Data Assimilation of Terrestrial Remote Sensing

       Global terrestrial remote sensing products are obtained from reflected radiation using inverse logic while climate models run forward using these products as input to calculate radiative and carbon fluxes for terrestrial systems. However, little resemblance exists between fluxes modeled and those used to derive these variables, due to uncertainties in both remote sensing products and climate model parameters/parameterizations. Remote sensing products are also not optimum because a priori information is not used and different algorithms have been developed to retrieve different vegetation properties without being constrained by other properties. These recognized uncertainties cannot be properly accounted for in applications in current climate models. 

 Here we are developing and testing a data assimilation approach as a dynamic system in a climate model from MODIS data to improve estimates of vegetation properties (e.g., leaf area index, fractional vegetation cover), by optimally combining various sources of information as provided by models and observations. Detail of this approach is documented in our proposal submitted to NASA. Our initial work for this approach includes development of more realistic canopy radiation models and testing of some simple data assimilation approaches. See details at four-stream approximation.

Figure. The 3-D vegetation canopy with their shadows

(Source: http://rami-benchmark.jrc.it/HTML/RAMI4PILPS/EXPERIMENTS3/OVERVIEW/SHRUBLANDS/SHRUBLANDS.php)