Naomi Nakagaki 200708 Version 2.0 raster digital data Reston, Virginia U.S. Geological Survey http://water.usgs.gov/lookup/getspatial?agpest97grd This spatial dataset consists of 219 1-kilometer (km) resolution grids depicting estimated agricultural use of 219 pesticides in 1997 for the conterminous United States. Each grid cell value in the national grids of this dataset is the estimated total kilograms (kg) of a pesticide applied to row crops, small grain crops and fallow land, pasture and hay crops, and orchard and vineyard crops within the 1- by 1-km area. Nonagricultural uses of pesticides are not included in this dataset. Of the 219 pesticidesrepresented in the grids, 96 are herbicides, 65 are insecticides, and 37 are fungicides. The remaining 21 compounds are composed of the category "other pesticides", which consist of fumigants, growth regulators, and defoliants. Although this dataset is referenced to 1997, it generally represents a composite of estimated pesticide use during the mid to late 1990s. The national grids of the agricultural use of 219 pesticides were developed to support hydrologic studies that are part of the U.S. Geological Survey's (USGS) National Water Quality Assessment (NAWQA) Program. The goal of the NAWQA program is to develop long-term consistent and comparable information on streams, ground water, and aquatic ecosystems to support sound management and policy decisions (U.S. Geological Survey, 2001; Gilliom and others, 1995). The NAWQA program, which began in 1991, consists of long-term cyclical studies in over 50 major river basins and aquifers in the United States. Pesticide-use grids are used in the NAWQA program to: 1) determine overall estimates of agricultural pesticide use in major watersheds (as documented in Nakagaki and Wolock, 2005) and other large study areas; 2) provide input into national models, such as the regression model for predicting concentration of pesticide distributions in unmonitored rivers and streams (as described in Larson and others, 2004); and 3) display the spatial distribution and use intensity of agricultural pesticides in the conterminous United States in USGS publications (as shown in Gilliom and others, 2006). Purpose References: Gilliom, R.J., Alley, W.M., and Gurtz, M.E., 1995, Design of the National Water-Quality Assessment Program--occurrence and distribution of water-quality conditions: U.S. Geological Survey Circular 1112, 33 p. Gilliom, R.J., Barbash, J.E., Crawford, C.G., Hamilton, P.A., Martin, J.D., Nakagaki, Naomi., Nowell, L.H., Scott, J.C., Stackelberg, P.E., Thelin, G.P., and Wolock, D.M., 2006, The Quality of Our Nation's Waters -- Pesticides in the Nation's Streams and Ground Water, 1992-2001: U.S. Geological Survey Circular 1291, 172 p. Larson, S.J., Crawford, C.G., and Gilliom, R.J., 2004, Development and application of watershed regressions for pesticides (WARP) for estimating atrazine concentration distributions in streams: U.S. Geological Survey Water-Resources Investigations Report 03-4047, 68 p. Nakagaki, Naomi, and Wolock, D.M., 2005, Estimation of agricultural pesticide use in drainage basins using land cover maps and county pesticide data: U.S. Geological Survey Open-File Report 2005-1188, 46 p. U.S. Geological Survey, 2001, The National Water Quality Assessment Program-- entering a new decade of investigations: U.S. Geological Survey Fact Sheet 071-01, accessed 1/30/06, at http://pubs.usgs.gov/fs/fs-071-01/ Update information: This dataset is referred to as "version 2" because 176 grids were added to the initial release (January 2007) of 43 1-km resolution grids of agricultural use of pesticides. All 219 pesticide grids were processed from the same data sources and methodology. Limitations of use: The pesticide grids are not intended to be used to identify pesticide use on a cell-by-cell basis but can be used to estimate use by major watershed or other defined region. This dataset is an interpretive product, created by combining county pesticide-use estimates with mapped agricultural land and crop types that were classified from satellite imagery. Users are warned that although data values are stored as floating numbers, this does not reflect the accuracy of the amount of pesticides applied in a 1- by 1 km area. Please read the "Use_Constraints" and "Entity_and_Attribute_Overview" sections of this metadata file prior to using this dataset. Background and detailed description of this dataset: County tabular data of agricultural pesticide use combined with a digital map of