| A combination of field observations, statistical modeling, and geographic information system technology is used to demonstrate a soil-landscape modeling process to produce soil drainage class maps. The objectives of this research are to (1) calibrate a statistically based soil-landscape model, (2) validate the model using independent field observations, (3) apply the model to a digital geographic database to produce soil drainage class maps, and (4) extrapolate the model beyond the calibration site and evaluate results.;This study is divided into three distinct phases. First, we calibrated the soil-landscape model using field observations of soil and landscape combinations from a training area. This model uses a combination of class sampling frequencies and multivariate discriminant analysis to predict soil drainage class. Landscape variables used include soil parent material, slope shape and gradient, and various stream and drainageway proximity variables. A 74% overall agreement rate with field observations of soil drainage class was found for the model compared to 69% for the published soil survey.;Second, digital geographic data layers of the landscape variables were created and stored in a geographic information system using a 30 x 30 meter raster format. We applied the soil-landscape model to the digital geographic database on a cell-by-cell basis to define landscape variable combinations and make soil drainage class predictions for each grid cell. The product is a digital map of predicted soil drainage class.;Third, when the model was applied to another site within the same physiographic province, it failed to identify poorly drained soils on colluvial footslopes since this situation did not occur at the model calibration site. Consequently, the model is not portable for regional applications in its current form and may require partial re-calibration to local conditions.;The soil-landscape modeling technique has potential uses for soil survey and has advantages pertaining to mapping consistency, documentation of decision criteria, the ability to update models and map products easily, and the use of geographic information system technology for data management. |
