Biomass estimates for shrubland-dominated ecosystems in southern California have, to date, been limited to national or statewide efforts which can underestimate the amount of biomass; are limited to one-time snapshots; or estimate aboveground live biomass only. We developed a consistent, repeatable method to assess four vegetative biomass pools from 2001-2021 for our southern California study area (totaling 6,441,208 ha), defined by the Level IV Ecoregions (Bailey 2016) that intersect with USDA Forest Service lands (Figure 1). We first generated aboveground live biomass estimates (Schrader-Patton and Underwood 2021), and then calculated belowground, standing dead, and litter biomass pools using field data in the peer-reviewed literature (Schrader-Patton et al. 2022) (Figure 2). Over half (52.3%) of the study area is shrubland, and our method accounts for three post-fire shrub regeneration strategies: obligate resprouting, obligate seeding, and facultative seeding shrubs. We also generate biomass estimates for trees and herbs, giving a total of five life form/life history types. These data provide an important contribution to the management of shrubland-dominated ecosystems to assess the impacts of wildfire and management activities, such as fuel management and restoration, and for monitoring carbon storage over the long term.

The biomass data are a key input into the online web mapping tool SoCal EcoServe, developed for US Department of Agriculture Forest Service resource managers to help evaluate and assess the impacts of wildfire on a suite of ecosystem services including carbon storage. The tool is available at https://manzanita.forestry.oregonstate.edu/ecoservices/ and described in Underwood et al. (2022).

REFERENCES

Bailey, R.G. 2016. Bailey's ecoregions and subregions of the United States, Puerto Rico, and the U.S. Virgin Islands. Forest Service Research Data Archive. (Fort Collins, Colorado). https://doi.org/10.2737/RDS-2016-0003

Schrader-Patton, C.C. and E.C. Underwood. 2021. New biomass estimates for chaparral-dominated southern California landscapes. Remote Sensing, 13, 1581. https://doi.org/10.3390/rs13081581

Schrader-Patton et al. 2022. “Estimating Wildfire Impacts on the Biomass of Southern California’s Chaparral Shrublands.” Proceedings for the Fire and Climate Conference May 23-27, 2022, Pasadena, California, USA and June 6-10, 2022, Melbourne, Australia. Published by the International Association of Wildland Fire, Missoula, Montana, USA.

Underwood et al. 2022. “Estimating the Impacts of Wildfire on Chaparral Shrublands in Southern California using an Online Web Mapping Tool.” Proceedings for the Fire and Climate Conference May 23-27, 2022, Pasadena, California, USA and June 6-10, 2022, Melbourne, Australia. Published by the International Association of Wildland Fire, Missoula, Montana, USA.

METHODS

We generated spatial estimates of above ground live biomass (AGLBM, in kg/m²) for 2000-2021 for our southern California study area. The study area, totaling 6,441,208 ha, is defined by the 42 Level IV Ecoregions (Bailey 2016) that intersect the four southern US Department of Agriculture (USDA) National Forests in southern California (Figure 1).

We created biomass raster layers (30m spatial resolution) by modeling a set of covariates (Normalized Difference Vegetation Index [NDVI], precipitation, solar radiation, actual evapotranspiration, aspect, slope, climatic water deficit, elevation, flow accumulation, landscape facets, hydrological recharge and runoff, and soil type) to predict AGLBM using 766 field plots of biomass from the USDA Forest Service Forest Inventory and Analysis (FIA); the Landfire Reference Database (LFRDB) plot data; and other research plots. The dates of field data spanned 2001-2012. The NDVI raster data were derived from Landsat TM/ETM+/OLI multispectral satellite data (onboard Landsat 5, 7, and 8, respectively). NDVI data were composited from all available Landsat images for the months of July and August for each year 2001-2021.  We also downloaded annual precipitation data for each water year (October 1 - September 30) 2001-2021 from PRISM (http://www.prism.oregonstate.edu/).  For each field plot, we extracted the raster values for all covariates; NDVI and precipitation data were matched to the year of plot visit.  We predicted AGLBM using the set of 17 covariates (Random Forest [RF] algorithm in R statistical computing software). To create an AGLBM raster surface for each year 2001-2021, we used NDVI and precipitation raster data specific to each year in the RF (using predict function in the R raster module) (see Schrader-Patton and Underwood 2021 for details).

