5. Processing the Data
Pesticide use reports are submitted to the agricultural commissioners' offices where county staff reviews them. There are two ways this information is entered into DPR's pesticide use database:
By mid-1997, 56 of the state's 58 counties enter all agricultural and nonagricultural pesticide use reports submitted to their offices into a computer database. (These counties represent more than 99 percent of the pesticide use in the state subject to reporting requirements. The limited pesticide use in the remaining two counties does not warrant an investment in electronic data processing.)
To process a use report, county employees enter the operator ID number. Matching grower data already in the system are displayed on the screen and the records are scanned until the correct site ID and commodity/crop combination are displayed. Data such as the site ID, commodity code, planted acreage, and location (section, township, range, base, and meridian) are copied to the use report record; information specific to the application is then entered (date and time, treated acres, application method, U.S. EPA or California registration number, and amount of pesticide used). After a record has been entered, it is transferred to a use report database on the county's computer. Extensive validity checks of the entered data are made against the grower's data files and other databases, such as the registered product database described below. Periodically, the use report data are downloaded to an electronic file that is then transferred to DPR's Pesticide Enforcement Branch via floppy disk or electronically via the Internet.
- For the two counties not electronically processing their data, county biologists manually review the reports for completeness. The reports are then mailed to DPR where they are entered into the database.
As the use reports are loaded into DPR's database, another round of up to 50 different validity checks is made against the data. In particular, the product registration number is verified and a check is made to validate that the commodity reported is a legal use of the pesticide product. Pesticide use records passing all validation checks are transferred to the main "production" database. An error listing of records failing one or more checks is generated. If the data were entered by a county, DPR sends the reports to the county for correction. If DPR staff entered the data, the physical use report is pulled and reviewed. If there was a coding or data entry error, it is corrected and reentered into the database. Reporting errors, such as an incorrect product registration number, are returned to the county for resolution.
The error checking typically removes less than one percent of the PUR records. Some of the remaining records have extremely high values that significantly affect total pounds applied of a pesticide. For example, the 1995 database contains one reported carbaryl application of 596,511 pounds on five acres of oranges. The median rate of carbaryl use on oranges in 1995 was 12 pounds per acre. In the past, this type of error was only identified by chance when the data were used for analytical projects.
DPR developed a statistical method to detect probable errors in the data fields for acres treated and the pounds of pesticide used. Called the outlier program, this method calculates pesticide use rates (pounds of active ingredient applied divided by acres treated) that are then examined using a variety of statistical methods. The records with highly unlikely use rates (outliers) are placed in a file which accompanies electronic versions of the annual use report, thereby serving to flag suspect pesticide use records. In addition, these records are returned to the counties for research and correction. (See Appendix A for a discussion of the creation of the outlier program.)
An integral part of the use reporting system is DPR's database on registered pesticide products (also called the product label database). Since the early 1970s, DPR has maintained a database on all pesticide products currently (and previously) registered for use in California. The database contains information on more than 45,000 pesticide products. Approximately 10,000 are active product registrations. An average of 1,000 new products is added to the database annually, and a similar number are inactivated due to nonrenewal, suspension, or cancellation. Between 2,000 and 2,500 label amendments are processed annually, many changing the kind and number of crops on the label. Data fields in the product label database include: U.S. EPA or California registration number; pesticide product name; type of registration; type of pesticide; formulation; active ingredients; percent of each active ingredient; specific gravity; all commodity/crop/sites on which the product may be used; health and environmental hazards; general categories of target pests; and application instructions. DPR staff and many outside groups use the database, which is available on DPR's web site. Non-confidential information is routinely made available to the public, registrants, county agricultural commissioners, poison control centers, the state Legislature, and other governmental agencies.
All use report data are run against the product label database as a check for accuracy and to convert the pounds of product applied to pounds of active ingredient(s), based on the formulation and percentage of each active ingredient in the product. In addition, the reported commodity/crop/site is checked against the list of registered uses on the label. Products no longer actively registered are retained in the database since continued use of those products is often allowed (while existing stocks remain with end users).