Interpretable Models Capture the Complex Relationship Between Climate
Indices and Fire Season Intensity in Maritime Southeast Asia
Abstract
There have been many extreme fire seasons in Maritime Southeast Asia
(MSEA) over the last two decades, a trend which will likely continue or
accelerate due to climate change. Fires, in turn, are a major driver of
atmospheric carbon monoxide (CO) variability, especially in the Southern
Hemisphere. Here we attempt to maximize the amount of CO variability
that can be explained via human-interpretable statistical models that
use only climate mode indices as predictor variables. We expand upon
previous work through the complexity at which we study the connections
between climate mode indices and atmospheric CO (a proxy for fire
intensity). Specifically, we present three modeling advancements. First,
we analyze five different climate modes at a weekly timescale, which
increases explained variability by 15% over models on a monthly
timescale. Second, we accommodate multiple lead times for each climate
mode index, finding that some indices have very different effects on CO
at different lead times. Finally, we model the interactions between
climate mode indices at weekly timescales, which provides a framework
for studying these interactions at a higher level of complexity than
previous work. Furthermore, we perform a stability analysis and show
that our model for the MSEA region is robust, which adds weight to the
scientific interpretation of the selected model terms. We believe that
the complex relationships quantified here will be useful for scientists
studying modes of variability in MSEA and for forecasters looking to
maximize the information they glean from climate modes.