county boundaries can be used to generate countywide choropleth maps of agricultural pesticide use. However, such a map does not show the spatial distribution of agricultural land to which the pesticides are applied, nor the application rate of the pesticides on agricultural land. In contrast, the agricultural pesticide grids in this dataset can be used to map the spatial distribution and estimated use intensity of pesticide applications on agricultural land in the conterminous United States. This dataset can also be used to estimate agricultural pesticide use in relatively large areas of interest by clipping it to the boundaries of study areas and summing the estimated grid cell values within the study areas. The data sources used to generate this national dataset were 1) a digital map of county boundaries, 2) county tabular file of agricultural pesticide use, and 3) a digital map of agricultural land. These data sources are described below, along with the manner in which they were used to generate the agricultural pesticide use grids. The source for a digital map of county boundaries was the 1990 county boundaries of the conterminous U.S., at the 30-meter (m) resolution (Jo Ann M. Gronberg, U.S. Geological Survey, written commun., 2005), which was derived from polygons of 1990 county boundaries (U.S. Department of Commerce, 1993) and the U.S. shoreline boundary available from the National Oceanic and Atmospheric Administration (1994). A 1-km resolution grid of county boundaries was produced from the 30-m resolution grid of county boundaries, both of which were used to develop the national pesticide grids. The source for agricultural pesticide use was the 1997 county tabular file of 220 pesticides applied to 87 crops in the conterminous U.S. (Thelin, 2005). Out of the 220 compounds in Thelin's (2005) dataset, 219 pesticides are included in this dataset. BT, an insecticide, which is included in Thelin's (2005) dataset is not available as a pesticide grid because the use estimates are zero for all counties. The 1997 county file of pesticide use (Thelin, 2005) was used to determine the total amount of pesticides applied to each of the 3 following groups of crops, or "crop classes": 1) row crops, small grain crops, and fallow land hereinafter abbreviated as "CGF"), 2) orchard and vineyard crops (hereinafter abbreviated as "ORCH"), and 3) pasture and hay crops (hereinafter abbreviated as "PAST"). (For a listing of individual crops by crop class, refer to table 3 in Nakagaki and Wolock, 2005.) The pesticide totals for the 3 crop classes were calculated to correspond to the agricultural land classifications in the digital map of land cover. The source for mapped agricultural land was an enhanced version of the 30-m resolution National Land Cover Dataset 1992, or "NLCD 92". The original NLCD 92 consists of 21 land classifications (U.S. Geological Survey, 1999) and the enhanced version (hereinafter abbreviated as "NLCDe 92") consists of 25 land classifications. The 4 added classifications in the enhanced version were created by combining the original NLCD 92 with selected land categories from USGS's Land Use and Land Cover (LULC) dataset (Fegeas and others, 1983; U.S. Geological Survey, 1990; U.S. Geological Survey, 1998), as enhanced by Price and others (2003, 2007). The original and enhanced NLCD are described in Vogelmann and others (2001) and Nakagaki and Wolock (2005), respectively. "Agricultural land" was based on the following six classifications in the NLCDe 92: "row crops," "small grains," "fallow," "pasture/hay," "orchards/vineyards/other," and "LULC orchards/vineyards/other" (land cover codes 82, 83, 84, 81, 61, and 62, respectively). The "row crops," "small grains," and "fallow," were combined into 1 classification (CGF); the 2 orchard/vineyard/other classes were also combined (ORCH); and the "pasture/hay" remained as a class by itself (PAST). A 30-m resolution grid for CGF, PAST, and ORCH was created, each of which were overlaid with the county boundaries map to compute county areas of CGF, PAST, and ORCH. To develop the agricultural pesticide use grids, the objective was to apportion, for each county, the amount of pesticide used for a group of crops, to the areas of land associated with the same group of crops. For example, 200 kilograms of atrazine use on CGF estimated for county A would be apportioned to areas of land classified as CGF in county A. While this approach worked for the majority of counties, it was not possible for counties that had pesticide use