Gap Filling of Biophysical Parameter Time Series with Multi-Output Gaussian Processes
This work provides a method for accurately filling gaps in biophysical parameter time series, which is crucial for environmental monitoring and agricultural applications, especially in remote sensing data where gaps are common.
This paper addresses the problem of filling gaps in biophysical parameter time series, specifically LAI and fAPAR over rice areas, which cannot be solved by standard single-output Gaussian Process models. The proposed multi-output Gaussian Process models successfully predict these variables even with high amounts of missing data by transferring information across domains.
In this work we evaluate multi-output (MO) Gaussian Process (GP) models based on the linear model of coregionalization (LMC) for estimation of biophysical parameter variables under a gap filling setup. In particular, we focus on LAI and fAPAR over rice areas. We show how this problem cannot be solved with standard single-output (SO) GP models, and how the proposed MO-GP models are able to successfully predict these variables even in high missing data regimes, by implicitly performing an across-domain information transfer.