The herbicide-use estimates provided by Gianessi and Puffer (1991) list acres treated and pounds of active ingredient applied for a given crop in each county for which use has been estimated. Cropping data are from the 1987 Census of Agriculture, and are subject to occasional suppressions of acreage estimates at the county level due to problems of confidentiality and census disclosure rules. The herbicide-use estimates included in this coverage are totals of use on all crops treated in a given county.
The polygons representing county boundaries in the conterminous United States, as well as lakes, estuaries, and other nonland-area features were derived from the Digital Line Graph (DLG) files representing the 1:2,000,000-scale map in the National Atlas of the United States (1970).
An automated procedure was developed to process the raw herbicide-use data into ARC/INFO coverage attributes. The procedure is summarized below:
(1) copy county2m coverage to coverage called herbicide%#%, and (2) run the AML herbadd.aml for each herbicide to be added.
The herbadd.aml program runs a fortran program to total estimates of herbicide use on all crops by county, then processes these data, finally adding them as three columns of attribute data to the county coverage. Other programs were developed to calculate summary statistics of the herbicide-attribute data and to produce maps that show attribute values across the United States.
COUNTY BOUNDARIES
This series of maps was published as part of the National Atlas of the United States (U.S.Geological Survey, 1970). The maps for the conterminous United States were digitized in 15 sheets and published in the Digital Line Graph (DLG) format as described by Domeratz and others (1983).
Each sheet was prepared by reading the DLG files of the political and water-bodies layers, converting them to ARC/INFO; extracting the county boundaries and the coastline, respectively; and joining the two layers. FIPS codes were assigned to all polygons by using available sources and were checked manually.
Boundaries with adjacent sheets of the 15-sheet set were edgematched manually; one of the sheets was chosen arbitrarily as the "correct" border. Egematching operations were used to adjust the linework as far as was necessary so that the coverages would fit to a tolerance of 100 meters (328.1 feet). The coverage (referred to herein as Version 1.0) was stored as 49 separate coverages (48 States and the District of Columbia) because the ARC/INFO software in use at the time could not process the entire coverage. Individual States could be joined by specifying a tolerance of 100 meters.
From time to time, adjustments were made to the State coverages to reflect changes in counties. The accuracy of these adjustments is believed to be comparable to that of the original linework.
For Version 2.0, all State coverages were rejoined and manually edited to produce a perfect edgematch between all States. For States on the original map-sheet boundaries, this adjustment averaged less than 20 meters and in no case was more than 100 meters. The whole coverage was CLEANed to a tolerance of 20 meters (65.6 feet), which resulted in few, if any, effects on small offshore islands. The coverage also was checked to ensure that it represented current counties or county equivalents.
The coverage in Version 1.0 ended at the coastline. No attempt was made to depict offshore areas. This created problems when the coverage was used to assign county codes to sampling stations located near the coast. To help in this matter, Version 2.0 includes offshore extensions of the county polygons. The (water) boundaries of many of these polygons are arbitrary.
The Canadian Great Lakes features are another new addition to Version 2.0. They were added to improve the utility of the coverage for visual displays. Although the Canadian Great Lakes are represented logically by a single polygon, practical considerations--the inability of some software to plot polygons with a large number of vertices--made it necessary to separate them into four polygons. The dividing lines are located in narrow channels to minimize interference with plotting patterns. Canadian islands within the Great Lakes also were included.
All tick marks were relocated to places that are easily visible on maps of the United States, to help in registering maps that otherwise may not have adequate registration information.
To expedite accessing parts of the coverage, certain items have been indexed with the procedure INDEX_COUNTY.AML. See Section 3 above. A spatial index also was created.
When this coverage is used to clip or intersect other coverages, a tolerance as low as 2 meters (6.6 feet) can be used.
The processing used to derive this coverage moved boundaries from their positions on the original maps. In cases of conflicting lines, preference was given to forming the correct topology. Strictly speaking, this coverage is not identical to the source materials. These changes were unavoidable in producing a continuous coverage of the conterminous United States.
