A Method for Multiscale Estimation of Leaf Area Index from Time-Series Multi-Source Remote Sensing Data

Satellite observations are affected by clouds, aerosol and other factors, resulting in temporal discontinuities and spatial incompleteness in leaf area index (LAI) products. Furthermore, the currently available LAI products are generally retrieved from mono-temporal remote sensing data acquired by a single sensor, without comprehensive utilization of multi-source satellite observations. This paper proposes a new data assimilation method to retrieve temporally continuous LAI at different spatial scales from time-series multi-source remote sensing data with different spatial resolutions using an ensemble multiscale tree model (EnMsT). A dynamic model was constructed to describe the change rule of LAI in time series. At each time-step, the forecast of LAI from the dynamic model was used to construct an initial EnMsT. Then, satellite surface reflectance data with different spatial resolutions were used to update the LAI at each node of the EnMsT using an ensemble multiscale filter (EnMsF) technique. The final results demonstrate that this new method can estimate temporally continuous LAI at different spatial scales and the retrieved LAI values are in good agreement with the field measurements.