How useful are the lake surface temperature estimates from a geostationary satellite (Himawari-8) to detect seasonal and diurnal changes?

Abstract

ABSTRACT The current geostationary satellite images have a time resolution of minutes. It is not clear how useful they are for detecting diurnal and seasonal changes. We tested this by comparing lake surface temperatures (LSTs) from Himawari-8 Advanced Himawari Imager (AHI) images with in situ measurements in Lake Kasumigaura, Japan, for one year. LST products of the Sentinel-3 satellites were also compared. We found that instantaneous LSTs can be estimated with an root mean square difference (RMSE) of 0.6°C. A large amount of data (N = 102–103) is available for each month, but N differs greatly in the range of 216–2771. A good agreement (RMSE of 0.9°C) was found between monthly mean LSTs from Himawari-8 and in situ measurements, and it is better than the RMSE of 1.2–1.7°C found for the Sentinel-3 LST products. The diurnal change detection was more difficult due to insufficient data numbers around noon. Practical remedies for this issue are suggested.