Area of Interest
• Lidar Tile Index
• Ground Survey Data
Rasters GeoTiffs (*.tif)
• Bare Earth Digital Elevation Model (DEM) –
1.5 ft resolution
• Highest Hit Digital Surface Model (DSM) – 1.5
ft resolution
• Normalized Digital Surface Model (nDSM) –
1.5 ft resolution
• Intensity Raster – 0.25 ft resolution
Digital Imagery GeoTiffs (*.tif) 9.75 cm Orthophotography • 5 spectral bands (RGB, NIR, Red Edge)
Table 4: Analytical products delivered for the Ellwood site.
Product Type File Type Product Details
Points LAS v 1.4 (*.las)
Height Normalized Point Cloud with Custom
Classification
• Standing dead wood (64), downed woody
debris (65), tree stems (67), ground (2), and
unclassified (1).
Vectors Shapefiles (*.shp)
• Point Density Profiles Summarized to 100 ft x
100 ft Ladder Fuel Grid
• Adjusted Plot Locations
Rasters GeoTiffs (*.tif)
• Point Density Profiles – 1 meter resolution
• Dead Material Volume Rasters
o Standing Dead Wood – 6-inch and
12-inch voxel
o Downed Woody Debris – 6-inch and
12-inch voxel
Tables Comma Separated Values
(*.csv)
Point Density Profiles Summarized to 100 ft x 100
ft Ladder Fuel Grid
Images JPEG (*.jpeg)
Simulated Fisheye Photos from Lidar Point Cloud
• Adjusted plot locations with points of view
(POV) located 1.75, 5, and 10 meters
above ground
• Historic Monarch cluster locations with
POV located at 5-foot intervals for the
range of heights measured for each
Monarch cluster.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 3
Figure 1: Location map of the Ellwood 2021 site in California.
2.0 ACQUISITION
2.1 PLANNING
In preparation for data collection, NV5 Geospatial reviewed the project area and developed a specialized flight plan to ensure complete coverage of the Ellwood Mesa lidar study area at the target point density of ≥50.0 points/m2. Acquisition parameters including orientation relative to terrain, flight altitude, pulse rate, scan angle, and ground speed were adapted to optimize flight paths and flight times while meeting all contract specifications.
Factors such as satellite constellation availability and weather windows must be considered during the planning stage. Any weather hazards or conditions affecting the flight were continuously monitored due to their potential impact on the daily success of airborne and ground operations. In addition, logistical considerations including private property access and potential air space restrictions were reviewed.
2.2 AIRBORNE SURVEY
2.2.1 Lidar
The lidar survey was accomplished using a Riegl VUX-1UAV system mounted on an DJI Matrice 600 Unmanned Aerial Vehicle (UAV) (Figure 2). Table 5 summarizes the settings used to yield an average pulse density of 50 pulses/m2 over the Ellwood Mesa project area. The Riegl VUX-1UAV laser system can record 8 range measurements (returns) per pulse at 122 meters nominal terrain height.
It is not uncommon for some types of surfaces (e.g., dense vegetation or water) to return fewer
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 4 pulses to the lidar sensor than the laser originally emitted. The discrepancy between first return and overall delivered density will vary depending on terrain, land cover, and the prevalence of water bodies. All discernible laser returns were processed for the output dataset.
Figure 2: NV5 Geospatial’s DJI M600 with Riegl VUX-1LR LiDAR Scanner.
Table 5: Lidar specifications and survey settings
Lidar Survey Settings &
Specifications Details
Acquisition Dates September 27, 2021
Aircraft Used DJI Matrice 600
Sensor Riegl
Laser VUX-1UAV
Maximum Returns 8
Resolution/Density Average 50 pulses/m2
Survey Altitude (AGL) 122 m
Survey speed 15 knots
Field of View 90⁰
Mirror Scan Rate 78.5 Lines Per Second
Target Pulse Rate 820 kHz
Pulse Length 3 ns
Laser Pulse Footprint Diameter 6.1 cm
Central Wavelength 1064 nm
Pulse Mode Multiple Times Around (MTA)
Beam Divergence 0.5 mrad
Swath Width 244 m
Swath Overlap 50%
Intensity 16-bit
Accuracy RMSEZ ≤ 10 cm
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 5 All areas were surveyed with an opposing flight line side-lap of ≥50% (≥100% overlap) in order to reduce laser shadowing and increase surface laser painting. To accurately solve for laser point position (geographic coordinates x, y and z), the positional coordinates of the airborne sensor and the attitude of the aircraft were recorded continuously throughout the lidar data collection mission.
Position of the aircraft was measured twice per second (2 Hz) by an onboard differential GPS unit, and aircraft attitude was measured 200 times per second (200 Hz) as pitch, roll and yaw (heading) from an onboard inertial measurement unit (IMU). To allow for post-processing correction and calibration, aircraft and sensor position and attitude data are indexed by GPS time.
2.2.2 Digital Imagery
Imagery was collected using a Micasense Red Edge-MX digital mapping camera (Table 6). The Red Edge –MX is a global shutter multispectral camera manufactured by Micasense. The system is mounted fixed to the UAS platform and collects multispectral imagery and Downwelling Light Sensor (DLS) data.
Table 6: Camera manufacturer’s specifications.
Camera Specification Details
Equipment Micasense Red Edge-MX
Focal Length 5.4 mm
Data Format Blue, Green, Red, NIR, Red
Edge
Pixel Size 8 cm @ 120 m
Image Size 1,280 x 960 pixels
Frame Rate 1 Hz
HFOV 47.2°
For the Ellwood Mesa site, images were collected in five spectral bands (red, green, blue, NIR, and Red Edge) with 90% along track overlap and 75% sidelap between frames. The acquisition flight parameters were designed to yield a native pixel resolution of ≤ 10 cm. Orthophoto specifications particular to the Ellwood Mesa project are in Table 7.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 6 Table 7: Project-specific orthophoto specifications.
Orthophoto Specification Details
Equipment Micasense Red Edge - MX
Spectral Bands Blue, Green, Red, NIR, Red Edge
Ground Sampling Distance (GSD) 9.4 cm pixel size
Along Track Overlap ≥90%
Cross Track Overlap ≥75%
Above Ground Level (AGL) 122 meters
GPS PDOP ≤3.0
GPS Satellite Constellation ≥6
Image 8-bit GeoTiff
2.3 GROUND SURVEY
Ground control surveys, including monumentation, aerial targets and ground survey points (GSPs) were conducted to support the airborne acquisition. Ground control data were used to geospatially correct the aircraft positional coordinate data and to perform quality assurance checks on final lidar data and orthoimagery products.
