Automated Crop Type Mapping from Landsat Imagery

Knowledge of the spatial distribution of crop types is important for many environmental and human health research studies. These studies may require crop type maps over large geographical regions for multiple years, which can be derived only from satellite images. Interpretation of Landsat images using traditional image classification techniques is not feasible for such applications because of the large number of images to be analyzed. We present a new method that automates the interpretation process resulting in a less expensive and more timely product. Software was developed to use readily available agriculture data to automatically extract spectral training statistics from target areas on the satellite images. The statistics are then used to process the remainder of the image, county by county, without intervention from the analyst. The Mahalanobis distance measurement is used in the final map to provide a measure of confidence – important for further modeling efforts. To demonstrate the feasibility of this approach, we produced a map for a single crop type (corn), using a Landsat Multispectral Scanner image in eastern Nebraska. Thirteen counties (3.35 million hectares) were classified in less than 15 minutes. The resulting map classifies the land area as either ‘highly likely to be corn’, ‘likely to be corn’, or ‘unlikely to be corn’. Ground reference data from three counties were used to assess the accuracy of our method. The resulting average classification accuracy of 89 percent is comparable to traditional methods. Introduction Knowledge of the spatial distribution of specific crop types is important for many environmental and health studies (Kellogg et al. 1992, Wood et al. 1995, Nuckols et al. 1996a,b), Gilliom and Thelin 1997. In many instances such studies need crop types maps over large geographical regions (e.g., multi-county, entire state) for multiple years in order to determine statistically significant relationships between environment and disease occurrence. For example, in a study of agricultural chemical use and occurrence of cancer, once location of crops can be determined, important parameters such as pesticide use can be estimated and incorporated into an environmental model for exposure assessment (Ward et al. 1999). Such maps covering extensive geographical regions can only be derived from satellite imagery such as Landsat TM. Landsat satellite imagery has been collected since the early 1970’s and has been successfully used to classify many different crop types (Myers 1983). However, the classification process can be very time consuming using traditional methods (i.e., supervised, unsupervised). In general, traditional methods