How close is close enough?

Modern sensor and GPS technologies are associated with an overwhelming amount of data, and who knows what to do with it? One aspect of information to be grasped is the processing and spatial aggregating of data to derive an idea about the scale-dependent change of information and appropriate management scale. The objective of this study was to study the spatial covariance and cross-correlation behavior of wheat grain yield and springmeasurements of NDVI at different scales. The underlying field experiment was a nitrogen fertilization response study in which different rates of nitrogen fertilizer were not randomly distributed but continuously varied across a 27 m wide by 644 m long strip in a farmer's field. Down to an aggregation scale of 6 by 10 m, yield and NDVI are scale-invariant with respect to their covariance and cross-correlation behavior. At resolutions of 4 by 5 m and 2 by 2 m, the variance increased and correlation decreased substantially. However, 4 by 5 m appears to be still a reasonably manageable scale. Under the assumptions of ANOVA, this fertilizer study would have barely shown a significant yield response to fertilizer. Nevertheless, the design and focus on spatial processes of integrative state variables in this study are obviously a promising way towards better understanding soil and crop processes at the field-scale.