SoCal EcoServe - Methods

Running SoCal EcoServe

After launching the SoCal EcoServe tool the map window on the left side of the screen shows USDA Forest Service (USFS) boundaries in green and the larger project area boundary in orange. Using the pulldown menu in the top left of the screen the user can change the background image and add watershed boundaries (USGS Hydrological Unit 12).

The right side of the screen shows a panel with 3 sequential steps with pulldown arrows to:

Select the year of the fire
Select the fire
Select one of six ecosystem services, or hotspots of change

The green question marks associated with the steps link to information on data and methods (see below).

Step 1. Select year

First, select the year of the target fire. SoCal EcoServe contains selected wildfires that burned on (or partially on) the Angeles, Cleveland, Los Padres, and San Bernardino national forests in the past five years. Note, however, that data on water runoff, groundwater recharge, sediment erosion, and carbon storage are available from 2000 to present. If you are interested in using these data, contact us (charlie.schrader-patton@usda.gov) .

Step 2. Select Fire

The pulldown menu lists the fires that occurred in the year specified in Step 1. The intention is that new fires will be included when the tool is updated. Most fires have two options in the dropdown list, one for the entire fire, and the other for just the area inside national forest lands listed with "(USFS only)" after the fire name. In cases where the fire spans national forest and adjacent lands, the user can select whether to query and report ecosystem services for the entire fire or just the portion on national forest land. Once the fire is selected the map in left panel refreshes to the fire extent. Please contact us if there is a specific wildfire that you are interested in viewing data for.

Step 3. Select ecosystem services

The pulldown menu lists the ecosystem services available to query for the target fire. The ecosystem services include: biodiversity, carbon storage, groundwater recharge, recreation, sediment export, and water runoff. If recreation is not listed, then the fire does not encompass any recreation sites (in our databases) within it. Hyperlinks allow Methods to be accessed which describe pre- and post-fire calculations (except for biodiversity where fire impacts are not accounted for), and relevant references. We also provide a hotspots of change in ecosystem services raster.

Table and graphic outputs

For water runoff, groundwater recharge, sediment erosion, and carbon storage services a table reports data on the quantity of that service pre- and post-fire and also a toggle button to switch between table to bar chart format. Where post-fire calculations are incorporated for hydrological and sediment erosion services using the percent canopy cover class lost due to fire (RAVG data, see below). The table includes data reported by these classes.

A table of the recreation sites (e.g., campsite, trailhead, day use, etc.) is provided and, for select sites from the USFS National Visitor Use Monitoring Survey (NVUM), it includes the predicted number of visitors a year, which has a corresponding bar chart. In the table for biodiversity services we summarize different biodiversity values and input data for the target fire as well as values for the four national forests for comparison (we do not assess the impact of fire on biodiversity).

Data download options include: ‘table’, ‘chart data’, ‘spatial data’, and ‘map’.

The ‘spatial data’ link downloads all the spatial data relating to the target fire regardless of which service is currently being queried, including: fire perimeter, the RAVG burn severity data, biodiversity (overall score and some of the individual biodiversity layers), and recreation (list of sites within perimeter). For water runoff, groundwater recharge, sediment export, and carbon storage the ‘spatial data’ option downloads geotiff rasters of the services in pre-fire condition, post-fire condition, and the change between pre-fire and post-fire (calculated as pre-fire minus post-fire). Spatial data are associated with each fire using the ‘Fire Descriptor’ which is the original fire name assigned to the fire. Details are described in the ReadMe file associated with the spatial data download.

The ‘table’ option downloads data in comma separated value (csv) format for the current service being queried. The ‘chart data’ option downloads the data associated with the chart of the current service, also in csv format For water runoff, groundwater recharge, sediment export, and carbon storage it provides the pre-fire year before, pre-fire average, and post-fire data; for recreation this csv file lists the NVUM sites within the fire perimeter and total number of visits per year, and for biodiversity it lists the proportion of the fire with high, medium, and low biodiversity scores (see Biodiversity below).

The ‘map’ option describes how a user can save the map in the left panel by using the print screen function on your computer and copy and paste it into PowerPoint or Word to crop.

Map viewer

For water runoff, groundwater recharge, sediment erosion, and carbon storage the user has the option to view maps of pre-fire and post-fire data by changing the radio buttons located at bottom right of the map view (the same class breaks are used for both pre-fire and post-fire maps to allow for easy comparison). A map of canopy cover loss (RAVG, see below) is also available; this layer was used in the estimates of fire impacts on water runoff, groundwater recharge, and sediment erosion services. For recreation, the map viewer displays recreation sites within the fire perimeter and for biodiversity, the map viewer shows an overall biodiversity score along with four other spatial datasets (see descriptions of individual services below for details).

Burn severity data

To estimate the impact of the fire on three of the ecosystem services 1 year post-fire (water runoff, groundwater recharge, and sediment export), as well as on carbon stored in litter, we integrated data on burn severity from the USDA Forest Service Rapid Assessment of Vegetation Condition after Wildfire (RAVG) https://fsapps.nwcg.gov/ravg . RAVG data are created for all fires >1,000 acres (405 ha) on federal lands.

From the RAVG data we used the five classes of the percent loss of canopy cover derived from the Relative Differenced Normalized Burn Ratio (RDNBR) calculated from pre- and post-fire Landsat imagery (Miller and Thode 2007). The five RAVG canopy cover classes were cross-walked to a single value for each class (0%, 20%, 40%, 60%, and 80%, see Table 1). The use of these values rather than the mid-point of the RAVG classes allows even intervals between the classes and a separate 0% class. In some cases, these cross-walked estimates of canopy cover loss will mean our estimate of the burn severity impact on services will be conservative. The canopy cover loss single values are then converted and interpreted as percent canopy cover values for integrating into the hydrology and sediment erosion models and biomass data.

Table 1. Cross-walk between RAVG percent change in canopy cover loss class and the percent canopy cover integrated into hydrology and sediment erosion models, and biomass calculations.

RAVG canopy cover loss class	RAVG canopy cover loss, class	Canopy cover loss, single value	Conversion and interpretation as % canopy cover

1	0% change	0%	100%
2	0% to <25% change	20%	80%
3	25% to <50% change	40%	60%
4	50% to <75% change	60%	40%
5	75% to 100% change	80%	20%

Climate influences on services

Weather in southern California is highly variable (Fig. 1), which drastically affects hydrological responses each year as water runoff and groundwater recharge depend on precipitation, evapotranspiration, soils, and geology. The removal of vegetation with combustion and reduced transpiration should increase hydrological services such as water runoff and groundwater recharge, however when drought is severe runoff and recharge can decrease post-fire. For runoff, recharge and sediment erosion services we averaged model outputs over 12 years (selected from 2000-2021) which captured a range of drought and normal climate conditions as indicated by the Palmer Drought Severity Index for the south coast ecoregion of California. Using the 12-year average values for pre- and post-fire, shows these services under normalized climate conditions. This approach has the advantage of facilitating comparisons of fires occurring in different years; increasing the influence of burn severity (RAVG) data over climate in determining fire impacts, and allowing fire impacts to be calculated as soon as burn severity data are available.

