A Computationally Efficient Spatio-Temporal Fusion Model for Reflectance Data

Fusing remotely-sensed reflectance data from different sources at different spatial and temporal scales is useful to monitor lake water quality. The nonparametric statistical downscaling model (NSD) [5] can account for a change of spatial and temporal support between two remote sensors, but it is computationally demanding for large datasets. This work proposes a method to improve the computational efficiency of the NSD model by endowing it with a Gaussian predictive process. The predictive performance and computational efficiency of both models are compared through simulation and using satellite reflectance data from Lake Garda.