Quantifying Interactive Effects of Fire and Precipitation Regimes on Catchment Biogeochemistry of Aridlands

# Abstract

Wildfire has increased in size, frequency, and severity across the arid and semi-arid western United States, yet predictions of aquatic response remain uncertain because fire effects are mediated by highly variable precipitation, intermittent hydrologic connectivity, and lagged transport of solutes and sediments to stream networks. The lter-sparc-fire-arid-streams project addresses this uncertainty by building a reproducible, catchment-scale synthesis workflow that links fire history, hydrologic context, and water chemistry across diverse aridland sites. The project is grounded in a database-first architecture that harmonizes multi-source chemistry and discharge records, derives site-level and event-level fire exposure metrics, and supports structured Bayesian inference of concentration-discharge dynamics before and after fire.

Our approach aligns with and operationalizes the conceptual framing that post-fire outcomes in aridlands emerge from interactions among accumulation, combustion, transport, and downstream propagation processes under pulsed hydroclimatic forcing (https://doi.org/10.1093/biosci/biae120). In practice, we standardize USGS and non-USGS chemistry records in PostgreSQL/PostGIS, aggregate fire exposure in the firearea schema (including fire timing windows and cumulative burn metrics), and integrate hydrologic covariates for model-ready exports. We then fit a progression of Stan models from baseline concentration-discharge relationships to pre/post-fire, delta, and unified formulations, with hierarchical extensions under active development for cross-site inference.

Support from the companion firearea repository (https://srearl.gitlab.io/firearea/index.html) is central to this workflow. firearea functions and vignettes provide reusable methods for catchment delineation, fire-perimeter preparation, fire-catchment intersection, and fire-to-channel distance quantification that feed directly into the database products and modeling pipeline. This integration enables consistent spatial attribution of disturbance exposure and hydrologic connectivity across watersheds.

The resulting synthesis framework is designed to quantify not only immediate post-fire responses, but also persistent and lagged stream-solute signals associated with intermittent precipitation regimes in arid lands (https://doi.org/10.1007/s10533-024-01154-y). By combining transparent data engineering with probabilistic CQ modeling, the project delivers reusable analytical products for hypothesis testing, cross-ecosystem comparison, and long-term prediction of fire impacts on aridland stream biogeochemistry.

Beyond site-specific assessment, the workflow supports comparative analyses across hydroclimatic gradients, disturbance histories, and ecosystem types, enabling stronger tests of when fire effects are amplified, buffered, or delayed. This capacity is intended to improve both mechanistic understanding and practical planning for water-quality risk in fire-prone aridland watersheds.