To estimate other shrubland biomass pools (standing dead, litter, and below ground) we employed a multi-step process:

1)     First, we segregated the study area by community type using the California Wildlife Habitat Relationships (CWHR) data (Mayer and Laudenslayer 1988).  The wildland vegetation of the study area (excluding agricultural, urban, water, and barren classes) contains 45 CWHR classes, 14 of which are >=0.75% of the wildland vegetation in the study area.  We divided these 14 classes into shrubland dominated versus non-shrubland dominated types (annual grass, oak, conifer, mixed hardwood) (Table 1).

Table 1. The Community types (WHR class) that are >= 0.75% of all wildland vegetation in the study area and their % area of the southern California ecoregion

2)     Second, for the shrubland types we determined the per pixel proportion of biomass by three plant life forms: tree, shrub, and herb. We further subdivided the shrub life form into three life history classes based on shrub post-fire regeneration strategies: Obligate Resprouters (OR), obligate seeders (OS), and facultative seeders (FS), providing five plant types in total. We created rasters depicting the proportion of biomass in each of the five plant types by first calculating the proportion of biomass in each type for the plots used in Schrader-Patton and Underwood (2021).  The plot data contained individual plant species, crown width and height measurements.  Using these measurements, we estimated the biomass for each individual plant within the plot by applying published allometric equations (see Schrader-Patton and Underwood 2021 for details). The individual plants in the plots were classified into the five plant types and the proportion of biomass in each type were calculated for each plot. A multinomial model was used to relate these proportions to a suite of geophysical and remote sensing variables which, in turn, was applied to raster surfaces of these predictors. The final outputs were raster maps of the proportion of biomass by life form (tree, shrub, herb) and, for shrubs, the proportion of biomass by life history type (OR, OS, and FS) (Underwood et al. in review).

3)  Third, we estimated the standing dead, litter, and below ground biomass pools by either applying fractions of AGLBM gleaned the available published literature or by using biomass estimates in existing spatial datasets. The specific method used was dependent on the percentage of the WHR class in the study area and the vegetation type (shrub or non-shrub) (Figure 2).   

a)     For shrubland types >= 0.75% of all wildland vegetation in the study area (Mixed Chaparral, Desert Scrub, Coastal Scrub, Chamise Redshank Chaparral, Montane Chaparral, and Sagebrush), we used the proportion of the five plant types as a basis for applying the AGLBM factors from the literature.  For litter estimates, we applied AGLBM factor of 0.78 (derived from Bohlman et al. 2018) to Mixed chaparral, Chamise-Redshank Chaparral, and Coastal scrub WHR classes. These shrubland types also contained tree and herb biomass. We estimated the litter and standing dead biomass for these plant types by multiplying the plant type proportion by AGLBM (Tree and herb AGLBM), or by the North American Wildland Fuels Database (NAWFD, Pritchard et al. 2018) litter biomass (Tree and herb litter and standing dead biomass), or by literature-derived factors (Tree and herb belowground biomass). Sagebrush, Montane chaparral, and Desert scrub were assigned litter biomass from the NAWFD data as these WHR types had no litter estimates in the literature.   

b)     For non-shrubland types >= 0.75% all wildland vegetation in the study area (Coastal Oak Woodland, Pinyon-Juniper, Montane Hardwood, Blue Oak Woodland, Sierran Mixed Conifer, Juniper, and Montane Hardwood-Conifer), the snag and litter NAWFD biomass estimates were used for standing dead and litter estimates, respectively. For belowground biomass, we used AGLBM factors from the literature based on the gross vegetation type (Oak, Conifer, or Mixed) and amount of total per pixel AGLBM.  For example, for Oak WHR types (Coastal Oak Woodland, Blue Oak Woodland) <= 7 kg/m2 we used an AGLBM factor of 0.46 (see Mokany et al. 2006 for breakdown by class breaks).

c)     For all the remaining WHR classes (each < 0.75% of all wildland vegetation in the study area) and Annual Grasslands, we used the NAWFD snag and litter estimates (standing dead and litter biomass), and the California Air Resources Board (CARB, Battles et al. 2014) for our belowground estimates.