2.3.1 Base Stations
Base stations were utilized for collection of ground survey points using real time kinematic (RTK) and static survey techniques. Base station locations (Table 8) were selected with consideration for satellite visibility, field crew safety, and optimal location for GSP coverage.
Table 8: Base station position for the Ellwood acquisition. Coordinates are on the NAD83 datum.
Monument ID Latitude Longitude Ellipsoid (meters)
1000 34° 25' 16.71413" -119° 53' 48.47240" -9.868
NV5 Geospatial utilized static Global Navigation Satellite System (GNSS) data collected at 1 Hz recording frequency for each base station. During post-processing, the static GNSS data were triangulated with nearby Continuously Operating Reference Stations (CORS) using the Online Positioning User Service (OPUS1) for precise positioning. Multiple sessions over the same monument were processed to confirm antenna height measurements and to refine position accuracy.
Ground control was collected using real time kinematic (RTK) survey techniques. For RTK surveys, a roving receiver receives corrections from a nearby base station via radio, enabling rapid collection of points with relative errors less than 1.5 cm horizontal and 2.0 cm vertical. RTK surveys record data while stationary for at least thirty, one-second epochs for aerial targets and five, one-second epochs for GSPs. All aerial targets were calculated by taking the average of two independent thirty, one- second epochs. All GSP measurements were made during periods with a Position Dilution of
1 OPUS is a free service provided by the National Geodetic Survey to process corrected monument positions. http://www.ngs.noaa.gov/OPUS.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 7 Precision (PDOP) of ≤ 3.0 with at least six satellites in view of the stationary and roving receivers. See Table 9 for Trimble unit specifications.
2.3.2 Ground Survey Points (GSPs)
GSPs were collected in areas where good satellite visibility was achieved on paved roads and other hard surfaces such as gravel or packed dirt roads. GSP measurements were not taken on highly reflective surfaces such as center line stripes or lane markings on roads due to the increased noise seen in the laser returns over these surfaces. GSPs were collected within as many flightlines as possible; however, the distribution of GSPs depended on ground access constraints and monument locations and may not be equitably distributed throughout the study area (Figure 3).
2.3.3 Air Targets
Ground survey data were collected by NV5 Geospatial to adjust aircraft positional and attitude data and to perform an accuracy assessment of final orthophoto products. NV5 Geospatial collected hard surface air targets typically on high visibility road markings or temporary white vinyl chevrons. Air target points were surveyed throughout the Ellwood Mesa study area, prior to imagery acquisition, using RTK techniques (Figure 3). Hard surface points consisted of high contrast, road markings such as stop bars, turn arrows, or cement corners. Typically, each corner of the road marking or temporary air target panel was surveyed, in this way only one point was used for aerial triangulation while the remaining points were used for quality assurance purposes.
Table 9: NV5 Geospatial ground survey equipment identification.
Receiver Model Antenna OPUS Antenna ID Use
Trimble R10 Integrated Antenna TRMR10 NONE Rover
Figure 3: Ground survey location map.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 8
3.0 LIDAR AND IMAGERY PROCESSING
3.1 LIDAR DATA
Upon completion of data acquisition, NV5 Geospatial processing staff initiated a suite of automated and manual techniques to process the data into the requested deliverables. Processing tasks included GPS control computations, smoothed best estimate trajectory (SBET) calculations, kinematic corrections, calculation of laser point position, sensor, and data calibration for optimal relative and absolute accuracy, and lidar point classification (Table 10). Processing methodologies were tailored for the landscape. Figure 4 shows a cross section of the lidar acquired in the Ellwood Mesa project area.
Figure 4: Cross section of the lidar point cloud in Ellwood.
Table 10: ASPRS LAS classification standards applied to the Ellwood dataset.
Classification
Number Classification Name Classification Description
1 Default/Unclassified Laser returns that are not included in the ground class, composed of vegetation and anthropogenic features
2 Ground Laser returns that are determined to be ground using automated and manual cleaning algorithms
3.2 DIGITAL IMAGERY
To produce a normalized multispectral orthomosaic in post-processing it was necessary to take radiance images of a calibration panel with known spectral reflectance in each of the collected spectral bands. These images are then used for spectral reflectance calibration, which normalizes the radiance values across all acquired imagery and calculates resulting reflectance values for each image.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 9 Images of each spectral band were matched to their respective camera station and composited into stacked 5 band images. In Agisoft Metashape, these images were then adjusted and orthorectified using the Lidar-derived ground model to remove displacement effects from topographic relief inherent in the imagery. The resulting orthos were mosaicked and exported. The final mosaic was inspected and edited for seam cutlines across above ground features such as buildings and other man-made features.
4.0 ACQUISITION RESULTS AND ACCURACY ASSESSMENT
4.1 LIDAR DENSITY
The acquisition parameters were designed to acquire an average first-return density of 50 points/m2.
First return density describes the density of pulses emitted from the laser that return at least one echo to the system. Multiple returns from a single pulse were not considered in first return density analysis. Some types of surfaces (e.g., breaks in terrain, water, and steep slopes) may have returned fewer pulses than originally emitted by the laser. First returns typically reflect off the highest feature on the landscape within the footprint of the pulse. In forested or urban areas, the highest feature could be a tree, building or power line, while in areas of unobstructed ground, the first return will be the only echo and represents the bare earth surface.
The density of ground-classified lidar returns was also analyzed for this project. Terrain character, land cover, and ground surface reflectivity all influenced the density of ground surface returns. In vegetated areas, fewer pulses may penetrate the canopy, resulting in lower ground density.
The average first-return density of lidar data for the Ellwood Mesa project was 277.81 points/m2 (Table 11). The statistical distribution of first return densities per 1 meter x 1 meter cell are portrayed in Figure 5.
Table 11: Average lidar point densities.
