Home / Use It
Three ways to get the Almanac into your work - stream it directly from the cloud, grab a single file, or pull data in bulk. Most people will want the first, and it's likely easier than you expect.
You may not need to download anything. You still have the option to download the full files onto a drive and work with them the way you're accustomed to - but there are quicker options now. You can stream the files directly into your existing software, work with the data, and save your outputs just as you're used to. The files live in the cloud, and modern GIS tools can read them in place - pulling only the specific areas and pixels you need over the network, without ever copying the whole file.
For an 18-property, 41-year dataset, that's the difference between downloading and managing hundreds of gigabytes on disk, or pointing your software at a URL and getting started immediately. This is what "cloud-native geospatial" means in practice, and it changes what's worth attempting. Instead of deciding up front which layers and years to download, you can explore the whole record interactively, pull just what a given analysis needs, and keep your work pointed at the single authoritative copy on Source Cooperative. The same approach scales from a quick look at one watershed to a statewide computation.
One source, streamed into the tools you already use. The Almanac lives in one place on Source Cooperative as Cloud-Optimized GeoTIFFs, described by a STAC catalog so tools can discover and stream it. Analysts pull only the layers, years, and area they need - directly into ArcGIS Pro, QGIS, Python, AGOL, or wherever they work - and what they learn in use, including reports of data issues, feeds back into each annual release. Both the work and the data improve over time.
The rest of this page shows how to hook up the pipes - streaming into ArcGIS Pro and the open-source stack, grabbing a single file, or pulling data in bulk. If you've never streamed cloud rasters before, start with Stream it; it's less work than downloading, not more.
Read rasters directly from the cloud in ArcGIS Pro, QGIS, or code. No download.
SimpleEvery file has a public HTTPS URL. Fetch it with a browser, curl, or wget.
At scalePull a whole layer or the entire dataset with the AWS CLI - no account needed.
The Almanac is published as two datasets at source.coop/wildland-almanac - California and CONUS - each containing versioned releases (initially just v2026.1). A version is a directory level, not part of the filename. The README and documentation live at each dataset root and describe the release. You can browse the directory organization directly on the Source website, and preview any layer in your browser - for example, scroll to the bottom of the 2025 aboveground-biomass file detail page to see it rendered.
# Source organization → dataset root → version → property → file source.coop/wildland-almanac/ └── california/ # dataset root ├── README.md # start here ├── WildlandAlmanac_CA_Documentation.pdf └── v2026.1/ # a release ├── Carbon_AGB/ │ ├── WildlandAlmanac_CA_Carbon_AGB_1985.tif │ ├── … (water years 1985–2025) │ └── WildlandAlmanac_CA_Carbon_AGB_2025.tif ├── Veg_TreeFrac/ └── … (18 properties)
Filenames follow WildlandAlmanac_CA_{Property}_{WaterYear}.tif. Substitute the property name and four-digit water year for any other file. Newer releases (v2027.1, …) add as parallel version directories; the older ones stay put, so published links keep working and you can always return to the exact dataset you used if you need to replicate your work in the future.
Because the files are COGs, GIS tools can read them straight from cloud storage over HTTP, transferring only the bytes needed for your current view and zoom. The URI form differs between ArcGIS Pro and the open-source / GDAL stack - both point at the same underlying file.
On any Source file-detail page, the Cloud URI is the S3 form (for ArcGIS) and the Source URL is the HTTPS form - prepend /vsicurl/ to the latter for GDAL-based tools.
Add Data → From Path → paste the s3:// URI above. The raster streams in; no download, no credentials.
Layer → Add Layer → Add Raster Layer → set Source type to a protocol/HTTP source and paste the /vsicurl/https://… URI.
AGOL Map Viewer does not yet support adding COGs directly by URL (this is an open Esri feature request as of 2026). The working path is to add the layer in ArcGIS Pro using the s3:// URI above, then publish it from Pro to your AGOL organization as a hosted imagery layer. Share permissions are then managed within your AGOL group as usual. The Esri JavaScript SDK can render COGs directly via ImageryTileLayer, so custom AGOL apps (Experience Builder, Web AppBuilder, custom widgets) can stream the files without going through Pro.
import rasterio
url = ("/vsicurl/https://data.source.coop/wildland-almanac/"
"california/v2026.1/Carbon_AGB/"
"WildlandAlmanac_CA_Carbon_AGB_2025.tif")
with rasterio.open(url) as src:
# windowed read - only the bytes you need are fetched
window = src.window(*src.bounds) # or a sub-bbox
agb = src.read(1, window=window)
print(src.crs, src.res, agb.shape)
library(terra)
url <- paste0("/vsicurl/https://data.source.coop/wildland-almanac/",
"california/v2026.1/Carbon_AGB/",
"WildlandAlmanac_CA_Carbon_AGB_2025.tif")
agb <- rast(url)
agb # prints extent, CRS (EPSG:5070), resolution
plot(agb)
−9999 for all layers except Fire_LCP, where it is 0 (FARSITE convention).Every file has a public HTTPS URL. No credentials needed. Use a browser, curl, wget, or PowerShell Invoke-WebRequest.