estimates for CGF, PAST, or ORCH but no areas of land classified as CGF, PAST, or ORCH, respectively. This problem stemmed from conflicting definitions and/or classifications of cropland between the 1997 Census of Agriculture (U.S. Department of Agriculture, 1999) and the NLCDe 92. The differing dates of data capture further contributed to the discrepancies in cropland areas. As a result, it was necessary to adjust the protocol for the problematic counties that were missing areas of land associated with a crop class but had estimated pesticide use for the same crop class. For these counties, the land areas associated with one of the other 2 crop classes served as a substitute on which the pesticide use would be distributed. For example, if there was pesticide use on CGF in a county and the same county lacked areas of land classified as CGF in the NLCDe 92, the 30- by 30-m areas coded as PAST (according to the NLCDe 92) served as the areas to which CGF pesticide would be applied. In order to identify the counties that would undergo this adjustment process, each county was checked for specific conditions. Any county that had pesticides applied to crops in the CGF, PAST, or ORCH classes but lacked area of land classified as CGF, PAST, or ORCH respectively, was evaluated to determine the alternative crop class to be used as a substitute. A hierarchy of rules was established to determine the alternative crop class for the problematic counties, as summarized below: Missing Crop Class ------1st Alternative-----2nd Alternative ORCH -----------CGF---------------PAST CGF -------------PAST-------------ORCH PAST ------------CGF--------------ORCH The first alternative crop class would be incorporated if it existed; otherwise, the 2nd alternative crop class would be incorporated. If both alternatives didn't fit the condition, the pesticide use for the county would be distributed evenly throughout the county (occurred for 2 counties missing PAST). The most common example of missing agricultural land in a county occurred in the ORCH crop class. (Refer to Process_Step #5 for additional information on the adjustment process.) The crop class adjustments were conducted in over 800 counties. Once an alternative crop class was assigned to the problematic counties, a 30-m resolution grid for each crop class was generated. The grid that was used to distribute CGF pesticide use ("CGF grid"), mostly consisted of the 30-m cells coded as "row crops," "small grains," or "fallow" in NLCDe 92, but also some 30-m cells coded as "pasture/hay"; the grid that was used to distribute ORCH pesticide use ("ORCH grid") consisted of 30-m cells coded as "orchards/vineyards/other" or "LULC orchards/vineyards/other" but also many cells coded as "row crops," "small grains," or "fallow"; and the grid that was used to distribute PAST pesticide use ("PAST grid") consisted mainly of 30-m cells coded as "pasture/hay" but also all 30-m cells in 2 counties in which there were no cells classified as PAST, CGF, or ORCH. Next, a 1-km resolution representation was created for each of the 30-m resolution CGF, PAST, and ORCH grids. These 1-km resolution grids were expressed as percentages so that each 1-km grid cell was equal to the percentage of 30-m cells that were found to represent the crop class. The 1-km resolution CGF, PAST, and ORCH percentage grids were then merged with the 1-km resolution county boundaries grid attributed with county application rates of each pesticide. (The county application rates were derived from the county totals of pesticide use (kg) by crop class divided by the county area of cropland by crop class.) Through a series of geoprocesses as described in the "Process_Step" section, the final product is a national grid of total kilograms of a pesticide applied to agricultural land within each square kilometer grid cell (nonagricultural uses are not included). Naming Convention of the pesticide grids: The 219 grids in this dataset are named "kg" concatenated with the 4-digit numeric code for a pesticide. The codes and names of the 219 pesticides in this dataset are listed below. Code Pesticide 8008 1,3-D 1302 2,4-D 1308 2,4-DB 6001 ABAMECTIN 6002 ACEPHATE 3000 ACETOCHLOR 1002 ACIFLUORFEN 1863 ALACHLOR 6003 ALDICARB 1982 AMETRYN 6091 AMITRAZ 9048 ASULAM 1980 ATRAZINE 2001 AZADIRACHTIN 6004 AZINPHOS-METHYL 2002 AZOXYSTROBIN 1362 BENEFIN 5001 BENOMYL 7009 BENSULFURON 1098 BENSULIDE 1287 BENTAZON 2003 BENZYLADENINE 6063 BIFENTHRIN 1809 BROMACIL 1116 BROMOXYNIL 2004 BUPROFEZIN 2005 BUTENOIC ACID 1839 