Revisions: COUNTY POLYGON DATA
Revision 1.0, 12/17/90. This revision represents many corrections and minor modifications made to this set of coverages from its construction in 1985 through the revision date.
Revision 2.0, 3/18/91. Major reworking of the coverage, combining all State coverages.
Reviews_Applied_to_Data: The herbicide-use data-processing procedure and attribute data have been peer reviewed in 1993 by Leonard Orzol and Barbara Ruddy, U.S. Geological Survey hydrologists.
The herbicide-use estimates prepared by Gianessi and Puffer (1991) were compared with National estimates provided by the Office of Pesticide Programs of the U.S. Environmental Protection Agency, and were reviewed by more than 20 individuals from the companies that registered the chemicals.
The county boundaries in this coverage were not reviewed formally. They have, however, been used in numerous applications where serious errors would have been obvious. Some State coverages were corrected following such use. The offshore polygon extensions and the Canadian Great Lakes polygons have not been reviewed.
Related_Spatial_and_Tabular_Data_Sets: This coverage is part of series of 1:2,000,000-scale base maps that cover the United States. Also available are:
COUNTY--County boundaries. STATE--State boundaries (formed from COUNTY). WATERBOD--Water Bodies. STREAM--Streams. HUC--Hydrologic cataloging units (basins).
Other_References_Cited: Domeratz, M. A., Hallam, C. A., Schmidt, W. E., and Calkins, H. W., 1983, USGS digital cartographic data standards--Digital line graphs from 1:2,000,000-scale maps: U.S. Geological Survey Circular 895-D, 38 p.
Gianessi, Leonard P. and Puffer, Cynthia, 1991, Herbicide use in the United States: Washington, D.C., Resources for the Future, December 1990 (revised April 1991), 128 p.
Nebert, Douglas D.,1994, Design of the distributed spatial data library for the Water Resources Division, U.S. Geological Survey: U.S. Geological Survey Open-File Report, 94-327, 30 p.
U.S. Department of Commerce, 1989a, Census of agriculture, 1987--Final county file: U.S. Department of Commerce, Bureau of the Census, [machine-readable data file]
U.S. Department of Commerce, 1989b, Census of agriculture, 1987--Final county file technical documentation: U.S. Department of Commerce, Bureau of the Census, Data User Services Division.
U.S. Geological Survey, 1970, National atlas of the United States of America: Washington, D.C., U. S. Geological Survey, 417 p.
Notes: DOCUMENTATION
The documentation for this coverage was generated by using the ARC macro language (AML) program document.aml available from the U.S. Geological Survey (Nebert, 1994).
Estimates of herbicide use by county were generated by Gianessi and Puffer (1991) by using the following procedure:
(1) collect statistics by State, by crop, on percentage of acres treated with a given herbicide, and average annual application rate of the herbicide from surveys sent to Extension Service weed scientists in 1987 and 1989;
(2) augment survey data with published information from some States;
(3) establish herbicide-use profiles, by State, by crop, containing the percentage of acres treated, and average annual application rates;
(4) apply data in herbicide-use profiles to county-level crop-acreage estimates from the 1987 Census of Agriculture (U.S. Department of Commerce, 1989a); and
(5) tabulate pounds of active ingredient of herbicides used by crop, by county.
Eighty-four crops were profiled, resulting in usage estimates for 96 herbicide active ingredients. This coverage contains estimates of use for the eighty-first through the ninety-sixth the most-used herbicides in the conterminous United States.
Although crop-acreage data represent the 1987 growing year, the herbicide-use estimates generally reflect 1989 usage amounts (Gianessi and Puffer, 1991).