Annual Precipitation X Mean Annual Temperature Graph

Fig. 1. Plot of annual precipitation and mean annual temperature for southern California for water years 2000-2019.

Methods for each Ecosystem Service

Water runoff and groundwater recharge

Data on water runoff and groundwater recharge are extracted from the Basin Characterization Model (BCM), a statewide water balance model (270 m resolution) (Flint et al. 2013). This assessment undertook hydrological modeling for the years 2001-2021 using precipitation and temperature data specific to each of those years.

· Pre-fire values (12 year average): Correspond to modeled outputs that assume undisturbed vegetation condition (Flint et al. 2013). Pre-fire estimates of water runoff and groundwater recharge represent an average generated from the BCM outputs for 12 different water years (which run from October to September and precede the calendar year) (see Climate Influences on Services section).

· Post-fire values (12 year average): We reran ran the BCM to integrate the fire-induced percent post-fire canopy cover loss data from RAVG (see ‘Burn severity data’ section above) for each target fire. To reduce the influence of year to year variation in drought conditions, we calculated the average of each BCM run using the canopy cover loss single value (i.e., 100%, 80%, 60%, 40% , 20%, 0%) over the 12 years that captured a range of drought conditions from 2000-2021. We report the normalized post-fire runoff and recharge quantities using the 12-year average outputs based on a target fire’s canopy cover loss in it’s RAVG data.

The maps in the viewing window show the value of the service per unit area. The table displays the total amount of the service summed over the fire area (i.e., volumetric total).

Caveats

· While other hydrological models may outperform the BCM at finer spatial scales such as plots, the BCM is widely accepted to be the most appropriate model for applications at the size of planning watersheds.

· The maps optimize display of spatial variation in the service by using distinct colors for classes which have equal areas. The same class breaks are used for both pre-fire and post-fire maps to allow for easy comparison. The class breaks are determined based on the pre-fire data, which will emphasize variability in the pre-fire versus post-fire data. Note class breaks are different between fires, so direct comparisons of legends of different fires is not possible. Map comparisons between different fires will require downloading the spatial data and creating layers with GIS software.

· In some cases pre-fire recharge and runoff can be substantially different (e.g. for the Thomas fire (2017, USFS boundary). Recharge is “greater than runoff in xeric climates because of the high evaporative demand and storage space in soils for water“ (Flints et al. 2021, pg 7) , i.e.., precipitation first goes into recharge before showing up as runoff if the groundwater is in deficit. In the case of the Thomas fire, 2017 had average precipitation; 2016 slightly drier; but 2014 and 2015 were drought years. Consequently, in 2017 more water goes into recharge than into runoff.

References

Flint, L. E., E. C. Underwood, A. L. Flint, and A. D. Hollander. 2019. Characterizing the influence of fire on hydrology in southern California. Natural Areas Journal 39(1):108-121.

Flint, L.E., Flint, A.L., and Stern, M.A. 2021. The basin characterization model—A regional water balance software package: U.S. Geological Survey Techniques and Methods 6–H1, 85 p., https://doi.org/10.3133/tm6H1. ISSN: 2328-7055 (online).

Carbon storage

The project developed biomass data for southern California (30 m spatial resolution) as shrubland biomass has generally not been adequately mapped in national and state-wide data sources (Schrader-Patton and Underwood 2021).

To predict above ground live biomass (AGLBM) across the study area, we compiled a suite of predictor raster surfaces at 30 m spatial resolution including: elevation, slope, aspect, and soil type, as well as annual precipitation, climatic water deficit, water runoff, groundwater recharge, and evapotranspiration. In addition, we used the Normalized Vegetation Index (NDVI) images processed from Landsat 5, 7, and 8 data for the months of July and August for each year between 2001 and 2021. A composite image of the maximum NDVI values for all Landsat images for these two months minimized pixels contaminated by clouds. We used a total of 959 plots, and extracted the values from each raster spatial layer after carefuly matching the plot visit year to the NDVI year. Plots included estimates of biomass based on allometric equations for US Forest Service Forest inventory and Analysis (FIA shrubland plots (688 plots), the Landfire Program’s Landfire Reference Database (LFRDB; 276 plots) and various academic studies (15 plots) (e.g., Uyeda et al. 2016). For species without an allometric equation, we used a general equation for shrub species from Lutes et al. (2006). Using a Random Forest ensemble machin learning algorithm, we mapped biomass across the study area using the Landsat NDVI data as the time-dependent predictor in the model.

Using AGLBM from the plots as the dependent variable and the predictor environmental data layers as the explanatory or independent variable, we built a regression model using the Random Forest ensemble machine learning algorithm. The maximum AGLBM for July/August of any given year was predicted across the study area using the prediction raster stack and the NDVI images for July/August of that year. We refer to this as the July/August AGLBM for a given year. This was done for each year between 2000 and 2021. For the NDVI raster layer, we used the months of July/August as they best capture information about shrubland biomass as many herbaceous species, e.g., non-native grasses, have senesced by the summer months (Henry and Hope, 1998). We cross-referenced the modeled biomass estimates with field estimates summarized in Bohlman et al. (2018) to ensure they approximated conditions on the ground (Schrader-Patton and Underwood 2021).

In addition to AGLBM data, we estimate other biomass pools which are not detected from our image-based biomass estimates. We used published studies from southern California shrublands to estimate the amount of aboveground dead, aboveground litter, and belowground biomass (Schrader-Patton and Underwood 2022). Often these pool estimates were factors of our AGLBM estimates.

The wildland vegetation in the study area was divided into two groups based on their spatial area and using 0.75% of the study area as a threshold size. Shrubland types that individually are >0.75% of the study area are Mixed Chaparral, Chamise-Redshank, Coastal Scrub, Sagebrush, Montane Chaparral, and Desert Scrub. Non–shrubland types that are individually >0.75% include Juniper, Pinyon-Juniper, Sierran Mixed Conifer, Blue-Oak Woodland, Coastal Oak Woodland, Montane Hardwood, Montane Hardwood–Conifer, and Annual Grasslands.