BUTYLATE 8017 CACODYLIC ACID 5014 CAPTAN 6006 CARBARYL 6007 CARBOFURAN 2006 CHLORETHOXYFOS 4008 CHLORIMURON 8000 CHLOROPICRIN 5007 CHLOROTHALONIL 6009 CHLORPYRIFOS 1913 CHLORSULFURON 7010 CLETHODIM 7204 CLOFENTEZINE 9001 CLOMAZONE 4002 CLOPYRALID 5011 COPPER 6010 CRYOLITE 1369 CYANAZINE 2007 CYCLANILIDE 2069 CYCLOATE 6081 CYFLUTHRIN 3008 CYMOXANIL 6011 CYPERMETHRIN 6012 CYROMAZINE 8015 CYTOKININS 5008 DCNA 1872 DCPA 2008 DELTAMETHRIN 9014 DESMEDIPHAM 6014 DIAZINON 1298 DICAMBA 1865 DICHLOBENIL 1005 DICLOFOP 6016 DICOFOL 6082 DICROTOPHOS 1374 DIFENZOQUAT 6064 DIFLUBENZURON 3001 DIMETHENAMID 7004 DIMETHIPIN 6017 DIMETHOATE 3006 DIMETHOMORPH 1950 DIQUAT 6018 DISULFOTON 1991 DIURON 5033 DODINE 4001 DSMA 6019 ENDOSULFAN 1948 ENDOTHALL 1414 EPTC 6020 ESFENVALERATE 9009 ETHALFLURALIN 7003 ETHEPHON 6022 ETHION 9012 ETHOFUMESATE 6023 ETHOPROP 6024 ETHYL PARATHION 5051 ETRIDIAZOLE 6025 FENAMIPHOS 5032 FENARIMOL 2009 FENBUCONAZOLE 6026 FENBUTATIN OXIDE 9003 FENOXAPROP 7203 FENPROPATHRIN 5017 FERBAM 9007 FLUAZIFOP 2010 FLUMETRALIN 3003 FLUMETSULAM 2011 FLUMICLORAC 1998 FLUOMETURON 2012 FLUTOLANIL 4010 FOMESAFEN 6028 FONOFOS 6071 FORMETANATE HCL 5031 FOSETYL-AL 8013 GIBBERELLIC ACID 1099 GLYPHOSATE 2013 HALOSULFURON 2070 HEXAZINONE 2014 HEXYTHIAZOX 7001 IMAZAMETHABENZ 2015 IMAZAPIC 4005 IMAZAQUIN 9000 IMAZETHAPYR 3004 IMIDACLOPRID 5006 IPRODIONE 4009 LACTOFEN 6083 LAMBDACYHALOTHRIN 6032 LINDANE 1993 LINURON 6033 MALATHION 8010 MALEIC HYDRAZIDE 5000 MANCOZEB 5009 MANEB 1305 MCPA 1889 MCPB 1477 MCPP 2016 MEFENOXAM 8007 MEPIQUAT CHLORIDE 5002 METALAXYL 6073 METALDEHYDE 8002 METAM SODIUM 6036 METHAMIDOPHOS 6037 METHIDATHION 6038 METHOMYL 6039 METHOXYCHLOR 8001 METHYL BROMIDE 6042 METHYL PARATHION 5029 METIRAM 1011 METOLACHLOR 1975 METRIBUZIN 4003 METSULFURON 1417 MOLINATE 1124 MSMA 5036 MYCLOBUTANIL 8003 NAA 8014 NAD 6044 NALED 1900 NAPROPAMIDE 1307 NAPTALAM 7007 NICOSULFURON 1018 NORFLURAZON 6049 OIL 1873 ORYZALIN 6045 OXAMYL 6046 OXYDEMETON-METHYL 4000 OXYFLUORFEN 5038 OXYTETRACYCLINE 6047 OXYTHIOQUINOX 1616 PARAQUAT 5021 PCNB 1419 PEBULATE 1629 PENDIMETHALIN 6048 PERMETHRIN 2220 PHENMEDIPHAM 6050 PHORATE 6051 PHOSMET 1051 PICLORAM 7008 PRIMISULFURON 6084 PROFENOFOS 1987 PROMETRYN 1888 PRONAMIDE 1191 PROPACHLOR 3007 PROPAMOCARB 1282 PROPANIL 6055 PROPARGITE 5020 PROPICONAZOLE 2017 PROSULFURON 2250 PYRAZON 2018 PYRIDABEN 7012 PYRIDATE 2019 PYRIPROXYFEN 2020 PYRITHIOBAC 7013 QUINCLORAC 7006 QUIZALOFOP 2027 RIMSULFURON 1910 SETHOXYDIM 1981 SIMAZINE 8004 SODIUM CHLORATE 2021 SPINOSAD 5037 STREPTOMYCIN 2022 SULFENTRAZONE 5004 SULFUR 8016 SULFURIC ACID 6059 SULPROFOS 2023 TEBUCONAZOLE 3005 TEBUFENOZIDE 2024 TEBUPIRIMPHOS 1963 TEBUTHIURON 6066 TEFLUTHRIN 1109 TERBACIL 6060 TERBUFOS 8006 THIDIAZURON 4004 THIFENSULFURON 1903 THIOBENCARB 6061 THIODICARB 5019 THIOPHANATE METHYL 5022 THIRAM 6067 TRALOMETHRIN 5015 TRIADIMEFON 1790 TRIALLATE 7011 TRIASULFURON 7002 TRIBENURON 8009 TRIBUFOS 1988 TRICLOPYR 2025 TRIFLUMIZOLE 1361 TRIFLURALIN 2026 TRIFLUSULFURON 5003 TRIFORINE 5012 TRIPHENYLTIN HYD 1432 VERNOLATE 5013 VINCLOZOLIN 5016 ZIRAM This dataset was developed using a geographic information system (GIS). The GIS software used to develop this dataset was Environmental Systems Research Institute's (ESRI) ArcInfo Workstation 9.1. Supplemental Information References: Fegeas, R.G., Claire, R.W., Guptill, S.C., Anderson, K.E., and Hallam, C.A., 1983, U.S. Geological Survey digital cartographic data standards-- Land use and land cover digital data: U.S. Geological Survey Circular 895-E, 21 p. Nakagaki, Naomi, and Wolock, D.M., 2005, Estimation of agricultural pesticide use in drainage basins using land cover maps and county pesticide data: U.S. Geological Survey Open-File Report 2005-1188, 46 p. National Oceanic and Atmospheric Administration, 1994, NOS80K/ALLUS80K Medium-resolution Digital Vector U.S. Shoreline, digital map, accessed October 2006, at http://woodshole.er.usgs.gov/pubs/of2005-1048/data/basemaps/usa/nos80k/nos80k.htm Price, C.P., Nakagaki, Naomi., Hitt, K.J., and Clawges, R.M., 2003, Mining GIRAS: Improving on a national treasure of land use data, in Proceedings of the 23rd ESRI International Users Conference, July 7-11, 2003: Redlands, Calif., 11 p. Available online at http://gis.esri.com/library/userconf/proc03/p0904.pdf. Price, C.P., Nakagaki, Naomi., Hitt, K.J., and Clawges, R.M., 2007, Enhanced historical land-use and land-cover data sets of the U.S. Geological Survey: U.S. Geological Survey Data Series 240, accessed April 9, 2007, at http://pubs.usgs.gov/ds/2006/240 Thelin, G.P., 