Comparing Geographic Information System (GIS) Calculated Acreage to Farmer Reported Acreage Utilizing a Mobile Mapping Instrument

The National Agricultural Statistics Service (NASS) conducts the annual June Area Survey (JAS), which is based on an area sampling frame. Segments of land comprise the JAS sampling units. Field enumerators use hard copy aerial photos that contain outlines of the sample unit or segment boundaries to locate and to interview all farmers within each sampled unit. Paper questionnaires are used to record the farmer’s agricultural activity occurring within each segment and additional information about their entire operation. June Area Survey sampled segment boundaries follow physical features (roads, railroads, rivers) on the ground. As a cost saving initiative, NASS is evaluating the use of a permanent grid area frame with sampling units of roughly equal size and area with data collection conducted using a mobile mapping instrument. The permanent grid frame would be based on the Public Land Survey System’s (PLSS’s) one-square-mile sections in the 30 states in which the PLSS is the primary surveying method. The sample units of the permanent grid frame are referred to as grid cells in this paper. In these 30 states, roads are often aligned with the PLSS section lines. However, exceptions to this rule and gaps in PLSS coverage exist. In states where land surveying is based on alternate systems (non-PLSS states), a grid frame, with one-square-mile sections would be generated using ESRI’s ArcGIS software. These grid cells would not be aligned with physical features on the ground. Due to these characteristics and the fact that grids cells commonly cut across agricultural fields, a mobile mapping instrument was developed to use Geographic Information Systems (GIS) technology to calculate the acreage of the fields located in the grid cells. The primary objective of this study is to determine whether a GIS calculation of acreage is comparable to the acreage reported by JAS farmers. If so, research into the use of a permanent grid sampling frame can move forward. Additionally, research into the use of a mobile mapping instrument for JAS data collection, based on grid cells or JAS segments, can advance with increased reliance on the GIS calculation of acreage. For this study, field enumerators used the aerial photos from previously collected 2013 JAS data to delineate field boundaries in the prototype mobile mapping instrument. The calculated GIS acreage, at the segment and crop field level, was then compared to the acreage reported by JAS 2013 farmers. Ninety segments in Indiana, Pennsylvania and Washington were included to assess segment and crop field acreage differences. Summary statistics and p-values from Wilcoxon signed-rank tests for total segment and crop field acreage differences between GIS calculated values and JAS farmer reported values are reported. To determine whether there was any bias in segment-level and crop-field-level acreage, estimated using the GIS software, the median difference in acreage was examined and was found to be zero. Wilcoxon signed-rank tests were performed for segment totals and at field level for three crop field categories. Results indicated that, for segment totals and for all crop field types in all states, with the exception of Indiana’s boundary fields (p-value = 0.04), the GIS calculation of acreage and the JAS farmer reported acreages were not statistically different.