A Numerical Evaluation of Soil-Landscape Models In the Sierra Foothill Region, CA: Implications for Digital Soil Mapping.

The Sierra Foothill Region in California represents an important interface between the agricultural and urban areas of the Great Valley and the wildlands of the Sierra Nevada. Although rangelands of the Sierra Foothills have been mapped for most of California, the spatial detail and map unit complexity do not always adequately describe these highly variable soil landscapes.  Considerable progress has been made in numerical approaches to soil mapping, however, little has been directly integrated into current soil survey operations in California. Our objective was to identify dominant soil-forming factors and processes that affect soil variability, at the landscape to watershed scale, within the Sierra Foothill Region. Two benchmark landscapes were established on representative lithologies (metavocanic and granitic) of the foothill region. Local (terrain-influenced) deviations from mean annual temperature and mean annual precipitation were evaluated by in-situ monitoring of volumetric water content and soil temperature. Indices of terrain shape, geomorphic elements and microclimate gradients were extracted from a 1-m elevation model built from real-time kinematic (RTK) GPS measurements within each benchmark landscape. Soils at 100 locations within the metavolcanic site and 15 locations within the granitic site were described and sampled by genetic horizon. Standard laboratory methods were used to measure particle size distribution, color, pH cation exchange capacity, total carbon, total nitrogen, exchangeable cations, acid-oxalate extractable Fe and Mn, and citrate-bicarbonate-dithionite (CBD) extractable Fe and Mn. Elemental concentrations data were obtained for each sample by X-ray fluorescence. Mean annual soil temperature (MAST) was effectively modeled at both catenas (R² = 0.89), with modeled beam radiance and canopy cover accounting for approximately the same proportion of variance. Soil moisture was very poorly modeled with the combination of (modeled) annual beam radiance and the compound topographic index (CTI), accounting for 1% (metavolcanic site) to 17% (granitic site) of the variance in mean percent of saturation over the 2008-2009 water year. Variation in soil "character" (clay content, pH, color, total carbon, cation exchange capacity, and Fe_o/Fe_d) was partitioned across suites of variables associated with terrain shape and microclimate, lithologic variability, and depth. Terrain shape accounted for 2% (metavolcanic site) to 24% (granitic site) of the variance in soil properties, while lithology accounted for about 20% at each site. Horizon depth accounted for 9% (metavolcanic site) to 18% (granitic site) of the variance in soil properties. The amount of variance partitioned into each category reflects how well each suite of variables may perform when used as predictors in regression models. In addition, variance proportions may suggest the relative importance of soil-forming factors in this region. Our findings suggest that effectiveness of commonly used indices of microclimate (modeled solar radiation) and effective precipitation (CTI) are highly variable within the Sierra Foothill Region. These indices do not always account for the expected differences in measured soil temperature and moisture; thus care should be taken when interpreting models that include these indices. We also found that landscape scale variability in lithology contributed more to soil variation than terrain shape in some cases. Several suggestions, based on our findings, will be presented within the context of pending soil survey update operations in the Sierra Foothill Region.