2005, 1997 County pesticide use estimates for 220 compounds, data table, accessed October 1, 2006, at http://water.usgs.gov/lookup/getspatial?pesticide_use97 U.S. Department of Agriculture, 1999, 1997 Census of Agriculture, United States, Summary and state data, Volume 1: Geographic Area Series Part 51, AC97-A-51, National Agriculture Statistics Service. U.S. Department of Commerce, Bureau of Census, 1993, 1:100,000-scale Counties of the United States, digital map, accessed September 2005, at http://water.usgs.gov/lookup/getspatial?county100 U.S. Department of Commerce, 1995, 1992 Census of Agriculture, United States, Summary and state data, Volume 1: Geographic Area Series Part 51, U.S. Department of Commerce, Bureau of the Census. U.S. Geological Survey, 1990, Land use and land cover digital data from 1:250,000- and 1:100,000-scale maps: U.S. Geological Survey Data User Guide, no. 4, 25p. U.S. Geological Survey, 1998, Land use and land cover digital data from 1:250,000- and 1:100,000-scale maps, accessed June 16, 2005, at http://edc.usgs.gov/products/landcover/lulc.html U.S. Geological Survey, 1999, National land cover data, digital files, accessed June 16, 2005, at http://edc.usgs.gov/products/landcover/nlcd.html Vogelmann, J.E., Howard, S.M., Yang, L., Larson, C.R., Wylie, B.K., and Van Driel, N. 2001. Completion of the 1990s National Land Cover Data Set for the conterminous United States from Landsat Thematic Mapper data and ancillary data sources, Photogrammetric Engineering & Remote Sensing 67: 650-662 1995 1998 Although this dataset is referenced to 1997, it generally represents a composite of estimated pesticide use during the mid to late 1990s: the source for pesticide use Thelin's county pesticide use dataset (Thelin, 2005), which was derived from county cropland areas from the 1997 Census of Agriculture (U.S. Department of Agriculture, 1999) and state average annual usage of agricultural pesticides from 1995 to 1998 (Gianessi and Marcelli, 2000), using methods described in Thelin and Gianessi (2000); the spatial distribution of agricultural land is primarily based on U.S. Geological Survey's National Land Cover Dataset 1992 (U.S. Geological Survey, 1999), which is comprised of satellite imagery collected predominantly in the early 1990s (Vogelmann and others, 2001). Currentness_Reference References: Gianessi, L.P., and Marcelli, M.B., 2000, Pesticide use in U.S.crop production: 1997, National Summary Report, National Center for Food and Agricultural Policy, Washington D.C., 101 p. Thelin, G.P., and Gianessi, L.P., 2000, Method for estimating pesticide for county areas of the United States: U.S. Geological Survey Open-File Report 00-250, 67 p. Thelin, G.P., 2005, 1997 County pesticide use estimates for 220 compounds, data table, accessed October 1, 2006, at http://water.usgs.gov/lookup/getspatial?pesticide_use97 U.S. Department of Agriculture, 1999, 1997 Census of Agriculture, United States, Summary and state data, Volume 1: Geographic Area Series Part 51, AC97-A-51, National Agriculture Statistics Service. U.S. Geological Survey, 1999, National land cover data, digital files, accessed June 16, 2005, at http://edc.usgs.gov/products/landcover/nlcd.html Vogelmann, J.E., Howard, S.M., Yang, L., Larson, C.R., Wylie, B.K., and Van Driel, N. 2001. Completion of the 1990s National Land Cover Data Set for the conterminous United States from Landsat Thematic Mapper data and ancillary data sources, Photogrammetric Engineering & Remote Sensing 67: 650-662 Complete None planned -128.307859 -65.143387 51.857984 22.736598 none inlandWaters National Water Quality Assessment (NAWQA) Program agricultural pesticide herbicide insecticide fungicide fumigant growth regulator defoliant none Conterminous United States none There are several limitations to this dataset and it is important they be carefully considered for each particular application. 1. The differences between county areas classified as agricultural land in the NLCDe 92 and in the 1997 county file of agricultural pesticide use (Thelin, 2005) can cause errors in the estimated distribution of use. The areas of agricultural land in the 1997 county agricultural pesticide use are primarily based on the 1997 Census of Agriculture (surveys) while the areas of agricultural land in the NLCDe 92 are based primarily on satellite imagery. For counties in which there are temporal and/or areal discrepancies between the two primary data sources, the probability of error increases in the pesticide grids. Ideally, mapped agricultural land identified by the actual crop grown rather than groups of crops (such as row crops, small grains, etc.) would be used with the pesticide information by crop, and the information on pesticide application and mapped cropland would correlate within the same time period. 