AUTHOR=Kramer Sasha J. , Bisson Kelsey M. , Mitchell Catherine TITLE=What data are needed to detect wildfire effects on coastal ecosystems? A case study during the Thomas Fire JOURNAL=Frontiers in Marine Science VOLUME=10 YEAR=2023 URL=https://www.frontiersin.org/journals/marine-science/articles/10.3389/fmars.2023.1267681 DOI=10.3389/fmars.2023.1267681 ISSN=2296-7745 ABSTRACT=

Wildfires are growing in frequency and severity worldwide, with anthropogenic climate change predicted to worsen the effects of wildfires in the future. While most wildfire impacts occur on land, coastal fires can also affect the ocean via smoke production and ash deposition. The impacts of wildfires on marine ecology and biogeochemistry have been studied infrequently, as it is difficult to conduct fieldwork rapidly and safely during unpredictable natural disasters. Increasingly, remote sensing measurements are used to study the impacts of wildfires on marine ecosystems through optical proxies. Given the optical impacts of smoke and in-water ash, these measurements may be limited in their scope and accuracy. Here, we evaluate the potential and limitations of remote sensing data collected from MODIS-Aqua to describe the effects of wildfires on optics and phytoplankton observations. Using samples collected in the Santa Barbara Channel (California, USA) during the Thomas Fire in December 2017, we found that MODIS-Aqua data were unsuited for interpreting ecosystem effects during a wildfire. Our results identified a persistent overestimation of chlorophyll-a concentration from MODIS-Aqua compared to in situ measurements. Optical models applied to in situ radiometry data overestimated the absorption by colored dissolved organic matter (CDOM) during the wildfire. Satellites will remain an important tool to measure the impacts of wildfires on marine ecosystems, but this analysis demonstrates the importance of in situ sampling to quantify the impacts of wildfires on ocean ecology and biogeochemistry due to the difficulty of interpreting remote sensing data during these events.