Analysis of Spatial Scaling Behavior of Soil Organic Carbon in Florida.

The spatial distribution of soil organic carbon (SOC) is influenced by multiple environmental and human factors acting at specific scales, including land use, hydrology, topography, and geology. These factors behave differently at varying scales, creating a range of possible SOC-landscape relationships that are a function of the scale in which they are observed, including the extent (i.e. size of the study area), grain (i.e. resolution, or pixel size), location (i.e. geographic position), and the spatial dependency of SOC (i.e. spatial autocorrelation). We investigated each of these scale parameters separately with the aim to identify an ideal scale to model SOC in Florida.

Our sampling design spanned across three nested spatial scales in Florida (State, ~150,000 km², 1193 samples; Watershed, ~3,585 km², 141 samples; and Field, ~5.58 km², 152 samples). We used stepwise regression to produce SOC models as a function of environmental factors to separately test the effects of the extent, grain, and location. We used variography (i.e. geostatistics) and fractals to compare the spatial dependency of SOC among scales.

Our results indicate that SOC models vary as a function of the extent, grain, and location, both in terms of accuracy, and in the selection of explanatory environmental factors. The only consistent environmental control of SOC observed at multiple scales was hydrology. Variography and fractals identified nested scales of SOC in the State of Florida, including short-, medium-, and long-range components observed at ~0-3, ~3-30, and ~30-360 km, respectively.

This study greatly contributed in our effort to identify an ideal scale to produce SOC models in Florida, one that reflects the multitude of SOC-landscape interactions, and the very spatial scale of SOC.