Biophysical and yield information for precision farming from near-real-time and historical Landsat TM images

The main goal of this study was to quantify within and between field variability in mapping agricultural crop types, their biophysical characteristics, and yield for precision-farming applications using near-real-time and historical (archival) Landsat Thematic Mapper (TM) images. Data for six crops (wheat, barley, chickpea, lentil, vetch and cumin) were gathered from a representative benchmark study area in the semi-arid environment of the world. Spectro-biophysical and yield models were established for each crop using a near-real-time TM image of 6 April 1998 acquired to coincide with an extensive ground data collection campaign. The models developed using this near-real-time acquisition were then used to extrapolate and quantify characteristics in the historical Landsat TM images of 5 April 1986 and 4 May 1988 acquired for the same area with limited ground data, thus adding scientific and commercial value to archival TM images. A farm-by-farm (or pixel-by-pixel) within and between field variability in agricultural land cover, biophysical quantities [e.g. biomass and Leaf Area Index (LAI)] and yield was established and illustrated. For the near-real-time image of 1998: (a) quantitative biophysical characteristics such as LAI and biomass were mapped at 81% overall accuracy ( K hat =0.76) or higher; (b) within field variability (commission errors) was mapped with an accuracy between 74-100%; and (c) between field variability (omission errors) was mapped with an accuracy between 76-100%. Temporal variability in biomass and LAI were mapped for the study area and highlighted for individual farms. Significant relationships existed between grain yields measured using field-based combine-mounted sensors and Landsat TM derived indices. The results demonstrate the ability of using near-real-time and historical Landsat TM images for obtaining quantitative biophysical and yield information that highlight within and between field variability, which is of critical importance in precision-farming applications.