Estimating these biomass pools was a two-step process: first determine the proportions of plant groups within each pixel and then applying a separate workflow for each one. In all pixels where shrublands are >0.75% we identify the proportion of each pixel that is one of three shrub life history types: Seeder (S), Facultative Seeder (FS), Resprouter (R), along with tree and herb life forms. We then implement a different workflow to estimate different biomass pools for each of these five groups; the majority of these use AGLBM as a foundation.

Pre- and post-fire carbon storage values in the EcoServe tool are assigned as follows and detailed in this schematic:

Pre-fire values :

· Aboveground live biomass (AGLBM) : We determine AGLBM using the maximum NDVI pixel value from all Landsat images collected in July/August. This time period was selected because it is after the primary shrubland growing season (March-June) and is when most grasses and forbs have senesced. We then determine which year of AGLBM to report for each target fire based on two fire start periods: January 1-September 1 and September 2-December 31. For fires that start between January 1 and September 1 we use the July/August AGLBM from the previous (calendar) year as the previous year represents pre-fire conditions. For fires that burn between September 2 and December 31 we use the July/August AGLBM of the current (calendar) year. For example, if the fire started on September 15 2020 we report July/August AGLBM for 2020, however if the fire started on June 15 2020 then we report July/August AGLBM for 2019. We then multiply the proportion of each of the five plant groups (Seeder, Facultative Seeder, Resprouter, tree and herb) within a pixel by AGLBM.

· Standing dead biomass : For the three shrub life history types (S, FS, and R) , we estimate standing dead biomass from published studies from southern California shrublands (Green 1970; DeBano and Conrad 1978; Regelbrugge and Conard 1996; Riggan et al. 1988, and the Natural Fuels Photo Series) https://www.fs.usda.gov/pnw/projects/digital-photo-series . Using the average value of standing dead biomass from the published studies, AGLBM is multiplied by 0.184, 0.359, and 0.344 for S, FS, and R respectively. To estimate the standing dead biomass for the tree pixel component, we assign the estimate for snags for the corresponding spatial location from the North American Wildland Fuels Database (2014 and 30 m spatial resolution; Pritchard et al. 2019). For the herb component, we multiply the proportion of herbs in the pixel by the NAWFD litter values. For pixels assigned to non-shrubland types that are >0.75% of the study area (e.g., annual grasslands, woodlands, mixed conifer, Pinyon-Juniper) we also assign standing dead and litter values using data from the corresponding spatial location in the NAWFD. NAWFD estimates are multiplied by 0.1 to convert from Mg/ha to kg/m2 for the SoCal EcoServe application.

· Belowground biomass : For each of the five plant groups, AGLBM is multiplied by a root to shoot proportion determined by findings in the literature. Proportions are 0.41, 1.26 and 2.41 for S, FS, and R respectively (Davis 1977; Miller and Ng 1977; Kummerow et al. 1977; Kummerow and Mangan 1981). For the tree proportion present in a pixel, AGLBM is multiplied by 0.302 which is the mean of all root to shoot ratios (all shoot biomass classes) for temperate oak, conifer, and other temperate broadleaf classes (Mokany et al. 2006). For belowground biomass associated with herbs, their pixel proportion is multiplied by 1.09 (Koteen et al. 2011). This is the average for non-native annual grass species as we assume the majority of grasslands in southern California are non-native (Park et al. 2018, Franklin 2002). For pixels assigned to non-shrubland types that are > 0.75% of the study area (annual grasslands, woodlands, mixed conifer, montane chaparral) we assign belowground biomass values from the California Air Resource Board (CARB) data (Battles et al. 2014; Gonzalez et al. 2015), as NAWFD does not include belowground estimates. CARB estimates were multiplied by 0.1 to convert from Mg/ha to kg/m2 for the SoCal EcoServe application.

· Litter biomass : We calculate litter biomass for mixed chaparral, chamise chaparral, and coastal sage scrub based on the ratio of litter to aboveground live biomass as estimated in Bohlman et al. (2018) for mixed chaparral (note; chamise redshank chaparral and coastal sage scrub types had 2 or fewer studies to inform them so did not provide sufficient data). Based on the Bohlman et al. (2018) mixed chaparral data, we multiply AGLBM by 0.78 for these three shrubland types. To estimate the litter biomass for trees and herbs present in shrubland-dominated pixels, we multiply the estimates of the proportion of trees and herbs by the litter values from NAWFD. For the remaining shrubland types that are >0.75% of the study area (sagebrush, desert scrub, and montane chaparral) we used the NAWFD litter values. For pixels assigned to non-shrubland types that are >0.75% of the study area (e.g., annual grasslands, woodlands, mixed conifer, Pinyon-Juniper) we also assign litter values from NAWFD. NAWFD estimates were multiplied by 0.1 to convert from Mg/ha to kg/m2 for the SoCal EcoServe application.

· Pre-fire values (7-year average) : represents the average biomass value from all vegetative pools for 2001 – 2007. We selected this period as these years were relatively normal climate years, without being dominated by drought or overly wet years (see Fig. 1).

For the 31 remaining wildland vegetation types covering < 0.75% of the study area were assigned biomass as follows:

Aboveground live biomass: values from the WWETAC-UCD data
Standing dead biomass: values from NAWFD snag biomass
Litter biomass: values from NAWFD litter biomass
Belowground biomass: values from CARB 2014 (no belowground biomass estimates made in NAWFD)

Post-fire values 1 year after :

We report post-fire values for the ‘year’ following the target fire, however, the length of this post-fire period varies depending on the date of the fire. Note that we use AGLBM measurements from July/August as the primary growing season is March through June, and there are two fire start periods (January 1-September 1 and September 2-December 31). Consequently, for a for a target fire that started on September 15 2020 the post-fire period would be almost one year (September 15 to July/August) but for a fire that started on June 15 2020 then the post-fire period would be about a month (June 15 to July/August).

· Aboveground live biomass : We report the AGLBM from July/August for the year following the fire if the fire burned between September 2-December 31. For fires that burned January 1-September 1 the post-fire AGLBM year is the same as the fire year.

· Standing dead biomass: Post-fire standing dead values are calculated using the post-fire AGLBM layer and the same methods described to calculate pre-fire values (see above section).

· Belowground biomass: Post-fire belowground values are calculated using the post-fire AGLBM layer and the same methods described to calculate pre-fire values (see above section).

· Litter biomass: Post-fire litter values are calculated using the post-fire AGLBM layer and the same methods described to calculate pre-fire values (see above section).

The amount of biomass from all pools is converted to carbon by multiplying by 0.47 (approximately half of woody AGLBM is carbon [McGroddy et al. 2004]). Pre- and post-fire carbon storage is reported in metric tons in the table (reported by different pools) and bar chart (total carbon).

