Geospatial Land Use and Land Cover Data for Improving Agricultural Area Sampling Frames

Abstract

This chapter describes the development of an automated stratification method, based on readily available geospatial Cropland Data Layers and its integration into the USDA National Agricultural Statistic Service operational stratification process. Automated stratification, based on the NASS Cropland Data Layers https://nassgeodata.gmu.edu/CropScape/, objectively, consistently, and rapidly stratifies United States land cover of the area frame primary sampling units, by percent cultivation. Subsequently, the integration of automated stratification into the traditional operational process further refines area frames with manual editing and review processes and significantly improves operational efficiency and area frame accuracy. A performance comparison is described between the NASS traditional stratification method, which is based on subjective visual interpretation of aerial or satellite data, and the Cropland Data Layer based automated stratification method. The effectiveness of both traditional and automated stratification methods is also assessed using in situ validation data collected at the segment level as part of the 2010 NASS June Area Survey. Though the automated stratification method improves efficiency and objectivity as well as accuracy in the intensively cropped areas, it inherits the Cropland Data Layer classification errors and has lower accuracies in low or non-agricultural areas. To further improve the accuracy of the automated method in low intensity agricultural areas and maintain its efficiency and objectivity, the automated method is integrated into the traditional stratification process to further refine area frames with manual editing and review procedures. The results of the integration have shown further improvement in stratification accuracy.