A downscaled monthly solar-induced chlorophyll fluorescence product at 0.5-degree resolution over east asia during 1995-2003

This paper proposes a hybrid downscaling method, named random forest area-to-area regression kriging (RFATARK), to improve the resolution of the Global Ozone Monitoring Experiment (GOME) solar-induced chlorophyll fluorescence (SIF) product over East Asia during 1995-2003. The RFATARK method effectively captures the nonlinear relationships between variables and incorporates spatial variability through residual prediction. Comparative analyses with two alternative downscaling methods demonstrate the superior performance of RFATARK in enhancing the spatial patterns of the downscaled SIF data. The reconstructed 0.5-degree SIF product offers a more detailed representation of the vegetation dynamics and carbon cycle in East Asia. The proposed methodological framework can also be applied to other coarse-resolution SIF products obtained from different satellite sensors, providing valuable datasets for regional vegetation monitoring and carbon cycle modeling.