Submitted to: Agro-Ecosystems Modelling International Congress Proceedings
Publication Type: Abstract only
Publication Acceptance Date: 9/5/2013
Publication Date: 9/27/2013
Citation: Thorp, K.R., Bronson, K.F. 2013. Geospatial Simulation: An open-source, model-independent geospatial tool for managing point-based environmental models at multiple spatial locations. Agro-Ecosystems Modelling International Congress Proceedings. 50:25-36. Interpretive Summary:
Technical Abstract: Many point-based models have been designed to simulate hydrology, gas flux, nutrient dynamics, and/or plant growth processes at a single point on the landscape. However, these environmental processes are known to be spatially variable. Simulations at different spatial locations therefore require adjustment of model input parameters to reflect specific conditions at each location. Increased availability of geospatial data sets, including remote sensing images, land cover maps, digital soil surveys, crop yield maps, and vehicle-based plant or soil maps, can support spatial model parameterization efforts. However, tools are needed for processing geospatial data layers and interfacing the data with the simulation model. The objective of this research was to develop software tools and data processing pipelines for managing data streams from a variety of instruments and merging site-specific information into environmental model simulations. A software 'plug-in' named Geospatial Simulation has been developed for the open-source Quantum geographic information system (GIS) to accomplish required geoprocessing tasks, including processing raster and vector data layers, running simulation models with site-specific data, and optimizing model simulations to site-specific locations. The plug-in was designed to be model independent, meaning the software can interact with any point-based simulation model that uses ASCII files for input and output. Testing and refinement of these geoprocessing tools were accomplished using data from two precision irrigation experiments for surface-irrigated cotton, conducted at Maricopa, Arizona during the summers of 2009 and 2011. Canopy spectral reflectance and canopy temperature data from airborne imagers were processed within predefined management zones. Site-specific soil texture information based on field sampling was interpolated and averaged within each management zone. Spatial data were integrated into site-specific simulations with the DSSAT-CSM-CROPGRO-Cotton model. Simulated annealing optimization was used to calibrate the model uniquely for each management zone by minimizing error between measured and simulated leaf area index (LAI), as estimated from normalized difference vegetation index (NDVI) and canopy height. Site-specific cotton yield and evapotranspiration simulations were compared with observations at the field site.