Classification Point Density
First-Return 277.81 points/m2
Figure 5: Frequency distribution of first return point density values per 1 x 1 meter cell.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 10
4.2 LIDAR ACCURACY ASSESSMENTS
The accuracy of the lidar data collection can be described in terms of absolute accuracy (the consistency of the data with external data sources) and relative accuracy (the consistency of the dataset with itself). See Acquisition Accuracy Controls for further information on sources of error and operational measures used to improve relative accuracy.
4.2.1 Lidar Non-Vegetated Vertical Accuracy
Absolute accuracy was assessed using Non-Vegetated Vertical Accuracy (NVA) reporting designed to meet guidelines presented in the FGDC National Standard for Spatial Data Accuracy2. NVA compares known ground check point data that were withheld from the calibration and post-processing of the lidar point cloud to the triangulated surface generated by the classified lidar point cloud as well as the derived gridded bare earth DEM. NVA is a measure of the accuracy of lidar point data in open areas where the lidar system has a high probability of measuring the ground surface and is evaluated at the 95% confidence interval (1.96 * RMSE), as shown in Table 12.
The mean and standard deviation (sigma ) of divergence of the ground surface model from quality assurance point coordinates are also considered during accuracy assessment. These statistics assume the error for x, y and z is normally distributed, and therefore the skew and kurtosis of distributions are also considered when evaluating error statistics. For the Ellwood Mesa survey, 41 ground check points were withheld from the calibration and post processing of the lidar point cloud, with resulting non-vegetated vertical accuracy of 0.078 feet compared to classified LAS, and
0.066 feet as compared to the bare earth DEM, with 95% confidence (Figure 6, Figure 7).
Table 12: Absolute accuracy results.
NVA, as compared to
classified LAS
NVA, as compared to
bare earth DEM
Sample 41 points 41 points
95% Confidence
(1.96*RMSE) 0.078 ft 0.066 ft
Average -0.015 ft -0.013 ft
Median -0.017 ft -0.012 ft
RMSE 0.040 ft 0.033 ft
2 Federal Geographic Data Committee, ASPRS POSITIONAL ACCURACY STANDARDS FOR DIGITAL GEOSPATIAL DATA EDITION 1, Version 1.0, NOVEMBER 2014.
https://www.asprs.org/a/society/committees/standards/Positional_Accuracy_Standards.pdf.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 11
Figure 6: Frequency histogram for lidar classified LAS deviation from ground check point values (NVA).
Figure 7: Frequency histogram for the lidar bare earth DEM surface deviation from ground check point values (NVA)
4.2.2 Lidar Relative Vertical Accuracy
Relative vertical accuracy refers to the internal consistency of the data set as a whole: the ability to place an object in the same location given multiple flight lines, GPS conditions, and aircraft attitudes.
When the lidar system is well calibrated, the swath-to-swath vertical divergence is low (<0.10 meters). The relative vertical accuracy was computed by comparing the ground surface model of each individual flight line with its neighbors in overlapping regions. The average (mean) line to line relative vertical accuracy for the Ellwood Mesa Lidar project was 0.079 feet (0.024 meters) (Table 13, Figure 8).
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 12 Table 13: Relative accuracy results
Relative Accuracy
Sample 17 surfaces
Average 0.001 ft
Median 0.001 ft
RMSE 0.083 ft
Total Compared Points 6,465,865
Figure 8: Frequency plot for relative vertical accuracy between flight lines
4.2.3 Lidar Horizontal Accuracy
Lidar horizontal accuracy is a function of Global Navigation Satellite System (GNSS) derived positional error, flying altitude, and INS derived attitude error. The obtained RMSEr value is multiplied by a conversion factor of 1.7308 to yield the horizontal component of the National Standards for Spatial Data Accuracy (NSSDA) reporting standard where a theoretical point will fall within the obtained radius 95 percent of the time. Based on a flying altitude of 122 meters, an IMU error of
0.016 decimal degrees, and a GNSS positional error of 0.019 meters, this project was compiled to meet 0.36 feet (0.11 m) horizontal accuracy at the 95% confidence level (Table 14).
Table 14: Lidar horizontal accuracy.
Unit RMSEr ACCr
Feet 0.21 ft 0.36 ft
Meter 0.06 m 0.11 m
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 13
4.3 ORTHOPHOTO ACCURACY
Image accuracy was assessed using the air target points from the aerial triangulation procedure.
Image accuracy results are shown in Figure 9 and Table 15.
Figure 9: Map of orthophoto accuracy assessments.
Table 15: Orthophotography accuracy statistics for Ellwood Mesa.
Count X Error (ft) Y Error (ft) Z Error (ft) XY Error (ft) Total (ft) 9 0.0541056 0.0969402 0.0521177 0.111017 0.122642
4.3.1 Ground Control
Air target points from the aerial triangulation adjustment are listed with their location and residuals in Table 16.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 14 Table 16: Location of air target check points withheld from aerial triangulation adjustment.
Label X Error (ft) Y Error (ft) Z Error (ft) Total (ft) Image (pix) 1008 0.035652 -0.176307 0.0661093 0.191639 6.390 (17)
1009 -0.0633332 0.0813204 -0.0578349 0.11819 10.845 (8) 1010 -0.0950713 0.00616302 -0.028731 0.0995088 6.922 (8) 1011 0.070953 -0.0208685 0.0536556 0.0913714 5.937 (11) 1012 0.0282189 0.0662476 -0.0564813 0.0915161 26.104 (9) 1013 -0.046341 0.00135106 -0.0213413 0.0510369 29.024 (8) 1014 -0.0175511 0.178497 -0.0674167 0.19161 23.088 (8)
1015 0.0499553 -0.0690696 0.0321876 0.0911163 7.289 (8) 1048 0.0352748 -0.0733733 0.0616433 0.102117 16.989 (8) Total 0.0541056 0.0969402 0.0521177 0.122642 16.189
4.3.2 Anomalies and Misfits
No anomalies or misfits detected.
4.4 ACQUISITION ACCURACY CONTROLS
Relative Accuracy Calibration Methodology:
Manual System Calibration: Calibration procedures for each mission require solving geometric relationships that relate measured swath-to-swath deviations to misalignments of system attitude parameters. Corrected scale, pitch, roll and heading offsets were calculated and applied to resolve misalignments. The raw divergence between lines was computed after the manual calibration was completed and reported for each survey area.
