Details

“Computation for Climate: Modeling Agriculture Under Uncertainty”

Abstract

Climate mitigation in agriculture is ultimately a modeling problem: how do we turn complex biology and sparse data into reliable decisions? Soil carbon and greenhouse gas emissions arise from nonlinear processes, incomplete records, and uncertain futures. Converting those dynamics into decision-ready tools requires integrating computer science, statistics, and domain science in ways that are rigorous, transparent, and scalable.

David LeBauer will describe his path from ecology to building climate modeling systems that support governments and startups working on carbon and greenhouse gas accounting in managed lands. He chose to build a career at the intersection of open-source software, quantitative modeling and public policy.

LeBauer will introduce the computational architecture behind these systems in accessible terms: how biological processes are encoded in models, how Bayesian calibration constrains parameters using noisy data, how ensemble simulations and sensitivity analysis propagate and decompose uncertainty, and how software design and open science practices determine whether tools are interpretable, reproducible, transparent and scalable for real-world decision-making.

The broader message is that impactful climate work requires scientific depth, disciplined software engineering, and a commitment to transparency and reproducibility, and that there are multiple viable paths to building such a career.

Speaker

David LeBauer (PhD, earth system science, UC Irvine) is a scientist and consultant working at the boundary of ecology, statistics, and scientific computing. He builds open, reproducible modeling and data infrastructures that couple process-based ecosystem models, field measurements, and Bayesian inference to make ecological predictions transparent, scalable, and decision-relevant. He is founder of The LeBauer Approach and Modeling Lead and Project Manager for California’s statewide cropland carbon monitoring and modeling initiative. Previously, he founded the Data Science team at the University of Arizona and led model calibration and validation at Indigo Ag in support of carbon credit protocols. He is creator and long-time co-lead of the open-source PEcAn ecosystem modeling platform and has led open data and cyberinfrastructure efforts supporting model–data synthesis, forecasting, and data-intensive agricultural research.