See more from this Session: Soil Change: Characterization and Modeling Across Scales: II
Monday, November 1, 2010
Long Beach Convention Center, Exhibit Hall BC, Lower Level
Spatiotemporally dynamic parameters such as soil moisture can be described via statistical models, but these can be difficult to translate into the mind’s eye. Visualization can aid exploration and interpretation. We characterized spatial correlation to interpolate continuous views of static soil physical properties determined at five depths (plow layer to 75 cm) from cores collected at 60 locations in a 12-ha North Carolina coastal plain agricultural field. We used profiling TDR probes to measure soil moisture at the five depths at various times over several years. Static properties were rendered as continuous 3D volumes and combined with soil moisture to calculate availability indices and estimate the least limiting water range. Results were time-animated as 3-D volumes, augmented with concurrent animations of the water table and moving 7-d total precipitation. Visualization effectiveness was examined by investigating interactions between spatial correlation of soil moisture, soil properties, and crop production. Preliminary results showed strong layering of soil physical properties, reflecting pedogenesis and management, and a substantial increase in horizontal spatial variability of soil moisture with time. Our working hypotheses are (1) increasing variability indicates complex interactions among plant growth, precipitation, evapotranspiration, and soil physical quality factors; and (2) visualization can clarify these interactions and lead to testable hypotheses. If we are successful, the methods may be useful for other investigations. Foreseeable applications include investigating precision agriculture treatment response in space and time. Field management may also benefit from visualization by the display of changes to soil physical quality through time. Important to the work performed was the use of open-source software, which can significantly reduce barriers to modeling, interpolation, and visualization. We demonstrate open-source tools using the R statistical computing environment to model spatial structure and interpolate continuous volumes. We demonstrate interactive visualization using VisIt, an open-source tool from Lawrence Livermore National Laboratory.