|Gelder, B - Iowa State University|
|Anex, R - Iowa State University|
|Kaspar, Thomas - Tom|
|Sauer, Thomas - Tom|
Submitted to: Soil Science Society of America Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/11/2011
Publication Date: 9/1/2011
Citation: Gelder, B.K., Anex, R.P., Kaspar, T.C., Sauer, T.J., Karlen, D.L. 2011. Estimating soil organic carbon using aerial imagery and soil surveys. Soil Science Society of America Journal. 75:1821-1828.
Interpretive Summary: Effectiveness of nitrogen fertilizers or herbicides can vary greatly depending on the soil organic carbon (SOC) concentration. As a result, implementation of precision or site specific nitrogen or herbicide application would benefit greatly from low-cost, high-quality, georeferenced soil organic carbon maps. Currently, the only way to produce these maps is through expensive soil sample collection and analysis. Georeferenced, aerial imagery is now widely available and inexpensive. Because soils higher in organic carbon appear darker than surrounding soils in aerial imagery after tillage, this data could be used for mapping relative SOC differences within fields. These maps, however, would not be quantitative, unless the range of SOC for a field is known. The SSURGO soil survey database can be used to estimate the SOC range for specific fields and by combining these range estimates with aerial imagery, quantitative field maps of SOC could be developed. To test this, SSURGO was used to estimate the SOC range for two fields in central Iowa. Soil organic carbon content across each field was then linearly interpolated within the estimated SOC range for each field using the brightness values at each pixel in a bare-soil aerial photograph as a scaling factor. Quantitative maps of SOC developed using this process accurately predicted SOC at specific sampling sites in the fields where measurements were taken and were superior to maps created from the SSURGO data or the aerial images separately. The impact of this process would be to make site-specific application of herbicide or nitrogen fertilizers based on SOC levels feasible for farmers and crop consultants and would result in more cost-effective and environmentally friendly application of these chemicals, which would benefit the general public.
Technical Abstract: Widespread implementation of precision agriculture practices requires low-cost, high-quality, georeferenced soil organic carbon (SOC) maps, but currently these maps require expensive sample collection and analysis. Widely available aerial imagery is a low-cost source of georeferenced data. After tillage, soils higher in organic carbon appear darker than surrounding soils in aerial imagery and this data could be used for mapping relative SOC differences within fields. The SSURGO soil survey data could be used to estimate the range of SOC for a specific field and then by combining these two data sources quantitative field maps of SOC could be developed. To test this, SSURGO was used to estimate the SOC range for two fields in central Iowa. Soil organic carbon content across each field was then linearly interpolated within the SSURGO estimated range for each field using the brightness values at each pixel in a bare-soil aerial photograph as a scaling factor. Measured SOC data from the two fields ranged from 3.4 to 50 g kg-1 and coefficients of determination between measured and estimated SOC concentrations were 0.72 and 0.75 with root mean square errors of 4.0 and 10.2 g kg-1, respectively. These coefficient of determination values were 0.39 and 0.26 higher, respectively, than those between SSURGO soil map unit SOC data and measured values. Measured SOC values and estimates based on aerial imagery and SSURGO data had similar distributions and their residuals were normally distributed, whereas SOC estimates based only on SSURGO data were not. These results imply that aerial imagery supplemented by SSURGO-estimated field SOC ranges can provide georefereced SOC estimates suitable for site-specific recommendations and analysis without field sampling.