Spatial Scale Requirements for Precision Farming: A Case Study in the Southeastern USA

Abstract

AbstractPrecision farming has created a critical need for spatial data on crop yield and related soil characteristics. However, because data are not without cost, users need practical guidelines for spatial resolution on which to collect soil and plant data. Our objectives were (i) to describe variation observed in crop response in the southeastern Coastal Plain of the USA, (ii) to compare it with variation in other regions, and (iii) to offer suggestions for precision farming practices in the southeastern Coastal Plain. From 1985 to 1995, corn (Zea mays L.), wheat (Triticum aestivum L.), soybean [Glycine max (L.) Merr.], and grain sorghum [Sorghum bicolor (L.) Moench] yields were mapped at from 10‐ to 20‐m resolution in an 8‐ha field at Florence, SC. Also available were topography (30‐m resolution), depth to clay (15 m), and in 1993, plant height on one date (9 m), canopy temperature on four dates (1.5 m), and detailed crop and soil information at selected sites. Yield of all crops in all years was significantly (P < 0.0007), though not strongly (median r2 = 0.3), correlated with soil map unit. In 1993, infrared thermometer canopy temperature minus air temperature (∆Tc) was correlated with soil map unit, even on the second day after a 46‐mm rain. Spherical semivariograms fitted to yields had ranges from 57 to 252 m (median = 79 m) and nugget/sill ratios from 0.00 to 0.56 (median = 0.32). Semivariograms for canopy temperature and plant height had ranges from 43 to 77 m. If the spatial structure for common soil characteristics matches the spatial structure for crop response, Coastal Plain soils may require study at finer resolution than the >100‐m grid that is commonly used in precision farming.