Shrub Types Within the Fire Boundary

To help interpret the table of Carbon Lost Over the Long-Term and select the appropriate Estimate of Carbon Recovery graph, we provide a table detailing the area (in hectares, acres, and the proportion) of the major shrub types within the selected fire. Note that these do not sum to 100% because we only list the three major shrub types (https://wildlife.ca.gov/Data/CWHR/Wildlife-Habitats):

(a) Mixed chaparral: dominated by scrub oak (Quercus berberidifolia), ceanothus (Ceanothus) and manzanita (Arctostaphylos).

(b) Chamise redshank chaparral: dominated by chamise (Adenostoma fasciculatum) and/or redshank ( Adenostoma sparsifolium );

(c) Coastal sage scrub: dominated by California sagebrush (Artemisia californica), purple sage ( Salvia leucophylla ), black sage (Salvia mellifera), and California buckwheat (Eriogonum fasiculatum)

Carbon Lost Over the Long-Term

When shrublands are within the historic fire return interval, e.g., 55 years for low-elevation shrubland (Keeley and Safford 2016, Van de Water and Safford 2011), biomass accumulates and shrub cover recovers after 10–14 years (Black et al. 1987, Bohlman et al. 2018). However, in many parts of southern California, the fire return interval has decreased, often in conjunction with an increase in non-native plant species, drought, and nitrogen deposition (Pratt et al. 2014, Allen et al. 2018, Syphard et al. 2019). Under these conditions, post-fire biomass recovery can be impeded and, in some cases, may result in type conversion from native shrubland to non-native grassland (Syphard et al. 2019). .

To capture the amount of biomass and carbon that is permanently lost post-fire, we identify shrubland pixels with low-regeneration potential based on two rules guided by the published literature (Zedler et al. 1983, Haidinger and Keeley 1993, Keeley and Brennan 2012, Syphard et al. 2018, Syphard et al. 2019, Underwood et al. 2021, Underwood and Safford 2021): (1) where there has been 3 or more fires in the last 40 years and (2) where time since last fire is <10 years (Underwood and Hollander 2023). For these low regeneration pixels, we assume shrubland biomass will not recover and will be replaced by annual (non-native) grasses. Accordingly, we assign a biomass value of 0.14 kg /m2 based on a query of the AGLBM data (2000-2021) in areas classified as annual grasslands (FRAP 2015) and cross-reference to the published literature on biomass values for grasslands (e.g., Bradley et al. 2006, Wang et al. 2014). We calculate the amount of biomass lost in these low regeneration pixels as the difference between the pre-fire shrubland AGLBM value and the post-fire annual (non-native) grass value. The user can compare the amount of carbon lost in shrub vegetation to the amount of carbon pre-fire. For example, for the Thomas fire (2017): pre-fire carbon in shrub vegetation was 1,945,741 metric tons and carbon lost was 54,138 metric tons, so the amount of carbon lost in shrub vegetation is about 2%. Alternatively, for the Cave fire (USFS only, 2019), the amount of carbon lost in shrub vegetation is 25%. These values are converted to carbon and reported in a table which summarizes pre- and post-fire carbon by the three main shrub community types.

Estimate of Carbon Recovery

To estimate the recovery of biomass after fire, we generated a series of plots in fires that burned in 2004 to analyze. We selected fires from 2004 as this year had a large spatial area burned and with the available stack of AGLBM layers from 2000-2020, it provided 4 years pre-fire and 16 years post-fire to explore biomass recovery. Using paired plots in burned and unburned areas removed the annual variability observed in biomass levels driven by climate. Other requirements were that burned plots had only burned once in 40 years and also burned at high intensity, as indicated by the RAVG canopy cover loss data. A total of 152 burned plots and 190 unburned plots were developed in the three major shrub community types: mixed chaparral, chamise-redshank chaparral, and coastal sage scrub. To assess the recovery of biomass, we focused on the % change in biomass from the first biomass year of 2000 on a plot-by-plot basis. This removes differences in the amount of biomass based on varying productivity of each site.

To guide the user on which recovery graph to focus on, the tool provides a table of the area and proportion of each of the three main shrub community types within the fire perimeter: mixed chaparral, chamise redshank chaparral, and coastal sage scrub ( https://wildlife.ca.gov/Data/CWHR/Wildlife-Habitats , see descriptions of each under 'Carbon lost over the long term' section).

Caveats

· Confidence in these estimates vary depending on the data source, for example, aboveground live biomass (AGLBM) was cross-validated with field data. However, inputs into estimating other carbon pools are based on the literature which can be limited in geographic scope or from other datasets with a coarse spatial resolution (e.g., NAWFD). These estimates have not been field validated.

· For some biomass pools, such as belowground, the temporal scale of fire impacts is unknown.

· Biomass in duff has not been accounted for owing to lack of data.

· Biomass in litter is likely to be overestimated in the immediate postfire year, as seeds of obligate seeding shrubs and fire-following annuals are stimulated by fire cues to germinate (Keeley 1987, Keeley and Keeley 1987) which would have less litter fall than shrubs.

· The maps optimize display of spatial variation in the service by using distinct colors for classes which have equal areas. The same class breaks are used for both pre-fire and post-fire maps to allow for easy comparison. Note class breaks are different between fires, so direct comparisons of legends of different fires is not possible. Map comparisons between different fires will require downloading the spatial data and creating layers with GIS software.

· In assessing long-term recovery of biomass by shrub community type, users must be aware that the accuracy of these estimates is underlain by the accuracy of the mapping of shrubland vegetation communities (FRAP 2015), as this guided the location where plots were placed for the biomass recovery analysis.

· The influence of drought on biomass should be noted. The years 2010 to 2020 were primarily drought hears, so biomass recovery in unburned plots might be disproportionately affected.

References

Allen, E. B., K. Williams, J. L. Beyers, M. Phillips, S. Ma, and C. M. D’Antonio. 2018. Chaparral restoration. In Underwood, E. C, H. D. Safford, N. A. Molinari, and J. E. Keeley (editors). Valuing Chaparral: Ecological, Socio-Economic, and Management Perspectives. Springer Series on Environmental Management. Springer, Cham, Switzerland.

Battles, J., P. Gonzalez, T. Robards, B. Collins, and D. Saah. 2014. California Forest and Rangeland Greenhouse Gas Inventory Development. Final Report. California Air Resources Board, Sacramento, California, USA. Pp 46.