2. The limitations of each type of data used to estimate use are inherited by this dataset: a) Please refer to Thelin and Gianessi (2000) for more information related to limitations of the county pesticide use estimates. An example of a limitation of the county-level pesticide use data is the non-disclosure of some census crop information to protect the identity of individual farmers (U.S. Department of Agriculture, 1999). Another limitation to consider is that state pesticide-use coefficients (average annual application rates and percent of a crop's acres in a state treated with a pesticide) are gathered from surveys and reports covering a 4-year period (Gianessi and Marcelli, 2000). The total amount of pesticide use by county, as determined from the pesticide grids, closely match the total amount of county pesticide use from Thelin's (2005) source data (differences are all less than 1 kg). Therefore, the grids reflect the pesticide use from the source county data correctly at the county scale. However, when sub-county areas (such as small watersheds) are evaluated, the potential for error for estimating pesticide use increases. b) The National Land Cover dataset 1992 (NLCD 92) is subject to misclassification of agricultural land in some areas despite the extensive use of ancillary information (Vogelmann and others, 2001). (For additional information related to accuracy assessment of the NLCD, see http://landcover.usgs.gov/accuracy/index.php). The pesticide use grid dataset incorporates an enhanced version of the NLCD 92 which includes reclassification in some areas based on U.S. Geological Survey's Land Use and Land Cover dataset of the 1970 and 1980's. Classification of agricultural land is further complicated by farmland that is used for multiple crops, or left fallow. 3) The limitations described in #1 and #2 above can cause some grid-cell values (estimated pesticide application) to be unrealistically high. For example, in a few counties in Texas, where much of the pasture lands appear to be classified as "shrubland" in the NLCDe 92, tens of thousands of kilograms of 2,4-D are distributed amongst the very small areas classified as "pasture/hay". 4) Conversely, the limitations described in #1 and #2 can also cause underestimation of pesticide use in areas, where, for instance, grasslands are misclassified as "pasture/hay". In these counties, the amount of pesticides are distributed over areas that may truly be a mix of grasslands and pastured land, which result in a low pesticide use intensity. In addition, this "dilution" effect is introduced in the 875 counties in which the CGF areas served to substitute ORCH areas due to the absence of land classified as ORCH. 5) The grid cells in all 219 agricultural pesticide grids are zero in 42 counties and statistically equivalent entities ("county equivalents"). The majority of these 42 counties and county equivalents are individual cities that comprise mostly urban land use (and therefore indicative of little or no agricultural pesticide use). The amount of agricultural pesticide use in these 42 counties and county equivalents are not excluded in the pesticide grids but rather combined with its adjacent or surrounding county. The counties and county equivalents that have pesticide use combined with its adjacent or surrounding county are: COUNTY STATE NAME OF COUNTY CODE (FIPS) 11001 DC District of Columbia 24510 MD Baltimore City 29510 MO St. Louis City 30113 MT Yellowstone National Park 51510 VA Alexandria City 51515 VA Bedford City 51520 VA Bristol City 51530 VA Buena Vista City 51540 VA Charlottesville City 51560 VA Clifton Forge City 51570 VA Colonial Heights City 51580 VA Covington City 51590 VA Danville City 51595 VA Emporia City 51600 VA Fairfax City 51610 VA Falls Church City 51620 VA Franklin City 51630 VA Fredericksburg City 51640 VA Galax City 51650 VA Hampton City 51660 VA Harrisonburg City 51670 VA Hopewell City 51678 VA Lexington City 51680 VA Lynchburg City 51683 VA Manassas City 51685 VA Manassas Park City 51690 VA Martinsville City 51700 VA Newport News City 51710 VA Norfolk City 51720 VA Norton City 51730 VA Petersburg City 51735 VA Poquoson City 51740 VA Portsmouth City 51750 VA Radford City 52190 VA Richmond City 51770 VA Roanoke City 51775 VA Salem City 51780 VA South Boston 51790 VA Staunton City 51820 VA Waynesboro City 51830 VA Williamsburg City 51840 VA Winchester City The Agricultural Census county FIPS codes and associated (=) U.S. county FIPS codes in the conterminous U.S. are identified below (Marlene Diehl, U.S. Department of Agriculture, written commun., 2007): 24033 = 24033, 11001 24005 = 24005, 24510 29189 = 29189, 29510 51003 = 51003, 51540 51005 = 51005, 51560, 51580 51015 = 51015, 51790, 51820. 