Estimating Surface Energy Fluxes via Assimilation of GOES Data into an Ensemble-Based Data Assimilation System

Abstract:

A number of studies have assimilated sequences of land surface temperature (LST) measurements into variational data assimilation (VDA) models to estimate surface energy fluxes model. However, the VDA systems suffer from several limitations: (1) It is difficult to derive adjoint models in the VDA models, (2) It is time-consuming to compute flow-dependent background error covariance, and (3) VDA schemes cannot directly provide the uncertainty of estimates. To overcome these shortcomings, an Ensemble Kalman smoother (EnKS) data assimilation framework was developed and tested extensively over six FLUXNET sites with grassland, cropland and forest land cover types. The state augmentation approach is used to estimate both the model parameters and state simultaneously. Land surface temperature (LST) data retrieved from geostationary operational environmental satellites (GOES) are assimilated into the EnKS scheme. The results illustrated that the EnKS model can predict surface energy fluxes well over different sites with various hydrological conditions and land cover types. Moreover, the results from EnKS are compared with those of VDA. The comparison shows that VDA performs slightly better than the EnKS model, implying that EnKS performs marginally suboptimal.