Assessing the Spatial Accuracy of the California Pesticide Use Reporting Database for Use in Exposure Assessment Studies

ehasl research project Exposure to agricultural chemicals has been associated with disease outcomes such as cancer, immune system disorders, adverse reproductive outcomes, developmental disorders, and neurological disease. Pesticide use data that provide accurate locations for where pesticides are applied are important for assessing relationships between health outcomes and pesticide exposure. The State of California has developed a Pesticide Use Reporting Database (CPUR), with the objective of providing complete agricultural pesticide use data for evaluating possible associations with health outcomes. Several recent epidemiological investigations have utilized the CPUR database. However, the spatial accuracy of the database has not been evaluated. The CPUR database was established in the 1950's with limited reporting and contains information on the pesticide type, pounds applied, date, method of application, and acres of crop treated. Beginning in 1990, a full use reporting system was instituted requiring commercial growers to report all pesticides used in agriculture. The reporting unit for the database is Public Land Survey System Section, which is approximately 1 mi2. We compared the CPUR pesticide application data by crop with high precision, 100% ground verified land-use data collected by the California Department of Water Resources (CDWR). CDWR identifies 83 specific crop types with a minimum mapping unit of 0.003 mi2. We used a Geographic Information System (GIS), to conduct a comparative analysis of the location of pesticide application by crop, as reported in CPUR, with the location of the same crops in the CDWR database. We conducted the comparative analysis for ten crops with the greatest number of pesticide applications for two counties, San Joaquin (1988) and Kings (1991). To assess the accuracy of the CPUR data, we developed a GIS analytical procedure, which computed the spatial agreement between the two datasets. The comparative analysis was performed at the Section extent (N = 3,856). We tested the resulting congruence estimates for statistical significance using a one-sided binomial test for two levels of assumed crop location error in the CDWR database (1% and 5% error). Overall agreement between CPUR and CDWR was relatively high, although statistically significant differences existed between the datasets assuming 5% error in CDWR data. Accuracy assessment indicates large variation in CPUR reporting accuracy, ranging from 73.1% for cherries to 95.1% for cotton. In general, overall agreement was significantly higher in Kings County for 1991 than for San Joaquin County in 1988 for both non-permanent crops (92.9% vs 81.9% respectively) and semi-permanent crops (87.4% vs 80.7% respectively). These results indicate spatial and temporal differences in accuracy of the CPUR dataset, at both the aggregate and individual crop level. The spatial accuracy of pesticide-use data can affect exposure assessment, and may result in misclassification of exposure in epidemiological investigations.

spatial accuracy

Miller, RS; JR Nuckols; R Gunier; A Hertz; P Reynolds. Assessing the Spatial Accuracy of the California Pesticide Use Reporting Database for Use in Exposure Assessment Studies. International Society of Exposure Analysis 12th conference and International Society for Environmental Epidemiology 14th conference, August 11-15, 2002. Vancouver, British Columbia, Canada.

View the Project in .PNG graphic Format ( 195 KB download )

(click here for help with .png files)