The above ground, litter, standing dead, and below ground biomass raster layers for each of the five plant types or WHR class were summed to produce the total biomass raster layers for each biomass pool.

To generate an annual biomass raster layer for each biomass pool for each year 2001-2021, we wrote a Python script (using the ESRI Application Programming Interface [API]). These 32 bit kg/m² geotiff files were multiplied by 1000 and rounded to create 16 bit unsigned integer raster files in g/m².

REFERENCES

Bailey, R.G. 2016. Bailey's ecoregions and subregions of the United States, Puerto Rico, and the U.S. Virgin Islands. Forest Service Research Data Archive. (Fort Collins, Colorado). https://doi.org/10.2737/RDS-2016-0003.

Bohlman, G.N., E.C. Underwood, and H.D. Safford. 2018. Estimating biomass in California’s chaparral and coastal sage scrub shrublands. Madroño 65:28–46. doi:10.3120/0024-9637-65.1.28.

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 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.

Franklin, J. 2002. Enhancing a regional vegetation map with predictive models of dominant plant species in chaparral. Applied Vegetation Science 5:135–146. doi.org/10.1111/j.1654-109X.2002.tb00543.x.

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

Koteen L.E., D.D. Baldocchi, 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.

Kummerow, J., D. Krause, and W. Jow. 1977. Root systems of chaparral shrubs. Oecologia 29:201-212.

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

Mayer, K.E., and W.F. Laudenslayer. 1988. A guide to wildlife habitats of California; California Department of Forestry and Fire Protection, Sacramento, CA.

Miller, P.C., and R. Ng. 1977. Root:shoot biomass ratios in shrubs in southern California and central Chile. Madroño 24:215-223.

Mokany, K., R.J. Raison, and S. Prokushkin. 2006. Critical analysis of root:shoot ratios in terrestrial biomes. Global Change Biology 12:84-96. doi.org/10.1111/j.1365-2486.2005.001043.x.

Park I.W., J. Hooper, J.M. Flegal, and D.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 Distribution 24: 497– 508. doi.org/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. doi.org/10.1029/2019JG005083.

Regelbrugge, J.C., and S.G. Conard. 1996. Biomass and fuel characteristics of chaparral in Southern California. 13th Conference on Fire and Forest Meteorology, Oct. 27-31, 1996, 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 58:155–176. doi.org/10.2307/2937023.

Schrader-Patton, C.C. and E.C. Underwood. 2021. New biomass estimates for chaparral-dominated southern California landscapes. Remote Sensing, 13, 1581. https://doi.org/10.3390/rs13081581.

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.

Underwood et al. 2022. “Estimating the Impacts of Wildfire on Chaparral Shrublands in Southern California using an Online Web Mapping Tool.” Proceedings for the Fire and Climate Conference May 23-27, 2022, Pasadena, California, USA and June 6-10, 2022, Melbourne, Australia. Published by the International Association of Wildland Fire, Missoula, Montana, USA.

USAGE NOTES

The biomass raster layers are packaged in zip files for each year using the following naming structure:

WWETAC_UCD_SoCal_Biomass_XXXX.zip

Where XXXX is the year of the biomass estimates. Within each zip file are the following files:

WWETAC_UCD_ <biomass pool>_XXXX_g_m2.tif

Where <biomass pool> is either aboveground, standing dead, litter, or belowground and XXXX designates the year. The dimensions of the geotiff raster files is 21243 columns by 13618 rows and the bounding box coordinates are 36.79, -121.96 (upper left) and 32.47, -115.23 (lower right), in decimal degrees. The rasters are unprojected and in the WGS84 (WKID 4326) geographic coordinate system (decimal degrees). Pixel size is .000317 x .000317 decimal degrees, approximately 35m x 29m depending on latitude.

Intended users of this dataset include resource managers, researchers who are carrying out biogeographic studies, and people needing vegetation biomass estimates across this landscape.

This dataset is made available under a CC0 license waiver.