A synthetic experiment to investigate the potential of assimilating LAI through direct insertion in a land surface model

Abstract

This study evaluates the potential of assimilating phenology observations using a direct insertion (DI) method by constraining the modeled terrestrial carbon dynamics with synthetic observations of vegetation condition. Specifically, observations of leaf area index (LAI) are assimilated in the Noah-Multi Parameterization (Noah-MP) land surface model across the continental United States during a 5-year period. An observing system simulation experiment (OSSE) was developed to understand and quantify the model response to assimilating LAI information through DI when the input precipitation is strongly biased. This is particularly significant in data poor regions, like Africa and South Asia, where satellite and re-analysis products, known to be affected by significant biases, are the only available precipitation data to drive a land surface model. Results show a degradation in surface and rootzone soil moisture after assimilating LAI within Noah-MP, but an improvement in intercepted liquid water and evapotranspiration with respect to the open-loop simulation (a free run with no LAI assimilation). In terms of carbon and energy variables, net ecosystem exchange, amount of carbon in shallow soil, and surface soil temperature are improved by the LAI DI, although canopy sensible heat is degraded. Overall, the assimilation of LAI has larger impact in terms of reduced systematic and random errors over the Great Plains (cropland, shrubland, and grassland). Moreover, LAI DA shows a greater improvement when the input precipitation is affected by a positive (wet) bias than the opposite case, in which precipitation shows a dry bias.

Keywords

• Data assimilation
• Direct insertion
• Leaf area index
• Dynamic vegetation model (DVM)
• Land surface model (LSM)
• Land information systems (LIS)