Automated Attitude Calibration: All data were tested and calibrated using TerraMatch automated sampling routines. Ground points were classified for each individual flight line and used for line-to- line testing. System misalignment offsets (pitch, roll and heading) and scale were solved for each individual mission and applied to respective mission datasets. The data from each mission were then blended when imported together to form the entire area of interest.
Automated Z Calibration: Ground points per line were used to calculate the vertical divergence between lines caused by vertical GPS drift. Automated Z calibration was the final step employed for relative accuracy calibration.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 15 Table 17: Lidar accuracy error sources and solutions.
Type of Error Source Post Processing Solution
GPS
(Static/Kinematic) Long Base Lines None
GPS
(Static/Kinematic) Poor Satellite Constellation None
GPS
(Static/Kinematic) Poor Antenna Visibility Reduce Visibility Mask Relative Accuracy Poor System Calibration Recalibrate IMU and sensor offsets/settings
Relative Accuracy Inaccurate System None
Laser Noise Poor Laser Timing None
Laser Noise Poor Laser Reception None
Laser Noise Poor Laser Power None
Laser Noise Irregular Laser Shape None
Operational measures taken to improve relative accuracy:
Low Flight Altitude: Terrain following was employed to maintain a constant above ground level (AGL).
Laser horizontal errors are a function of flight altitude above ground (about 1/3000th AGL flight altitude).
Focus Laser Power at narrow beam footprint: A laser return must be received by the system above a power threshold to accurately record a measurement. The strength of the laser return (i.e., intensity) is a function of laser emission power, laser footprint, flight altitude and the reflectivity of the target.
While surface reflectivity cannot be controlled, laser power can be increased and low flight altitudes can be maintained.
Reduced Scan Angle: Edge-of-scan data can become inaccurate. The scan angle was reduced to a maximum of ±45o from nadir, creating a narrow swath width and greatly reducing laser shadows from trees and buildings.
Quality GPS: Flights took place during optimal GPS conditions (e.g., 6 or more satellites and PDOP [Position Dilution of Precision] less than 3.0). Before each flight, the PDOP was determined for the survey day. During all flight times, a dual frequency DGPS base station recording at 1 second epochs was utilized and a maximum baseline length between the aircraft and the control points was less than 13 nm at all times.
Ground Survey: Ground survey point accuracy (<1.5 cm RMSE) occurs during optimal PDOP ranges and targets a minimal baseline distance of 4 miles between GPS rover and base. Robust statistics are, in part, a function of sample size (n) and distribution. Ground survey points are distributed to the extent possible throughout multiple flight lines and across the survey area. 50% Side-Lap (100% Overlap): Overlapping areas are optimized for relative accuracy testing. Laser shadowing is minimized to help increase target acquisition from multiple scan angles. Ideally, with a 50% side-lap, the nadir portion of one flight line coincides with the swath edge portion of overlapping flight lines. A minimum of 50% side-lap with terrain-followed acquisition prevents data gaps.
Opposing Flight Lines: All overlapping flight lines have opposing directions. Pitch, roll and heading errors are amplified by a factor of two relative to the adjacent flight line(s), making misalignments easier to detect and resolve.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 16
5.0 ANALYTICAL METHODOLOGY
The goal of the analytical portion of this project is to create lidar derived products that support the management of the Ellwood Mesa eucalyptus stand and the characterization of monarch butterfly habitat within the project area. Three products were identified and designed to meet these needs:
lidar point density profiles for ladder fuel management, standing dead and downed woody debris identification to guide dead material removal and plan replanting, and simulated fisheye photos from the lidar point cloud to support monarch habitat characterization throughout the eucalyptus stand.
5.1 LIDAR POINT DENSITY PROFILES
To assess the prevalence and location of ladder fuels along the boundary of the eucalyptus grove and the nearby structures in the Ellwood Mesa site, point density profiles were generated from the lidar point cloud. Point density profiles highlight the location of dominant, subdominant, and understory vegetation by dividing the above ground space within the AOI into height profiles or slices and measuring the number of lidar returns within each profile. The density profiles generated for this project can be used to map ladder fuels near buildings as well as clumps and gaps of canopy cover within the eucalyptus grove.
Prior to generating the point density profiles, the point cloud was height normalized. Height normalization creates a point cloud where all ground returns have an elevation of 0 and the elevation of all features captured in the point cloud have heights relative to ground level. The height normalized point cloud was then filtered to remove ground classified points and sliced into 1-meter tall height slices (or profiles). The percentage of points within the profile range compared to the total number of points within the pixel footprint were summarized by pixel. Functionally this is a representation of the relative point density within each 1-meter voxel that covers the area of interest.
This dataset shows where returns are located within the specified height ranges so that ladder fuels can be identified or gaps in forest cover can be targeted for planting.
5.1.1 Products
A 37-layer raster product (at 1-meter resolution) was needed to fully characterize the height variability in the point cloud. Each layer, one through thirty-seven, was populated with a percentage of the total points captured in that 1-meter cube (or voxel) relative to all returns present in its respective vertical point cloud column. A refined area of interest was used in the ladder fuels assessment to target areas that are close to structures and require management to reduce wildfire risk. The ladder fuels AOI is shown in Figure 10. Each point density profile raster was summarized to the gridded polygons of the ladder fuels AOI to represent the mean point density of each polygon for all 37 density profiles (Figure 11). The summarized ladder fuels AOI can be used to identify regions of the AOI that may need management for fire risk.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 17
Figure 10: [Left] Demonstration of lidar point density percentage from 1-5 meters above ground. Houses, shrubs, leaning eucalyptus stems, and the lower levels of the tree canopy are visible in this map. [Right] Ladder fuels AOI within the Ellwood project area.
Figure 11: Summarized ladder fuels AOI showing average lidar point density for the 1-2 meter profile.
