Monitoring spatial structure of crop canopies using semivariograms

The aim of the study is to analyze the seasonal development of crop canopies from the perspective of spatial domain of remotely sensed imagery for monitoring agricultural fields. The study site is located in eastern Nebraska, USA. Multidate high resolution aerial imagery using Kodak DCS420IR (including red and NIR bands) was acquired across the growing season over corn and soybean crops in 2000. Image datasets were first coregistered and then normalized using pseudoinvariant features, and NDVI was then computed. Spatial structure analyses were carried out using semivariograms. Results indicate that sill+nugget variance for NDVI component bands decreases as the crop develops for both corn and soybeans, suggesting the crop canopies become increasingly even, the range for red band also decreases for both crops but not for NIR band, whose range first increases but then decreases. Differential spatio-temporal patterns were found between corn and soybean NDVI for both sill+nugget variance and range, likely due to the difference in their developmental stage and canopy architecture. These findings suggest that NDVI is sensitive to the development of crop canopies and thus can be applied effectively to capture changing spatial structure of reflectance during the crop development