B034-0007
Plant Hydraulic Traits Mediate Live Fuel Moisture’s Importance for Wildfire Risk Forecasting

Wednesday, 9 December 2020
Poster
Krishna Rao1, Park Williams2, Noah S Diffenbaugh1, Marta Yebra3,4 and Alexandra G. Konings5, (1)Stanford University, Stanford, CA, United States, (2)Columbia University, Lamont -Doherty Earth Observatory, Palisades, NY, United States, (3)The Australian National University, Fenner School of Environment and Society, Acton, Australia, (4)Bushfire and Natural Hazards Cooperative Research Centre, Melbourne, Australia, (5)Stanford University, Department of Earth System Science, Stanford, CA, United States
Abstract:
Wildfires pose an immense danger to human lives and structures, making it imperative to forecast their risk accurately. Current near-term forecasts of wildfire danger rely on meteorological aridity metrics such as vapor pressure deficit, fire weather index, and precipitation to estimate live fuel moisture (LFM)— a key determinant of fire ignition probability and fire spread. However, LFM can vary substantially from meteorology depending on species composition and their hydraulic traits, leading to inaccurate estimates of LFM and ultimately wildfire danger.

In this work, we test the hypothesis that LFM can improve near-term wildfire danger forecasts compared to what can be achieved through meteorological aridity metrics alone. We further hypothesize that the magnitude of this improvement is mediated by plant hydraulic traits. The hypothesis was tested using LFM estimates over the western United States recently derived from microwave and optical remote sensing without assuming any dependence on meteorological aridity.

Testing our hypothesis on historic wildfire records from 2016 - 2020 using a Moderate Resolution Imaging Spectroradiometer (MODIS) burned area product (MCD64A1) revealed that wildfire occurrence patterns could be predicted with greater accuracy if LFM, rather than only meteorological factors, was accounted for. Including LFM improved prediction accuracy by 2-9% depending on land cover (length of color bars in Fig. 1) indicating that meteorological factors alone were insufficient to capture LFM's heterogeneity. Moreover, the improvement was mediated by hydraulic traits of the vegetation cover. We show that depending on fire size, different plant hydraulic traits like stomatal control (anisohydricity), xylem pressure corresponding to 50% loss of conductivity (P50), and maximum plant rooting depth explain the importance of including LFM in predicting spatial patterns of wildfires. Incorporating observationally-driven LFM estimates may thus lead to improved wildfire danger forecasts, especially in drought-resilient ecosystems.

<< Previous Abstract | Next Abstract >>