The lidar point density deliverable dataset allows a user to peel back layers of the point cloud to identify where ladder fuels may be present within the forest. Figure 12 provides a comparison between the canopy height model and the individual density profiles to demonstrate the concept of looking through the canopy to the inner layers of the point cloud. In the map shown at the top of Figure 12 the canopy height model showcases the upper canopy layers, rooftops, and exposed ground locations. A canopy height model shows a surface model of the highest returns captured in the height normalized point cloud. The first density profile layer (second from the top in Figure 12) represents the proportion of lidar returns present from 0-1 meter above ground and excludes lidar
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 18 returns that are classified as ground. This layer shows where low lying vegetation is present and shows values near 100 in regions where no lidar points are obscured by taller features like houses or canopy. The next two density profile layers shown in Figure 12 demonstrate where lidar point returns are located for the 1-2 meter profile and the 2-3 meter profile, respectively. In the 1-2 meter profile, returns from fences, shrubs, and leaning debris are visible. In the 2-3 meter profile, returns begin to capture roof edges and taller shrub features. Figure 13 shows an overview of the point density profile rasters for the lowest six point density profile layers (0-1 meters, 1-2 meters, 2-3 meters, 4-5 meters, and 5-6 meters).
Figure 12: Map displaying lidar point density profiles for an example region of Ellwood. Top to bottom: Canopy Height Model shows a surface model of the highest returns captured in the height normalized point cloud. The density profiles from 0-1 meters above ground, 1-2 meters above ground, and 2-3 meters above ground, respectively.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 19
Figure 13: Overview of lidar point density profile rasters for the Ellwood Mesa AOI. Top to bottom and left to right: 0-1 meter density profile, 1-2 meter density profile, 2-3 meter density profile, 3-4 meter density profile, 4-5 meter density profile, and 5-6 meter density profile.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 20
5.2 STANDING DEAD AND DOWNED WOODY DEBRIS CLASSIFICATION Standing dead trees and downed woody debris were identified in this project to support forest rehabilitation, fire risk mitigation, maintenance, and safety within the Ellwood project area.
Deliverable products related to the identification of dead material within the AOI were designed to provide mapped locations and volume estimates of dead material to aid in the planning and execution of hazard removal.
The standing dead and downed woody debris classification was implemented using imagery and lidar derived inputs in three main phases: an intermediate classification of woody material lidar returns, spectral classification of standing dead trees, and a threshold-based classification of downed woody debris with manual review. Downed woody debris is considered to be dead tree material within 0.5-3.5 feet (approximately 0.15-1 meters) above ground level. Any dead material identified higher than 3.5 feet above ground level was considered to be standing dead material.
5.2.1 Woody Material Identification
A preliminary step involved in the classification of the lidar point cloud for standing and downed dead material was a filtering process used to isolate woody material lidar returns, such as tree stems and branches. The lidar point cloud was filtered using non-ground lidar returns, lidar return intensity, and eigen values relating each point to its five nearest neighbors. Lidar returns identified as woody material in this process were classified as class 67. Standing dead material and downed woody debris were classified from this subset of woody material classified points. The woody material filtering procedure was very influential in the classification of standing dead and downed trees because it reduced the false positive classification of leafy material near dead trees.
The woody material class (67) is considered an intermediate classification because it was not manually edited or evaluated after assignment. The classification procedure was tuned to the specific attributes of the Ellwood lidar dataset and is expected to contain false positive and false negative classifications. The woody material classification was retained in the final deliverable dataset to support future analysis by the client, and it should be evaluated for appropriateness of use prior to inclusion in any additional work.
5.2.2 Standing Dead
The standing dead classification was focused on identifying and quantifying the volume of standing dead trees. To perform the standing dead classification, a model was trained to identify dead trees represented in the orthoimagery. Dead trees were identified as lacking green and leafy material.
They are characterized by exposed branches and light and brown colored returns from the stem and branch structures. The full spectrum of potential tree health statuses were not considered in this process, a binary live or dead classification scheme was used.
Spectral training points were selected by a photo interpreter and used to build a random forest model. The five imagery bands (R, G, B, NIR, Red edge) and additional spectral indices were incorporated as predictor variables. Once the model was applied across the area of interest, the results were vectorized, manually reviewed, and edited where necessary. Upon finalization of the two-dimensional standing dead vector layer, the classification was applied to the lidar point cloud, converting any points greater than 3.5 feet above ground and classified as 67 into class 65:
standing dead material. Figure 14 shows a map of standing dead features identified in the orthoimagery and Figure 15 shows a map of the identified standing dead trees within the Ellwood Mesa project area.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 21
Figure 14: Example of dead trees identified in the orthoimagery.
Standing Dead Trees in Ellwood Mesa
Figure 15: Mapped results of the identified standing dead trees within the Ellwood area of interest.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 22
5.2.3 Downed Woody Debris
The process for classifying downed woody debris was built upon the woody material classification described in Section 5.2.1. After the lidar point cloud was processed to identify woody material, all woody material returns (class 67) between 0.5 feet and 3.5 feet above the ground were reclassified as downed woody debris (class 64). The regions of classified downed woody debris were then extracted and manually edited. Manual review of the downed woody debris classification focused on reducing the false positive classification of shrubs and bushes and filling in gaps of downed stems and branches. Fine resolution vegetation density profiles were used to aid in the digitization of downed woody debris to visualize the forest floor through the canopy. The edited downed woody debris identification layer was then applied to lidar returns between 0.5 and 3.5 feet tall to create the final downed woody debris classification (class 64). Figure 16 shows a map of the identified downed woody debris within the Ellwood Mesa project area.
Downed Woody Debris in Ellwood Mesa
Figure 16: Mapped results of the identified downed woody debris within the Ellwood area of interest.
5.2.4 Dead Material Volume Estimates
Dead material volume estimates were generated from the classified lidar point cloud using voxels to estimate the area occupied by standing dead trees or downed woody debris. Voxelization of the lidar point cloud converts the point returns into 3-dimensional cubes of a user defined size. Multiple lidar returns can occupy one voxel, and all points are represented by a voxel. A six-inch voxel size was used for volume estimation for the Ellwood project because it created realistic visualizations of the point cloud that did not overinflate the areas represented by the stems in this forest. Figure 17 shows examples of the voxelized downed woody debris for a lidar tile to highlight the impact of voxel size on the overall volume estimate. A 12-inch voxel size creates a wood volume model treats every stem as at least 12 inches in diameter. This is not realistic for the eucalyptus trees within the Ellwood project area.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 23
Figure 17: Voxel representations of classified downed woody debris for a lidar tile. A six-inch voxel size is shown on the left and a 12-inch voxel size is shown on the right. Voxel visualizations are colored by elevation above ground. Cool colors represent features closer to ground level, while warmer colors represent features farther from the ground.
