Submitted to: Water Resources Research
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/1/2012
Publication Date: 5/17/2012
Citation: Nearing, G.S., Crow, W.T., Thorp, K.R., Moran, M.S., Reichle, R., Gupta, H.V. 2012. Assimilating remote sensing observations of leaf area index and soil moisture for wheat yield estimates: An observing system simulation experiment. Water Resources Research. 48 W05525. Interpretive Summary: Remote sensing technology provides rapid observation of crop growth variability across agricultural regions. This information can be used to drive crop growth simulation models for temporal and spatial analysis of factors that affect crop production, such as impacts of climate change and drought on agricultural productivity. Currently, we lack robust techniques for effectively combining remote sensing and crop modeling technologies for crop production assessments. In this research, we have conducted a simulation study with the CSM-CROPSIM-CERES-Wheat model to understand how to use information from remote sensing images to adjust the crop growth simulation and more accurately estimate crop yield. The study also tested two statistical approaches for merging remote sensing information into the model. Results of the study will be useful in the effort to develop information systems that use remote sensing images and crop growth models to monitor regional impacts of drought and climate change on agricultural productivity. In particular, the study supports the development of NASA's Soil Moisture Active Passive (SMAP) satellite for global soil moisture monitoring. The study demonstrates how information from such satellite systems can be used for assessing the impact of soil moisture deficits (drought) on agricultural production.
Technical Abstract: We develop a robust understanding of the effects of assimilating remote sensing observations of leaf area index and soil moisture (in the top 5 cm) on DSSAT-CSM CropSim-Ceres wheat yield estimates. Synthetic observing system simulation experiments compare the abilities of the Ensemble Kalman Filter (EnKF) and the Sequential Importance Resampling Filter (SIRF) to attenuate the effects of various isolated sources of modeling uncertainty including weather, soil parameters and initial conditions, cultivar parameters, and model state updating equations. Results indicate that data assimilation is effective at improving yield estimates when crops are water stressed and uncertainty is confined to soil parameters, weather, and states – and not due to cultivar parameters. The EnKF is more effective than the SIRF at assimilating soil moisture observations when soil parameters and weather drive uncertainty. The SIRF is better at assimilating observations of leaf area index and is effective when uncertainty is due to vegetation related components such as weather, perturbations to state updating equations including development stage transition timing state variables. Neither filter offers significant improvement to yield estimates when crops are not water stressed or when there is uncertainty in cultivar parameters.