Authors: Zhengwei Yang*, USDA - National Agricultural Statistics Service, Robert Seffrin, USDA - National Agricultural Statistics Service
Topics: Agricultural Geography, Land Use and Land Cover Change, Spatial Analysis & Modeling
Keywords: Remote Sensing, Cropland Cover Data, GIS Administrative Data, Crop Acreage Estimation
Session Type: Paper
Start / End Time: 1:10 PM / 2:50 PM
Room: Tyler, Marriott, Mezzanine Level
Presentation File: No File Uploaded
The US Department of Agriculture (USDA) National Agricultural Statistics Service (NASS) is developing a new method for estimating planted acreage from its Cropland Data Layer (CDL). NASS currently runs a regression using data from square-mile sized agricultural land areas that are enumerated in the June Area Survey (JAS) and data from tabulated pixels from the CDL for the same agricultural area to generate state level estimates. The regression adjusts for biases in the CDL, but there are few or no representative data from many counties. This research replaces the JAS data with Farm Service Agency’s Common Land Unit (CLU), which is a field or multiple field areas with reported crops. These CLUs are aggregated into approximately one square-mile units and are the basis for summing reported crops and tabulating pixels from the CDL. This increases the coverage from about two percent to about ninety-five percent of agricultural land. The CDL crop data are regressed against CLU crop data using robust regression methods to identify and remove outliers. In this study, 2017 data from the State of Nebraska is used. The preliminary results show that the proposed method enables county estimation for the minor crops such as dry beans, sorghum and oats, for which the estimates were previously not available. Moreover, the root-mean-square-error of individual county estimates significantly improved (by at least 30%) and the correlation coefficients of corn, soybeans and wheat crops were high at 0.976, 0.977 and 0.94, respectively.