51019 = 51059, 51515. 51031 = 51031, 51680. 51041 = 51041, 51570. 51059 = 51059, 51510, 51600, 51610. 51069 = 51069, 51840. 51077 = 51077, 51640. 51081 = 51081, 51595. 51087 = 51087, 52190. 51089 = 51089, 51690. 51095 = 51095, 51830. 51121 = 51121, 51750. 51143 = 51143, 51590. 51149 = 51149, 51670, 51730. 51153 = 51153, 51683, 51685. 51161 = 51161, 51770, 51775. 51163 = 51163, 51530, 51678. 51165 = 51165, 51660. 51175 = 51175, 51620. 51177 = 51177, 51630. 51191 = 51191, 51520. 51195 = 51195, 51720. 51199 = 51199, 51650, 51700, 51735. 51550 = 51550, 51740. 51810 = 51810, 51710. Use_Constraints References: U.S. Geological Survey Ask USGS -- Water Webserver Team mailing

445 National Center

Reston VA 20192 USA 1-888-275-8747 (1-888-ASK-USGS) http://answers.usgs.gov/cgi-bin/gsanswers?pemail=h2oteam&subject=GIS+Dataset+agpest97grd http://water.usgs.gov/GIS/browse/agpest97.jpg Illustration of data set kg8008 jpg Gail P. Thelin originated the main source data, the 1997 county pesticide use estimates; Curtis V. Price, James A. Falcone, Kerie J. Hitt, and Barbara C. Ruddy contributed to the development of the enhanced land cover data; David M. Wolock assisted with developing the methods used to generate the pesticide grids; Jo Ann M. Gronberg, Michaela R. Johnson, Nancy T. Baker, and Doug A. Freehafer reviewed the metadata and portions of selective pesticide grids. Microsoft Windows XP Version 5.1 (Build 2600) Service Pack 2; ESRI ArcCatalog 9.1.0.722 The values of selective grid cells were verified for computational accuracy (steps 1,9,12, and 13 of the Process steps) by the author and reviewers. In addition, grid cell values were summed by county to ensure the county totals derived from the grids equate to the county totals from the source data. The insecticide, BT, is excluded from this dataset because 1997 county estimates are unavailable for BT (Thelin, 2005). Thelin, Gail P. 20051004 Version 2.0, October 06, 2005 tabular digital data http://water.usgs.gov/lookup/getspatial?pesticide_use97 online 1995 1998 publication date pesticide_use97 county pesticide use by crop (in pounds) U.S. Geological Survey 1999 raster digital data http://edc.usgs.gov/products/landcover/nlcd.html online 1992 publication date NLCD 92 land cover classifications, "row crops," "small grains," "fallow," "orchards/vineyards/other," "pasture/hay" Price, C.P., Nakagaki, Naomi., Hitt, K.J., and Clawges, R.M. 2007 raster digital data Data Series 240 Rapid City, South Dakota U.S. Geological Survey http://pubs.usgs.gov/ds/2006/240 online 1970 1985 publication date none land cover classification, "orchards, groves, vineyards, nurseries, and ornamental horticultural areas" U.S. Department of Commerce, Bureau of the Census 1993 vector digital data http://water.usgs.gov/lookup/getspatial?county100 online 1990 publication date county100 county boundary delineations and county code (FIPS) National Oceanic and Atmospheric Administration (NOAA), National Ocean Service (NOS), Office of Coast Survey, and the Strategic Environmental Assessments (SEA) Division of the Office of Ocean Resources Conservation and Assessment (ORCA) 1994 vector digital data http://woodshole.er.usgs.gov/pubs/of2005-1048/data/basemaps/usa/nos80k/nos80k.htm online 1988 1992 publication date allus80k shoreline delineations 1. The 1997 county pesticide by crop file was used to determine total pesticide use by county for each of the following 3 crop classes: a) row crops, small grain crops, and fallow land (CGF) b) pasture and hay crops (PAST) c) orchard and vineyard crops (ORCH) An INFO data table was created for each pesticide, consisting of the following fields: FIPS = numeric county code CGF_KG = total kilograms of the pesticide applied to crops in the "row crops/small grains/fallow land" crop class PAST_KG = total kilograms of the pesticide applied to crops in the "pasture/hay" crop class ORCH_KG = total kilograms of the pesticide applied to crops in the "orchard/vineyard" crop class This procedure was followed for all 219 pesticides in this dataset. 