Sub-daily live fuel moisture content estimation from Himawari-8 data

Abstract

Live fuel moisture content (LFMC) is a crucial variable affecting fire ignition and spread. Satellite remote sensing has been effective in estimating LFMC over large spatial scales, but continuous sub-daily (e.g., every 10 mins to hourly during daylight) LFMC monitoring from space is yet to be accomplished. Using the geostationary satellite Himawari-8 temporally dense observations every 10 mins, this study designed a generalized reduced gradient (GRG) numerical optimization method coupled with PROSAILH_5B radiative transfer model (RTM) to track the sub-daily LFMC dynamics. This method simultaneously accounted for the changing sun-target-sensor geometry bi-directional reflectance distribution function (BRDF) effect on Himawari-8 AHI reflectance. LFMC field measurements from Australia and China validated the LFMC estimation from Himawari-8 AHI. In addition, they were also compared to estimates from two broadly used polar-orbiting satellites, the Landsat-8 OLI and Terra+Aqua MODIS. At the sub-daily scale, the LFMC estimated using the GRG method from Himawari-8 AHI yielded reasonable accuracy (R2 = 0.61, rRMSE = 20.78%). When averaged to a daily scale, the accuracy of LFMC estimation based on the Himawari-8 AHI was lower (R2: 0.60–0.61, rRMSE = 25.38%–26.58%) than that based on the Landsat-8 OLI (R2: 0.68–0.79, rRMSE = 18.11%–25.89%) and Terra+Aqua MODIS (R2: 0.63–0.76, rRMSE = 19.73%–25.84%). However, after removing some heterogeneous measurements, the difference in the accuracy of LFMC estimates among these three data sources got smaller and improved (R2: 0.72–0.82, rRMSE = 17.96%–23.84%). Furthermore, the method proved its feasibility and applicability to identify fire danger conditions through two wildfire case studies: one in Queensland (Australia, 2019) and another in Xichang (China, 2020). These studies showed that the wildfires started when the Himawari-8 AHI-based sub-daily LFMC reached its daily minimum. Therefore, this study serves as a foundational step toward estimating sub-daily LFMC dynamics, an important yet overlooked factor in assessing sub-daily fire danger and behavior.