Application of multiple constraints model-data assimilation techniques to coupling satellite passive microwave and thermal imagery for estimation of land surface variables and energy fluxes

[摘要] Executive Summary: This report demonstrates the application of a multiple constraints model-data assimilation (MCMDA) scheme to the combination of AMSR-E soil moisture content (SMC) and MODIS land surface temperature (LST) in a coupled biophysical model for the tropical savannas of northern Australia, and discusses the difficulties and error sources encountered. The investigations showed that AMSR-E SMC data on their own were poor constraints to the model. Incorporating LST data via the MCMDA framework the scheme was able to ameliorate deficiencies in the SMC data and resulted in enhanced characterisation of land surface soil moisture and energy balance. The model predicted on average cooler LST’s (~1.7 K) and wetter SMC values (~0.04 g cm-3) than the satellite image products. Mismatch between model-predicted and satellite-observed SMC and LST is computed by an objective function in the MCMDA scheme and regulated by the model parameter gmax—the canopy-averaged maximum stomatal conductance. A range of optimisation routines were investigated for the task of estimating gmax. Agreement (±0.1 mm s-1) between the methods gave confidence in the global optimality of the final estimated value of gmax = 9.4 mm s-1 for the study region. Ensuing estimates of canopy stomatal conductance, gC, and latent heat flux, ëE, were compared to those obtained in a MODIS evapotranspiration product of Mu et al. (2006). Good agreement was observed (RMSE for gC = 0.5 mm s-1 and for ëE = 18 W m-2) with differences attributable to a greater canopy-to-air vapour pressure gradient obtained from a more realistic partitioning of soil surface and canopy temperatures in the MCMDA scheme. In terms of RMSE, an optimal fit to both satellite image datasets resulted in an increase by 84% for predicted SMC and a 0.06% increase for LST relative to the fit to each dataset separately. The scheme’s capability for handling multiple incommensurate datasets is of key importance for improved forecasting of land-air exchanges of scalars such as water, energy and carbon in tropical savanna systems.

[发布机构] CSIRO

[学科分类] 地球科学(综合)