# curl / wget curl -O https://data.source.coop/wildland-almanac/california/v2026.1/Carbon_AGB/WildlandAlmanac_CA_Carbon_AGB_2025.tif # Windows PowerShell Invoke-WebRequest -Uri "https://data.source.coop/wildland-almanac/california/v2026.1/Carbon_AGB/WildlandAlmanac_CA_Carbon_AGB_2025.tif" -OutFile "Carbon_AGB_2025.tif"
You usually don't need the entire layer. Because the files are COGs, GDAL can read one straight from the cloud and write out just your area of interest, fetching only the bytes for that window over the network. This is the same partial-read mechanism the streaming section uses - here it writes a small clipped GeoTIFF to disk instead of reading into memory.
# Clip a lat/long box (W S E N) and write a small local GeoTIFF. # GDAL reads only the tiles overlapping the box over HTTP - the source file is never fully downloaded. gdalwarp \ -te -120.5 37.0 -119.0 38.2 -te_srs EPSG:4326 \ /vsicurl/https://data.source.coop/wildland-almanac/california/v2026.1/Carbon_AGB/WildlandAlmanac_CA_Carbon_AGB_2025.tif \ Carbon_AGB_2025_clip.tif # If you already have coordinates in the data's projection (EPSG:5070, meters), # gdal_translate -projwin is simpler (note order: xmin ymax xmax ymin): gdal_translate \ -projwin -2000000 2000000 -1900000 1900000 \ /vsicurl/https://data.source.coop/wildland-almanac/california/v2026.1/Carbon_AGB/WildlandAlmanac_CA_Carbon_AGB_2025.tif \ Carbon_AGB_2025_clip.tif
-t_srs EPSG:4326 to gdalwarp if you need the clip reprojected to lat/long.For multiple files or whole layer directories, the AWS CLI is fastest. No AWS account or credentials are required - pass --no-sign-request. The CLI is available for Windows, Mac, and Linux at aws.amazon.com/cli.
# one file aws s3 cp \ s3://us-west-2.opendata.source.coop/wildland-almanac/california/v2026.1/Carbon_AGB/WildlandAlmanac_CA_Carbon_AGB_2025.tif . \ --no-sign-request # one layer directory (all water years) aws s3 sync \ s3://us-west-2.opendata.source.coop/wildland-almanac/california/v2026.1/Carbon_AGB/ \ ./Carbon_AGB/ --no-sign-request # an entire release aws s3 sync \ s3://us-west-2.opendata.source.coop/wildland-almanac/california/v2026.1/ \ ./wildland-almanac-ca-v2026.1/ --no-sign-request
The Fire_LCP landscape is delivered as eight single-band COGs - five that vary per year (fuel model, canopy cover, canopy height, canopy base height, canopy bulk density) and three static topographic bands (elevation, slope, aspect). Single-band delivery lets you pull any one band on its own - for example canopy height - or clip and stack all eight to build a landscape for your area of interest.
# 1. virtually stack the 8 bands for one year, in LCP order, straight from the cloud gdalbuildvrt -separate Fire_LCP_2020.vrt \ /vsicurl/https://data.source.coop/wildland-almanac/california/v2026.1/Fire_LCP/WildlandAlmanac_CA_Fire_LCP_Elevation.tif \ /vsicurl/https://data.source.coop/wildland-almanac/california/v2026.1/Fire_LCP/WildlandAlmanac_CA_Fire_LCP_Slope.tif \ /vsicurl/https://data.source.coop/wildland-almanac/california/v2026.1/Fire_LCP/WildlandAlmanac_CA_Fire_LCP_Aspect.tif \ /vsicurl/https://data.source.coop/wildland-almanac/california/v2026.1/Fire_LCP/WildlandAlmanac_CA_Fire_LCP_FBFM40_2020.tif \ /vsicurl/https://data.source.coop/wildland-almanac/california/v2026.1/Fire_LCP/WildlandAlmanac_CA_Fire_LCP_CC_2020.tif \ /vsicurl/https://data.source.coop/wildland-almanac/california/v2026.1/Fire_LCP/WildlandAlmanac_CA_Fire_LCP_CH_2020.tif \ /vsicurl/https://data.source.coop/wildland-almanac/california/v2026.1/Fire_LCP/WildlandAlmanac_CA_Fire_LCP_CBH_2020.tif \ /vsicurl/https://data.source.coop/wildland-almanac/california/v2026.1/Fire_LCP/WildlandAlmanac_CA_Fire_LCP_CBD_2020.tif # 2. cut your area of interest (EPSG:5070 meters: xmin ymax xmax ymin) to a local 8-band GeoTIFF gdal_translate -projwin -2000000 2000000 -1900000 1900000 Fire_LCP_2020.vrt Fire_LCP_2020_aoi.tif
.lcp) from rasters like these; GDAL can also write an .lcp directly with gdal_translate -of LCP. The per-band units (canopy height and base height in decimeters, bulk density ×100, cover in percent, FBFM40 categorical) are listed in the documentation PDF - set them when you build the landscape. For CONUS, swap california → conus and use a snapshot year (1990, 2000, 2010, 2020, 2025).We don't have time to offer much technical support at the moment, but notes on errors and use cases are genuinely valuable and shape future releases. Please file feedback and reports of errors on the GitHub issue tracker.