Estimates of herbicide use by county were generated from a combination of data collected from surveys of weed scientists and from surveys of farmers. Herbicide-use profile data were generated from ancillary data sources or were calculated from profiles in neighboring states when no responses to surveys or published reports were available (Gianessi and Puffer, 1991). Sampling and statistics were used to account for nonresponding farm operations. Thus, the information describing crop acreages is subject to sampling variability as well as reporting and coverage errors (U.S. Department of Commerce, 1989b). Census disclosure rules also prevent the publication of information that would disclose the operation of individual farms.
Before herbicide-use estimates entered into this coverage, the total use of individual herbicides on all crops for each county was calculated.
COUNTY BOUNDARIES
The original files for this map were provided in 15 sections. Boundaries near the edges of sections have been adjusted in edgematching.
Polygons that extend into the water (an ocean or the Great Lakes) should be considered arbitrary.
Herbicide-use attributes are from tabular data files provided by Gianessi and Puffer (1991). Three attributes are added to an ARC/INFO coverage for each of the 96 herbicides. Attributes with names ending in .ACR are estimates of the number of acres treated with a given herbicide. Attributes with names ending in .LBS are estimates of the number of pounds of active ingredient applied of a given herbicide. Attributes ending in .USE are estimates of a county-level herbicide-use rate in pounds per square mile.
Attribute names are keyed to four-digit herbicide code numbers provided in the table below. A ranking of national herbicide-use amounts (Gianessi and Puffer, 1991) was used to divide herbicide-use data into five coverages (herbicide1, herbicide2, herbicide3, herbicide4, and herbicide5)
Table 1. Herbicide codes for herbicide5 coverage
Code Herbicide _______________________ 4009 Lactofen 1867 Isopropalin 1913 Chlorsulfuron 1988 Triclopyr 2053 Chloroxuron
1865 Dichlobenil 4004 Thiameturon 1176 Barban 1889 MCPB 4003 Metsulfuron
1477 MCPP 5003 Fenoxaprop 4002 Cylopyralid 1885 CDAA 1397 Fluchloralin
1984 Siduron _______________________
Herbicide-use estimates in counties represented by one polygon are equal to county totals. For counties described by more than one polygon, attribute values are provided for each polygon and are calculated as the county total multiplied by the ratio of the area of each county polygon divided by total area of the county.
A missing-value code of -99.0 is used when no herbicide-attribute data are given for a county and for all nonland-area polygons. The missing-value code can indicate any of the following conditions:
-- herbicide use not present in county (for example, no atrazine used in county) -- crop data withheld because of census disclosure rules -- crop data not available to the census -- crop data not published because county contains fewer than 10 farms.
COUNTY POLYGON ATTRIBUTES
Each county is identified by a five-digit Federal Information Processing Standard (FIPS) code. The first two digits represent the State and the three remaining represent the county within the State. County names (in upper case) and the two-character State abbreviation also are included.
All polygons are coded by polygon type (PLYTYPE) as follows: 0 = U.S. land. 1 = Extensions of counties into the ocean and estuaries. Some boundaries may be arbitrary. 2 = U.S. Great Lakes. Some boundaries between counties may be arbitrary. 3 = Canadian Great Lakes. 4 = Canadian islands in Great Lakes, included for visual cartographic purposes. 9 = External polygon.
All arcs are coded by boundary type (BNDTYPE) as follows: 0 = Coastline of United States. 1 = State-State border located on land. 2 = County-county border, within State, located on land. 3 = State-State border located on the ocean, an estuary, or a Great Lake. Some of these borders are arbitrary. 4 = County-county border, with State, but located on an ocean, an estuary, or Great Lake. Some of these borders are arbitrary. 5 = International border located on land. 6 = International border located on water. 7 = Canadian shoreline of the Great Lakes. 8 = Arbitrary separation line between the Canadian portions of the Great Lakes. 9 = Closure line with the external polygon. This is arbitrary.
All BNDTYPE values are determined from the adjoining polygons by the routine INDEX_COUNTY.AML. The coverage is indexed (ARC: INDEXITEM) on FIPS code, State FIPS code, State abbreviation, county name, polygon type, and boundary type.