Black, C. H. 1987. Biomass, nitrogen, and phosphorus accumulation over a southern California fire cycle chronosequence. In Tenhunen, J.D., F. M. Catarino, O. L. Lange, and W. C. Oechel (Eds). Plant response to stress. NATO ASI Series G (vol 15): Ecological Sciences. Springer. Berlin/Heidelberg, Germany.

Bohlman, G. N., E. C. Underwood, and H. D. Safford. 2018. Estimating biomass in California’s chaparral and coastal sage shrublands. Madroño 65(1): 28-46.

Bradley, B. A., R. A. Houghton, J. F. Mustard, and S. P. Hamburg. 2006. Invasive grass reduce aboveground carbon stocks in shrublands of the Western US. Global Change Biology12: 1815–1822. doi: 10.1111/j.1365-2486.2006.01232.x

Davis, E. A. 1977. Root system of shrub live oak in relation to water yield by chaparral. Hydrology and Water Resources in Arizona and the Southwest Vol 7: 241-248.

Debano, L. F. and C. E. Conrad. 1978. The effect of fire on nutrients in a chaparral ecosystem. Ecology 59(3): 489-497.

Debano, L.F. 1990. Effects of fire on the soil resource in Arizona chaparral. In Krammes, J. S. (tech. coord). Effects of fire management of southwestern natural resources. Gen. Tech. Rep. RM-191. USDA Forest Service, Rocky Mountain Forest and Range Experiment Station. Fort Collins, Colorado

Franklin, J. 2002. Enhancing a regional vegetation map with predictive models of dominant plant species in chaparral. Applied VegetationScience5 : 135–146. https://doi.org/10.1111/j.1654-109X.2002.tb00543.x

FRAP [Fire and Resource Assessment Program]. 2015. California Department of Forestry and Fire Protection’s CALFIRE Fire and Resource Assessment Program (FRAP). Fveg15_1 vegetation data. http://frap.fire.ca.gov/data/frapgisdata-sw-fveg_download .

Gonzalez, P., J. Battles, B. Collins, T. Robards, and D. Saah. 2015. Aboveground live carbon stock changes of California wildland ecosystems, 2001–2010. Forest Ecology and Management 348. https://doi.org/10.1016/j.foreco.2015.03.040

Green, L. R. 1970. An experimental prescribed burn to reduce fuel hazard in chaparral. Research Note PSWRN-216. U.S. Department of Agriculture, Forest Service, Pacific Southwest Forest and Range Experimental Forest, Berkeley, CA.

Haidinger, T. L. and J. E. Keeley. 1993. Role of high fire frequency in destruction of mixed chaparral. Madroño 40:141-147.

Henry, M. and A. Hope. 1998. Monitoring post-burn recovery of chaparral vegetation in southern California using SPT XS data. International Journal of Remote Sensing 19(16):3097-3107.

Keeley, J. E. 1987. Role of fire in seed germination of woody taxa in California chaparral. Ecology 68 (2): 434–443.

Keeley, J. E. and S. C. Keeley. 1987. Role of fire in the germination of chaparral herbs and suffrutescents. Madrono 34: 240–249.

Keeley, J. E., and T. J. Brennan. 2012. Fire-driven alien invasion in a fire-adapted ecosystem. Oecologia 169(4):1043-1052. https://doi.org/10.1007/s00442-012-2253-8.

Keeley, J. E. and H. D. Safford. 2016. Fire as an ecosystem process: Chapter 3. In Mooney, H. A., and E. S. Zavaleta (Eds). Ecosystems of California . University of California Press. Oakland, California, USA.

Koteen, L. E., Baldocchi, D. D. and J. Harte. 2011. Invasion of non-native grasses causes a drop in soil carbon storage in California grasslands. Environmental Research Letters 6(4): 044001.

Kummerrow, J. D. Krause, and W. Jow. 1977. Root Systems of Chaparral Shrubs. Oecologia 29, 163-177.

Kummerow and Mangan. 1981. Root systems in Quercus Dumosa dominated chaparral in southern California. Acta Oecologica 2(16), 177-188.

Lutes, D. C., R. E. Keane, J. F. Caratti, C. H. Key, N. C. Benson, S. Sutherland, and L. J. Gangi. 2006. FIREMON: Fire effects monitoring and inventory system. General Technical Report RMRS-GTR-164-CD. USDA Forest Service, Rocky Mountain Research Station. Fort Collins, Colorado.

McGroddy, M.E., T. Daufresne, and L. O. Hedin. 2004. Scaling of C:N:P stoichiometry in forests worldwide: implications of terrestrial Redfield‐type ratios. Ecology 85: 2390–2401.

Miller, P. C. and E. Ng. 1977. Root:shoot biomass ratios in shrubs in southern California USA and central Chile. Madrono 24:215–223.

Miller, J. D. and A. E. Thode. 2007. Quantifying burn severity in a heterogeneous landscape with a relative version of the delta Normalized Burn Ratio (dNBR). Remote Sensing of the Environment 109:66-80.

Mokany, K., J.R. Raison, and A.S. Prokushkin. 2006. Critical analysis of root:shoot ratios in terrestrial biomes. Global Change Biology 12: 84-96..

Park, I., J. Hooper, J. Flegal, and G. Jenerette. 2018. Impacts of climate, disturbance and topography on distribution of herbaceous cover in Southern California chaparral: Insights from a remote-sensing method. Diversity and Distributions 24. DOI:10.1111/ddi.12693.

Prichard, S. J., M. C. Kennedy, A. G. Andreu, P. C. Eagle, N. H. French, and M. Billmire. 2019. Next‐generation biomass mapping for regional emissions and carbon inventories: Incorporating uncertainty in wildland fuel characterization. Journal of Geophysical Research: Biogeosciences, 124. https://doi.org/10.1029/2019JG005083

Pratt, R. B., A. L. Jacobsen, A. R. Ramirez, A. M. Helms, C. A. Traugh, M. F. Tobin, M. S. Heffner, and S. D. Davis. 2014. Mortality of resprouting chaparral shrubs after a fire and during a record drought: Physiological mechanisms and demographic consequences. Global Change Biology20: 893–907.

Regelbrugge, J. C. and S. G. Conard. 1996. Biomass and fuel characteristics of chaparral in southern California. 13th Fire and Forest Meteorology Conference. Lorne, Australia.

Riggan, P. J., S. Goode, P. M. Jacks, and R. N. Lockwood. 1988. Interaction of fire and community development in chaparral of southern California. Ecological Monographs 58(3): 155-176.

Schrader-Patton, C. and E. C. Underwood. 2021. New biomass estimates for chaparral-dominated landscapes. Remote Sensing, 13, 1582.