5.2.5 Products
The primary deliverable products for this dead wood classification analysis are geospatial rasters representing the mapped location and voxel count of the downed woody debris and standing dead tree classifications. A raster format was chosen to offer flexible visualization of the project results and the potential for custom summarization of specific areas of interest within the larger project area. To calculate the volume of standing dead or downed woody debris for a specific region of Ellwood, the sum of the pixel values within that geographic region (the total number of classified voxels) should be calculated and multiplied by 0.125 to measure the cubic footage of dead material (either standing dead or downed woody debris depending on which raster is being summarized). The total volume of standing dead trees within the Ellwood project area is estimated to be 222,953 cubic feet and the total volume of downed woody debris within the Ellwood project area is estimated to be 207,252 cubic feet. Figure 18, Figure 19, Figure 20, and Figure 21 show examples of the raster- based dead wood volume deliverables created for the Ellwood project area.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 24
Figure 18: Standing dead volume raster for an example region of Ellwood. The raster visualization is colored by the number of voxels representing standing dead material within a given pixel footprint.
Figure 19: Downed woody debris volume raster for an example region of Ellwood. The raster visualization is colored by the number of voxels representing standing dead material within a given pixel footprint.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 25
Figure 20: Standing dead volume raster for an example region of Ellwood. The raster visualization is colored by the number of voxels representing standing dead material within a given pixel footprint.
Figure 21: Downed woody debris volume raster for an example region of Ellwood. The raster visualization is colored by the number of voxels representing standing dead material within a given pixel footprint.
The downed woody debris and standing dead tree raster products were generated from the classified and voxelized point cloud results. To support further modeling and analysis in the Ellwood Mesa project area the classified lidar point cloud was also delivered to the client. The classification scheme for the delivered point cloud is outlined in Table 18. Class 67 was used as an intermediate classification for identifying woody material (as outlined in Section 5.2.1). However, it was not specifically tuned for accuracy beyond its needs for the dead material classification and should be evaluated for appropriateness of use prior to inclusion in any additional analyses. Figure 22 shows an example tile of the classified point cloud displaying only the standing dead and downed woody debris for a particularly unhealthy region of the project area.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 26 Table 18: Las classification scheme.
Classification Definition
1 Unclassified Points
2 Ground
64 Standing Dead (Z > 3.5 ft)
65 Downed Dead (Z <= 3.5 ft)
67 Woody Returns
Figure 22: Example image of the classified lidar point cloud showing the identified standing dead material and downed woody debris for a particular unhealthy region of the Ellwood project area.
5.3 FISHEYE PHOTO SIMULATION FROM THE LIDAR POINT CLOUD Simulated fisheye photos were targeted as deliverables for this project in order to supplement field collected photos and support seamless quantification of solar exposure and potential wind penetration within the Ellwood eucalyptus stand. Historically, monarch habitat suitability has been assessed at this site and others using fisheye photos processed with Hemiview software to characterize the canopy gap fraction and quantity and quality of light penetration into the canopy.
Appropriate wind shelter and solar exposure are important elements of quality monarch habitat.
Exhaustive fisheye photography field campaigns are limited by resource availability and by the physical limitations of camera placement positions. Simulated photos allow the photo point of view to be placed anywhere in x, y, and z space within the project area. This creates the potential for photos to be simulated in the middle of the canopy at monarch cluster locations or throughout the canopy to measure how canopy closure changes at different heights above ground. NV5G’s current investigation into simulated fisheye photos from a lidar point cloud is being complemented by an investigation into drone spherical photography by Althouse and Meade, Inc.
Fisheye photographs were collected and provided to NV5G by the client in designated field plots across the Ellwood project area. Photos were taken on a tripod approximately 5 feet tall and image locations were recorded using GPS. For each of the field photos, a lidar derived simulated photo was
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 27 created to facilitate comparison and evaluation of performance for the simulated photos. A prerequisite to simulating the fisheye photo in each plot, was to adjust for GPS error and ensure that photo locations were geolocated accurately and that features in the lidar matched with features captured in the field photographs.
5.3.1 Plot Adjustments
Thirty field plots were collected throughout the Ellwood project area by Stu Weiss, Creekside Science, and his team. For each plot, the plot center was flagged 2 meters south of a central reference tree and its GPS location was recorded. Then dominant trees within a 1/10th acre plot were selected and their distance and azimuth from plot center were measured. Additional information about these trees was also recorded, such as diameter at breast height, species, total height, crown condition, and health. A fisheye photo was also collected at each plot center location.
Manual adjustment of the plot geolocation was necessary to align the field collected data to the features captured in the lidar point cloud. Plot adjustment corrects for GPS error that occurs during field collection and allows for precise alignment of lidar derived products with features on the ground. To perform plot adjustments, the 3-5 dominant reference trees collected for each plot were projected around the plot center using the measured distances and azimuths to create tree constellations. The tree constellations were then used to perform plot location adjustments by matching the arrangement and heights of reference trees to features seen in the lidar point cloud.
Plot adjustments were tested by simulating fisheye photos at the adjusted locations and visually comparing the simulated photos with photos captured in the field. The same forest features (presence or absence of trees and patterns of view to sky) are expected to be seen in both the field and simulated photographs. Table 19 contains the plot adjustment changelog with the calculated bearings and distances of each plot adjustment. The most extreme adjustment in terms of distance moved was approximately 50 feet (Plot Number 5), however average adjustments were roughly 17 feet. Five of the thirty reference plots could not be confidently matched with the lidar features. Plots typically could not be aligned with the point cloud features due to high forest density in the region the plot was taken, or due to a small number of reference trees collected. In these cases, the reference tree patterns and heights were not unique enough to identify within the point cloud data.
Table 19: Plot adjustment change log. Bearings are recorded in degrees from North and distances are recorded in feet to match the project projection units.
