A Spatio-Temporal Brightness Temperature Prediction Method for Forest Fire Detection with MODIS Data: A Case Study in San Diego

Early detection of forest fire is helpful for monitoring the spread promptly, minimizing loss forests, wild animals, human life, and economy. The performance brightness temperature (BT) prediction determines accuracy detection. Great efforts have been made on BT model building, but there still remains some uncertainty. Based widely used contextual (CM) temporal-contextual (TCM), we proposed a spatio-temporal (STCM), which involves historical images to contrast correlation matrix between pixel be predicted its background pixels within dynamic window, spatial distance factor was introduced modify matrix. We applied STCM fire-prone area in San Diego, California, US, compared it with CM TCM. found that average RMSE 12.54% 9.12% lower than TCM, standard deviation calculated by reduced 12.04% 15.57% respectively. In addition, bias concentrated around zero range 88.7% 15.3% results demonstrated can obtain highest most robust performance, followed performed worst. Our research potential improving potentially useful other environmental variables high temporal autocorrelation. However, requirement high-quality continuous data will limit application cloudy rainy areas.