Schrader-Patton, C.C. and E.C. Underwood. 2022. Annual biomass data (2001-2021) for southern California: above- and below-ground, standing dead, and litter. Dryad, Dataset, https://doi.org/10.5061/dryad.qz612jmjt

Syphard, A. D., T. J. Brennan, and J. E. Keeley. 2018. The ecological chaparral landscape conversion in Southern California. Pages 323-346 in Underwood, E. C., H. D. Safford, N. A. Molinari, and J. E. Keeley, editors. Valuing chaparral: ecological, socio-economic, and management perspectives. Springer Series on Environmental Management. Springer, Cham, Switzerland.

Syphard, A. D., T. J. Brennan, and J. E. Keeley. 2019. Extent and drivers of vegetation type conversion in southern California chaparral. Ecosphere 10: e02796

Underwood, E. C., A. D. Hollander, N. A. Molinari, L. Larios, and H. D. Safford. 2021.. Identifying priorities for post-fire restoration in California chaparral shrublands. Restoration Ecology doi: 10.1111/rec.13513

Underwood, E. C. and H. D. Safford. 2021. Appendix 5: Postfire restoration prioritization tool for chaparral shrublands. Pages 183-185 in Meyer, M.D., J.W. Long, and H.D. Safford, editors. Postfire restoration framework for national forests in California. Gen. Tech. Rep. PSW-GTR-270. U.S. Department of Agriculture, Forest Service, Pacific Southwest Research Station. Albany, CA. 204p.

Underwood, E.C., and A.D. Hollander. 2023. Areas of low natural regeneration potential post-fire in shrublands of southern California (selected years between 2008 and 2020). Dryad, Dataset, https://doi.org/10.25338/B8CH2T

Uyeda, K., D. Stow, J. O’ Leary, C. Tague and P. Riggan. 2016. Chaparral growth-ring analysis as an indicator of stand biomass development. International Journal of Wildland Fire 25(10).

Van de Water, K. M. and H. D. Safford. 2011. A summary of fire frequency estimates for California vegetation before Euro-American settlement. Fire Ecology7: 26–58.

Wang, L., L. Li, X. Chen, X. Tian, X. Wang, and G. Luo. 2014. Biomass allocation patterns across China’s terrestrial biomes. PLoS ONE 9(4): e93566. https://doi.org/10.1371/journal.pone.0093566

Zedler, P. H., C. R. Gautier, and G. S. McMaster. 1983. Vegetation change in response to extreme events – the effect of a short interval between fires in California chaparral and coastal scrub. Ecology 64(4):809-818. Doi.org/10.2307/1937204

Sediment export (erosion)

Although the ecosystem service is 'sediment erosion retention' we report the amount of sediment ‘eroded’ or ‘exported’. Specific inputs were developed to tailor it to southern California including soil erodibility data from the county scale gridded Soil Survey Geographic Database (gSSURGO) soils map. Data on sediment export (30 m spatial resolution) are derived using the Natural Capital Project’s InVEST sediment erosion module (Hamel et al. 2015). Specific inputs were developed to tailor it to southern California including soil erodibility data from county scale SSURGO soils map; monthly precipitation data from the (hydrological) Basin Characterization Model (BCM); a rainfall erosivity index; and generating a vegetation cover index based on the Landsat-derived Normalized Difference Vegetation Index (NDVI, for 2014-15) (see Underwood et al. 2018 for details). The InVEST outputs provide an estimate of the amount of sediment delivered to a stream.

· Pre-fire values (12 year average): estimates of sediment export are generated on a calendar year basis, the value being taken from the sediment export estimate for the year the fire occurred assuming undisturbed vegetation conditions. Pre-fire estimates of sediment erosion represent an average generated from the InVEST outputs for 12 different years (see Climate Influences on Services section).

· Pre-fire values (30 year average): represents average sediment export data from 1981-2010. Data are generated from the Basin Characterization Model (Flint et al. 2013). This does not account for any landscape disturbances, such as fire, that occurs within this period. The average data is not available to view in the tool, but this raster can be downloaded using the 'Spatial Data' download button.

· Post-fire values (12 year average): We ran the InVEST module to integrate post-fire canopy cover loss data from RAVG (see ‘Burn severity data’ section above) for each target fire. To reduce the influence of year to year variation in drought conditions, we calculated the average of each InVEST run using the canopy cover loss single value (i.e., 100%, 80%, 60%, 40% , 20%, 0%) over the 12 years that captured a range of drought conditions from 2000-2021. We report the normalized post-fire sediment erosion quantity using the 12-year average output based on a target fire’s canopy cover loss in it’s RAVG data.

The maps in the viewing window show the value of the service per unit area. The table displays the total amount of the service summed over the fire area (i.e., volumetric total).

Caveats

· The InVEST sediment erosion module only considers annual precipitation and does not account for the intensity of individual winter storm events that characterize southern California and cause substantial erosion.

· In areas of the ecoregion with exposed rock (e.g., labelled 'acid igneous rock' in SoilWeb datasbse https://casoilresource.lawr.ucdavis.edu/gmap/ , there will be no erodibility factor available in the SSURGO database. Consequently, these areas will appear with zero sediment erosion.

· It should be noted that the loss of soil also reduces ecosystem productivity upslope as well as impacts infrastructure downstream, which are not reflected in this ecosystem services estimate.

References

Hamel, P., R. Chaplin-Kramer, S. Sim, and C. Mueller. 2015. A new approach to modeling the sediment retention service (InVEST 3.0): case study of the Cape Fear catchment, North Carolina, USA. Science of the Total Environment 524-525:166-177.

Underwood, E. C., A. D. Hollander, P. R. Huber, and C. Schrader-Patton. 2018. Mapping the Value of National Forest Landscapes for Ecosystem Service Provision. In Underwood, E. C, H. D. Safford, N. A. Molinari, and J. E. Keeley (editors). Valuing Chaparral: Ecological, Socio-Economic, and Management Perspectives. Springer Series on Environmental Management. Springer, Cham, Switzerland.

Recreation

We report the following sources of recreation data:

· We report ‘Recreation Opportunities’ data on USFS public lands from the USFS Geodata Clearinghouse https://data.fs.usda.gov/geodata/help.php . This national scale recreation database contains spatial data on recreational sites, areas, activities and facilities.