Plot Number Bearing (degrees) Distance (feet) Comment
1 58 17.7 -
2 345 0.89 -
3 64 4.83 -
4 184 9.69 -
5 199 57.46 -
6 180 3.77 -
7 328 20.02 -
8 244 30.05 -
9 114 13.63 -
10 149 4.66 -
11 198 27.25 -
12 213 12.04 -
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 28 Plot Number Bearing (degrees) Distance (feet) Comment
13 199 7.02 -
14 220 7.73 -
15 43 33.45 -
16 120 33.35 -
17 16 20.62 -
18 190 12.47 -
19 181 26.75 -
20 22 10.46 Unable to confidently match plot with point cloud features
21 309 24.84 Unable to confidently match plot with point cloud features
22 153 4.31 -
23 44 11.75 -
24 205 2.08 Unable to confidently match plot with point cloud features
25 200 3.79 Unable to confidently match plot with point cloud features
26 168 18.78 -
27 198 29.48 -
28 180 7.09 -
29 350 29.16 -
30 97 15 Unable to confidently match plot with point cloud features
5.3.2 Photo Simulation
Simulated fisheye photos were created from the lidar point cloud by spherically projecting the point cloud, converting the points to voxels, flattening the image, and enforcing its orientation. Simulated photos were designed to contain equiangular distortion and a 180-degree viewing angle that matches the lens information provided by Stu Weiss, Creekside Science. The fisheye photo simulation procedure is built around the location of the target point of view. The point of view represents the simulated camera location and can be placed anywhere within the point cloud along the x, y, and z axes. For a given point of view, a subset of the full point cloud is extracted with the point of view at its center. Equiangular distortion of the simulated image is achieved by projecting the target point cloud to a sphere (Figure 23 and Figure 24A). The 180-degree viewing angle is achieved by cropping the projected point cloud at a height equal to the sphere’s center location (the location of the point of view) (Figure 24B). The upper hemisphere of the point cloud is then rotated and aligned so that the point of view is rendered looking up to the sky with North at the top of the simulated image. The point cloud is then voxelized in order to determine which pixels of the simulated photo are blocked by forest features and which have a clear view to sky (Figure 24C).
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 29
Figure 23: An example of a point cloud projected to a sphere. The yellow sphere located in the center of the unmodified point cloud (left) represents the center of the sphere and the point of view for the simulated photo.
Figure 24: (A) Projected Point Cloud: An example of a point cloud projected to a sphere. The small yellow sphere located in the center of the projected point cloud represents the center of the projection sphere and the point of view for the simulated photo. (B) Cropped Projected Point Cloud: The projected point cloud cropped at the height of the point of view. (C) Point of View Enforced Simulated Photo: The cropped point cloud is viewed from below and the orientation of the photo is enforced to ensure that North is on top.
Point cloud projection and voxelization parameters were tested by simulating fisheye photos with varying tile sizes, radii for spherical projection, and voxel sizing. A sphere radius of 400 feet with a voxel size of 2 feet was selected to most closely approximate the look and feel of field photographs based on visual interpretation (Figure 25). In order to balance memory limitations and feature inclusion, a lidar tile size of 400 feet was selected as the input data extent for each simulated photograph. Figure 26 demonstrates examples of three different tile sizes that were tested for this project. As the tile size increases, more distant features are included in the simulated photo and the number of lidar points used to build the image increases. As a result, more pixels of the fisheye photo along the image perimeter are obscured. This effect is demonstrated by the increase in dark pixels along the bottom left of each simulated photo shown in Figure 26 where tile size increases by 200 feet in each simulation from left to right.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 30
Figure 25: Image matrix showing various sphere radius and voxel size pairs. The selected parameters with the best look and feel compared to field photographs was a 400 foot radius and 2 foot voxel size (shown in the bottom right).
Figure 26: Image matrix showing how the input data tile size changes the resulting simulated image. A tile size of 400 feet was selected to balance the inclusion of features represented in the image and memory limitations.
5.3.3 Products
Products created during the fisheye photo simulation procedure include jpeg images representing simulated hemispherical photography with a 180-degree viewing angle and equiangular distortion.
Fisheye photos were simulated from the lidar point cloud at multiple locations throughout the Ellwood project area. Photos were simulated in a location that matched each field plot center with a point of view at 1.75 meters above ground, 5 meters above ground, and 10 meters above ground.
Photos were also simulated at all historical monarch cluster locations the client had documented.
Monarch cluster photos were simulated with points of view between 20 and 60 feet above ground depending on the recorded height of the monarch cluster.
5.3.3.1 Fisheyes for Field Plots
Fisheye photos were simulated at each field plot center at multiple points of view (POV). • POV 1.75 meters above ground – 30 photos
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 31 • POV 5 meters above ground – 30 photos
• POV 10 meters above ground – 30 photos
Deliverable images are named by plot number, tile size, sphere radius, and voxel size (EX:
‘P01_400ft_400_v2.jpeg’= Plot #1, 400 ft tile size, 400 ft sphere radius, 2ft voxel). The folders indicate their height above ground (EX: ‘Fisheyes_01_75_meters’ indicates 1.75 meters above ground). Table 20 contains a comparison of the fisheye photographs collected in the field and those simulated using the lidar point cloud.
Some features are represented differently between the field and simulated images. For example, powerlines and sparsely leaved branches at the top of the canopy block more view to sky in the simulated photos than the field photos. Additionally, due to a lack of lidar point density in the lower regions of the forest, primarily on stems, the eucalyptus trunks that block light very definitively in the field photos are less present or missing in the simulated fisheye photos.
5.3.3.2 Fisheyes for Monarch Cluster Locations
Fisheye photos simulated from the lidar point cloud at points of view (POV) determined by the attributed height of monarch cluster points.
• When the attributed height of the cluster point was represented by a range, a fisheye photo was generated at each 5-foot interval between the minimum and maximum height listed.
Files are named by their cluster GlobalID and height above ground (EX: ‘0a8017df-eebe-456d-8d75- 030427c9a047__25ft.jepg’ represents a cluster captured with a point of view 25 feet above ground).
Figure 27: Examples of simulated fisheye photos centered on a historic monarch cluster site and moving upwards through the canopy. Left to right images show the camera point of view situated 20 feet, 30 feet, 40 feet, and 50ft above the ground.
6.0 DISCUSSION
The data created as a part of the Ellwood Mesa Monarch Butterfly Habitat Characterization project will facilitate a fine-scale analysis of the eucalyptus grove used as a monarch over-wintering site. The point density rasters, dead debris classification, and fisheye photo simulations will support the assessment, planning, and management of forest rehabilitation. These data will also support future analytical work and provide a baseline for assessing change over time with the collection of future lidar datasets.
