Home / About
How the Wildland Almanac is produced, how the project got started, and how to reach the team behind it.
The Wildland Almanac is built from the full Landsat archive through a unified pipeline of roughly thirty programs - written mostly in MATLAB and R - that run sequentially under one code base. The pipeline is built to be re-run end-to-end for each annual update, which is what makes it possible to reprocess the entire 1985-onward time series consistently with every release rather than splicing new years onto old. Production runs on UCI’s HPC3 cluster and consumes approximately one million core-hours annually and 800 TB of storage.
The methodological backbone is straightforward: Landsat observations are processed through the COLD change-detection package to identify disturbances and to synthesize gap-free imagery for each pixel and year; those synthetic images are combined with top-tier training data (the ALS-derived canopy-height model of Allred et al. for height; FIA plots for biomass; AmeriFlux towers for water and carbon fluxes; and so on); and the resulting properties are estimated with gradient-boosted machine learning (XGBoost and LightGBM) or regressions. Additional information on training data and methods for each property is in The Data and in the documentation PDF.
The Wildland Almanac is a spinoff from the Center for Ecosystem Climate Solutions (CECS), a four-year (2019-2023) research program at the University of California, Irvine. CECS was supported by California's Strategic Growth Council Climate Change Research Program, with funds from California Climate Investments. It involved more than twenty scientists across the UC system and other universities, and produced both peer-reviewed publications and data products to inform California land-management strategies.
CECS was the prototype phase. It identified the data needs of land managers and the questions they ask, worked out many of the methods, the approach to reconciling multiple ecosystem properties on one consistent footing, and the practical questions of how to produce data at this scale. Michael Goulden, Roger Bales, John Battles, Jim Randerson, and Jon Wang made foundational contributions to that work and remain close collaborators.
CECS ended in 2023. It was a university research program rather than sustained infrastructure: there was no structural mechanism to keep updating and distributing the data. The Wildland Almanac represents a structure and commitment to update and expand these datasets. The goal is a durable, openly available, regularly updated dataset that planners, managers, and researchers can rely on, separate from the academic-grant cycle that produced it.
The Wildland Almanac is led by Michael L. Goulden in the Department of Earth System Science at the University of California, Irvine. For questions about the data, errors, or use cases, please file a GitHub issue; issues are tracked and folded into future releases. The data are offered with no promise of technical support. Reports of errors and notes on real-world use are welcome and shape future releases.
Contact: mgoulden@uci.edu
Department of Earth System Science, University of California, Irvine