· We also report data on surveyed recreation sites from the USFS National Visitor Use Monitoring Survey (NVUM), USDA Forest Service 2015, English et al. 2002). Started in 2004 the NVUM dataset collects data from visitors in national forests across the US. A detailed 5-minute questionnaire is conducted with visitors at picnic areas, trailheads, visitor centers and parking lots. A study by Garnache and Lupi (2018) used data from the NVUM survey to develop an economic model of visitation to specific national forests in the vicinity of Los Angeles, California, including the Angeles, Cleveland, southern Los Padres, and San Bernardino National Forests. Survey data from 179 sites (including day use areas, wilderness areas, trailheads, and also more general road stop sites) across portions of the four National Forests were used to generate demand equations for each site in the model. The Recreation Demand Model relates travel costs to get to and from a site (e.g., the marginal time and driving costs per mile) to site characteristics such as distance to water and fire history to determine the economic benefit to recreationists from recreating at each site. Since fire histories were found to have only a minor effect on site visitation, the main effects of a fire on recreation value would be through closures. We report the predicted total number of visits per year to each site for any surveyed recreation sites within the fire perimeter. NVUM sites can be identified in the EcoServe table for recreation with a value in the ‘NVUM Site Id’ column, data from the USFS Geodata Clearinghouse are attributed as ‘no data’. Also, the Site Id appears next to associated name label on the map.

Caveats

· Only recreation sites within the fire perimeter are reported since we assume these will be impacted by the fire. However, the tool does not account for recreation sites outside the fire perimeter that could also be affected indirectly by fire (e.g., road closures that prevent access or impacted viewsheds).

· Only developed recreation sites are reported (e.g., campsites, picnic sites). It does not account for dispersed recreation such as hiking in wilderness areas.

· For some fires there are no surveyed recreation sites to report.

· In some cases, the NVUM sites are also ‘recreation opportunity’ sites, but not always. In cases where the recreation site is both an NVUM site and in the Recreation Opportunity database, priority is given to reporting the information in the NVUM site database.

References

English, D., S. Kocis, S. Zarnoch, and J. Arnold. 2002. USDA Forest Service national visitor use monitoring process: research method documentation. General Technical Report SRS-GTR-57. USDA Forest Service, Southern Research Station, Asheville, North Carolina, USA.

Garnache, C., and F. Lupi. 2018. Identifying the effect of fire on the value of forest recreation services, working paper presented at World Congress of Environmental and Resource Economists, Gothenburg Sweden, June 2018.

USDA Forest Service. 2015. National visitor use monitoring: visitor use report. Database queried for: Angeles National Forest (2006, 2011), Cleveland National Forest (2009, 2014), Los Padres National Forest (2009, 2014), and San Bernardino National Forest (2009, 2014). https://apps.fs.usda.gov/nfs/nrm/nvum/results .

Biodiversity

We map and report a composite biodiversity value generated using a conservation planning software, as well as the individual datasets used to generate it. In the table we report the values for the target fire as well as the entire Angeles, Cleveland, Los Padres, and San Bernardino national forests, to permit comparisons.

Marxan Biodiversity Layer
An overall score for biodiversity for the southern California study area was generated with the conservation planning software Marxan (Ball et al. 2009, http://marxan.org ). Guided by USDA Southern California National Forest land management plan (USDA 2004) we identified conservation targets to represent biodiversity (Table 2), including locations of rare and endangered species, wildlife linkages (Huber et al. 2010), native and rare species richness from the statewide Areas of Conservation emphasis (ACEII, CDFW), and natural land cover type (FRAP 2015) (Underwood et al. 2018). For each conservation target we specified a corresponding goal for protection, these were higher for targets where there were fewer records or they covered a smaller spatial area (the assumption being that we prioritize having a diverse spatial set of TES species).

Marxan works by using a simulated annealing algorithm to explore many configurations of planning units, incrementally moving towards solutions that meet inputted conservation objectives in 'low cost' ways. In this study we defined 'cost' as the suitability of a given planning unit for inclusion in a final conservation network. Unlike focusing on a single species or proxy such as intact habitat, this irreplaceability score takes account of multiple biodiversity data inputs and identifies multiple sets of planning units that comprise relatively low cost (i.e., most suitable) solutions to meet a user’s conservation objectives, thereby prioritizing the landscape into a single value.

Using a variety of input data layers (see below) we used analysis units (formed within ecological units for southern California, see Underwood et al. 2018) that ranged in size from 4 - 6,475 ha (10 - 1600 acres). Urban areas and water bodies were omitted. For each unit we developed a cost based on its area and its native species richness and rare species richness scores from the California Department of Fish and Wildlife Areas of Conservation Emphasis (ACEII). Higher richness scores translated to lower costs for the unit, making it more attractive for selection in the Marxan analysis.

The Marxan software determined the irreplaceability of each ecological unit in the southern California study area using these biodiversity targets and goals. A value from 0-100 was assigned, representing the number of runs in which each unit is selected as part of an optimal solution (the higher the score the higher the irreplaceability for meeting the specified conservation goals). This score is shown in the ‘mean Marxan biodiversity’ score. We also report the proportion of each fire that falls into different categories of biodiversity value: high biodiversity value (76-100), medium (51-75), and low (0-50). Without a clear understanding of how different taxonomic groups are impacted by wildfire, we have not attempted to model post-fire impacts.

Caveats

· These data are calculated in a pre-fire state and do not account for fire or other disturbances on landscape.

· Note that the input data vary in the date they were created or generated. Below we list the date of the final report, publication, or the date provided in the digital year reference.

Native Species Richness
Statewide data on native species richness is from CDFW [California Department of Fish and Wildlife]. 2010. ACE-II: Areas of Conservation Emphasis Map. Extent: California. https://map.dfg.ca.gov/ace/ . While there are a variety of variables associated with these data, we only report the mean values. We report an index (0-1) of the native species richness in the area of a 2.5 square mile hexagonal grid cell. These data are from 2010 (year given in digital data reference).

Rare Species Richness
Statewide data on rare species richness is from CDFW [California Department of Fish and Wildlife]. 2010. ACE-II: Areas of Conservation Emphasis Map. Extent: California. https://map.dfg.ca.gov/ace/ . While there are a variety of variables associated with these data, we only report the mean values. We report an index (0-1) of the native species richness in the area of a 2.5 square mile hexagonal grid cell. These data are from 2010 (year given in digital data reference).

High Value Aquatic Biota
Data are from the Watershed Condition Class which is a USFS national assessment of USGS Hydrological Units (HUC12 watersheds) that contain federal lands. We report how much of the fire has “good” quality Aquatic Biota Index which reflects the life-forms present, native species, and non-native or invasive species.
http://www.fs.fed.us/biology/resources/pubs/watershed/maps/watershed_classification_guide2011FS978.pdf .
Note, if there are not high value aquatic watersheds within the fire perimeter then there is no option for drawing them on the map display. These data are from 2011 (year of USFS Watershed Condition Class report).