The work done for this project has explored a variety of methods for transforming and analyzing lidar and imagery collected across the Ellwood Mesa project area. The deliverable dataset demonstrates the high level of flexibility and analytical opportunity that comes with a high-density three-
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 32 dimensional dataset. A key achievement of this project was the development of a method for simulating fisheye photographs from lidar point clouds. However, the simulated fisheye photos should be evaluated for classification accuracy once the fisheye photos captured in the field have been processed into classified images. The relationship between simulated fisheye photos and field- collected photos needs to be established so that analysts can appropriately interpret and understand the gap fraction and solar exposure values calculated from a simulated photo.
Additionally, the relationship between wind and solar exposure measured from fisheye photos and other lidar-derived indices should be explored to support smooth interpolation of monarch habitat characteristics across the project area. Future work could expand upon the methods developed in this project to create datasets that are planning-ready and would require fewer client-side analytical resources. One opportunity to stream-line data analysis is to investigate the capacity for open source software to programmatically analyze simulated fisheye photos. This would allow for the provision of simulated fisheye photos set at any desired point of view and processed information regarding canopy gap fraction and solar exposure potential at each location.
A future area of investigation is to understand whether similar products and results can be generated with lower density lidar. The UAS collected lidar had excellent canopy penetration and high first return densities (average 277.81 points/m2) creating an ideal foundation for analytical work.
Future efforts should study the capabilities of downed woody debris detection and fisheye photo simulation using lower density lidar datasets.
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 33
7.0 FISHEYE PLOT PHOTO COMPARISON
Table 20: Field photo and simulated fisheye photo comparison. Simulated fisheyes are generated from the lidar point cloud at 1.75-meter height above ground. The asterisk (*) on a plot number indicates that a confident plot geolocation adjustment could not be made and that there is not confidence in the simulated fisheye results.
Plot
Number Field Photo Simulated Fisheye Photo
1 2
3 ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 34 Plot
Number Field Photo Simulated Fisheye Photo
4 5
6 ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 35 Plot
Number Field Photo Simulated Fisheye Photo
7 8
9 ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 36 Plot
Number Field Photo Simulated Fisheye Photo
10 11
12 ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 37 Plot
Number Field Photo Simulated Fisheye Photo
13 14
15 ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 38 Plot
Number Field Photo Simulated Fisheye Photo
16 17
18 ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 39 Plot
Number Field Photo Simulated Fisheye Photo
19 20*
21*
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 40 Plot
Number Field Photo Simulated Fisheye Photo
22 23
24*
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 41 Plot
Number Field Photo Simulated Fisheye Photo
25*
26 27
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 42 Plot
Number Field Photo Simulated Fisheye Photo
28 29
30*
ELLWOOD MESA MONARCH BUTTERFLY HABITAT CHARACTERIZATION 2021 NV5.COM/GEOSPATIAL | 43
8.0 GLOSSARY
1-sigma (σ) Absolute Deviation: Value for which the data are within one standard deviation (approximately 68th percentile) of a normally distributed data set.
1.96 * RMSE Absolute Deviation: Value for which the data are within two standard deviations (approximately 95th percentile) of a normally distributed data set, based on the FGDC standards for Non-vegetated Vertical Accuracy (NVA) reporting.
Accuracy: The statistical comparison between known (surveyed) points and laser points. Typically measured as the standard deviation (sigma ) and root mean square error (RMSE).
Absolute Accuracy: The vertical accuracy of lidar data is described as the mean and standard deviation (sigma σ) of divergence of lidar point coordinates from ground survey point coordinates. To provide a sense of the model predictive power of the dataset, the root mean square error (RMSE) for vertical accuracy is also provided. These statistics assume the error distributions for x, y and z are normally distributed, and thus we also consider the skew and kurtosis of distributions when evaluating error statistics.
Relative Accuracy: Relative accuracy refers to the internal consistency of the data set; i.e., the ability to place a laser point in the same location over multiple flight lines, GPS conditions and aircraft attitudes. Affected by system attitude offsets, scale and GPS/IMU drift, internal consistency is measured as the divergence between points from different flight lines within an overlapping area. Divergence is most apparent when flight lines are opposing. When the lidar system is well calibrated, the line-to-line divergence is low (<10 cm).
Root Mean Square Error (RMSE): A statistic used to approximate the difference between real-world points and the lidar points. It is calculated by squaring all the values, then taking the average of the squares and taking the square root of the average.
Data Density: A common measure of lidar resolution, measured as points per square meter.
Digital Elevation Model (DEM): File or database made from surveyed points, containing elevation points over a contiguous area. Digital terrain models (DTM) and digital surface models (DSM) are types of DEMs. DTMs consist solely of the bare earth surface (ground points), while DSMs include information about all surfaces, including vegetation and man-made structures.
Intensity Values: The peak power ratio of the laser return to the emitted laser, calculated as a function of surface reflectivity.
Nadir: A single point or locus of points on the surface of the earth directly below a sensor as it progresses along its flight line.
Overlap: The area shared between flight lines, typically measured in percent. 100% overlap is essential to ensure complete coverage and reduce laser shadows.
Pulse Rate (PR): The rate at which laser pulses are emitted from the sensor; typically measured in thousands of pulses per second (kHz).
Pulse Returns: For every laser pulse emitted, the number of wave forms (i.e., echoes) reflected back to the sensor.
Portions of the wave form that return first are the highest element in multi-tiered surfaces such as vegetation. Portions of the wave form that return last are the lowest element in multi-tiered surfaces.
Real-Time Kinematic (RTK) Survey: A type of surveying conducted with a GPS base station deployed over a known monument with a radio connection to a GPS rover. Both the base station and rover receive differential GPS data and the baseline correction is solved between the two. This type of ground survey is accurate to 1.5 cm or less.
Post-Processed Kinematic (PPK) Survey: GPS surveying is conducted with a GPS rover collecting concurrently with a GPS base station set up over a known monument. Differential corrections and precisions for the GNSS baselines are computed and applied after the fact during processing. This type of ground survey is accurate to 1.5 cm or less.
Scan Angle: The angle from nadir to the edge of the scan, measured in degrees. Laser point accuracy typically decreases as scan angles increase.
Native Lidar Density: The number of pulses emitted by the lidar system, commonly expressed as pulses per square meter.