Linkage Zones
Linkage zones reflect data from a number of regional studies (Huber et al. 2010, Spencer et al. 2010). These studies developed species-specific least cost corridors for linking large core areas for wildlife in the central coast, south coast, and the central valley. Note, if there are no wildlife linkage zones within the fire perimeter then there is no option for drawing these data on the map. These data are from 2010 (year of the two cited publications).

Number of Rare Plants, Birds, Reptiles, Amphibians, Fish, and Invertebrates
The number of rare species in different taxonomic groups is from the California Natural Diversity Database (2015 version, https://www.wildlife.ca.gov/Data/CNDDB ). This database provides information on the status and location of California's rare and endangered plants, animals, and natural communities. It is regularly updated, with updates focused on areas with active Habitat Conservation Plans and Natural Community Conservation Planning as well as areas deemed high priority by CDFW biologists. Access to the database is provided on a subscription basis. Rarity is assessed in the database using standards used in the NatureServe ranking system. We are unable to provide the spatial locations of these records, but the name of the species by taxonomic group can be exported in the ‘Download data, table’ option. These data are from 2015 (year given in digital data reference).

Table 2. Summary of conservation targets and associated conservation goals used in the mapping of biodiversity services in the southern California

Conservation target	Source	Description	Target by ecological unit	Conservation goal assigned (%)
Land cover	FRAP 2015*¹	Largely native vegetation types (with exception of annual grasslands)	Area of each land cover type calculated for each ecological unit	0-100 acres = 100% 100-1,000 acres = 75% 1,000-10,000 acres = 50% >10,000 acres = 25%
Sensitive species	CNDDB and NRIS*²	Selected 203 plant and animal species listed in the USFS Southern California Forest Plan Revision	Point data for species were summarized for each unit Polygon data were used to assign areal extent within each unit	1-5 species pts or 0-10 acres = 100% 6-10 pts or 10-100 acres = 75% 11-20 pts or 100-500 acres = 50% >20 pts or >500 acres = 25%
Landscape connectivity	Central Coast, South Coast, and Central Valley critical linkage studies*³	Species-specific least cost corridors calculated for linking large core areas	If the centroid of the unit intersected with connectivity area, the unit was given a 1, else 0	Any unit with connectivity present = 25%
Steelhead Trout	National Marine Fisheries Service*⁴	Occurrence data from the National Marine Fisheries Service (NMFS)	Total length of steelhead habitat reaches within each unit	Any unit with steelhead habitat present = 25%
Watershed Condition Class	USFS national assessment of HUC12 watersheds that contain federal lands*⁵	Used Aquatic Biota Index which reflects life forms presence, native species, and exotic or invasive species	Identified HUC12 watersheds classed as ‘good’ quality and selected units whose centroids were located in these watershed	Any unit with ‘good’ aquatic biota present = 25%

Data sources:

*1 http://frap.cdf.ca.gov/data/frapgisdata-sw-fveg_download

*2 USDA Forest Service 2005

*3 Huber et al. 2010, Spencer et al. 2010

*4 http://www.nmfs.noaa.gov/pr/species/criticalhabitat.htm

*5 http://www.fs.fed.us/biology/resources/pubs/watershed/maps/watershed_classification_guide2011FS978.pdf .
Note: 88% of total study footprint encompassed

References

Ball, I. R., H. P. Possingham, and M. Watts. 2009. Marxan and relatives: software for spatial conservation prioritisation. Pages 185-195 in A. Moilanen, K. A. Wilson, and H. P. Possingham, editors. Spatial conservation prioritisation: quantitative methods and computational tools. Oxford University Press, Oxford, UK.

Huber, P. R., S. E. Greco, and J. H. Thorne. 2010. Spatial scale effects on conservation network design: trade-offs and omissions in regional versus local scale planning. Landscape Ecology 25:683-695.

Spencer, W. D., P. Beier, K. Penrod, K. Winters, C. Paulman, H. Rustigian-Romsos, J. Strittholt, M. Parisi, and A. Pettler. 2010. California essential habitat connectivity project: a strategy for conserving a connected California. Prepared for California Department of Transportation, California Department of Fish and Game, and Federal Highways Administration, https://www.wildcalifornia.org/blog/2014-annual-report/4 .

Underwood, E. C., A. D. Hollander, P. R. Huber, and C. Schrader-Patton. Mapping the Value of National Forest Landscapes for Ecosystem Service Provision. 2018. In Underwood, E. C, H. D. Safford, N. A. Molinari, and J. E. Keeley (editors). Valuing Chaparral: Ecological, Socio-Economic, and Management Perspectives. Springer Series on Environmental Management. Springer, Cham, Switzerland.

USDA [US Department of Agriculture]. 2004. Southern California National Forests land management plan. US Department of Agriculture, Forest Service, Vallejo, California, USA. http://www.fs.usda.gov/Internet/FSE_DOCUMENTS/fsbdev7_007721.pdf .

Hotspots of Change in Services

To assist in identifying areas of potential interest for further exploration, we present data on the amount of change in four ecosystem services: carbon storage, water runoff, groundwater recharge, and sediment erosion services.

For water runoff, groundwater recharge, and sediment erosion the change raster is generated by using the 12-year average pre-fire and the 12-year average post-fire rasters:

(mean post-fire – mean pre-fire) / mean pre-fire * 100

The resulting change raster is then rendered into the following classes: <= 0%, 0-20%, 20-40%, 40-60%, 60-80%, 80-90%, 90-95%, 95-100%, and >100%. To identify hotspots of change, we used an 80% threshold and the table reports the area (acres) and proportion of the target fire that (a) experienced between >80% change in an ecosystem service (for water runoff, groundwater recharge, and sediment erosion); or (b) identified as low regeneration pixels (see ‘Carbon Lost Over the Long-Term’ section). The change information is also shown in the graph option. The ecosystem services layers in the map display reflect the proportion of change class for each pixel.

For water runoff, groundwater recharge and sediment erosion the change layer represents an increase in that service between pre- and post-fire, for carbon storage the change is a decrease associated with the permanent loss of carbon. For a handful a pixels in some fires, the change calculation resulted in zero or negative values, which were assigned to the <= 0% class. Based on a subsample of fires, these pixels accounted for about 1% or less of any fire area and appeared to be associated with waterbodies, urban and agricultural cover types.

Caveats

· Variation in the amount of area identified as a hotspot( > 80% change) varies between each service. This is, in part, due to the services being generated by different models and approaches, (i.e., water runoff and groundwater recharge from the Basin Characterization Model and sediment erosion retention from InVEST).

· The interpretation of the change hotspots depends on the goals of the user.