2. Using the 30-m resolution national grid of land cover ("NLCDe 92", as described in the Supplemental_Information section), a national 30-m resolution grid was generated for each of the 3 crop classes: a) The CGF (crop class) grid consisted of grid cells with a value of "0" or "1", "1" if the NLCDe 92 grid cells were classified as 82 (row crops), 83 (small grains) , or 84 (fallow) ; otherwise assigned "0". b) The ORCH (crop class) grid consisted of grid cells with a value of "0" or "1", "1" if the NLCDe 92 grid cells were classified as 61 (orchards/vineyards/other) or 62 (LULC orchards/vineyards/other); otherwise assigned "0". c) The PAST (crop class) grid consisted of grid cells with a value of "0" or "1", "1" if the NLCDe 92 grid cells were classified as 81 (pasture/hay); otherwise assigned "0". 3. The 3 grids generated in step 2 were overlaid with a 30-m resolution 1990 county boundaries grid to determine cell count totals by county and crop class. Results were stored in an INFO data table. 4. The output INFO files in step 1 and step 3 were merged to identify which counties would lose pesticide use loads in the pesticide grids due to the absence of land classified as agricultural land in the land cover map (NLCDe 92 grid). This step was carried out for each of the 3 crop classes and 219 of the 220 compounds in Thelin's (2005) county pesticide data file were evaluated (BT was excluded). 5. The 30-m national grids of each crop class (output of step 2) were adjusted to accommodate the counties identified in step 4. The table of cell count totals by county and crop class (output of step 3) was used to determine which crop class would serve as a substitute in counties that were lacking agricultural land (pertaining to a crop class). Here is a summary of the decisions that were made based on this evaluation: Crop # of counties Substituted land classes Class adjusted ORCH 875 "row crops," "small grains," "fallow" ORCH 6 "pasture/hay" CGF 8 "pasture/hay" PAST 2 none qualified so entire county area was used instead The 30-m national grids for each crop class were adjusted accordingly. The results were three 30-m resolution grids that represented the distribution of its original crop class mixed with another crop class. The sole purpose of these grids were to apply the pesticide use by crop class to its respective, adjusted crop class grid. To create the adjusted crop class grids, the following steps were followed 4 times, once for each set of conditions in the table shown above. a. Using the 30-m resolution county boundaries grid, select the counties in which the adjustment process would take place and make a new grid from this selection by coding all cells within these selected counties equal to "1". (This new grid will be referred to as "selcntyg".) So for the 875 counties that were missing ORCH areas, all 30-m cells within the boundaries of the 875 counties of interest were coded as "1" in the "selcntyg" grid; all other cells were coded "0". b. Using "selcntyg" to identify where geoprocessing would take place, create another temporary grid to flag the substituted crop class cells within the counties of interest. (This new grid will be referred to as "subcropclassg".) The flagged cells were coded "1" and all other cells were coded "0". In the working example, all 30-m cells classified as "row crops," "small grains," or "fallow" in the CGF grid which are located within the boundaries of the 875 counties of interest (coded as "1" in selcntyg") were coded as 1 in "subcropclassg". c. Sum the original crop class grid with "subcropclassg" to create the final adjusted crop class grid. In the working example, the original 30-m resolution CGF grid (created in step 2) was summed with the "subcropclassg" created in the previous step 5b. In regard to the 2 counties missing mapped pasture/hay lands, because these counties had neither mapped orchard and vineyard areas nor mapped row crops, small grains, or fallow land, the amount of pesticide used in these 2 counties were evenly distributed throughout the county. 6. The adjusted 30-m national grids of CGF, PAST, and ORCH (output of step 5c) were re-gridded to 1-km resolution grids, 1 for each crop class: (grid) CGF_PCT, (grid) PAST_PCT, and (grid) ORCH_PCT. Each 1-km grid cell represented the percentage of the 1-by-1 km area that consisted of the agricultural land pertaining to the crop class. The main steps to generate the 1-km percentage crop class grids from the 30-m crop class grids are described below: a. Resample the 30-m resolution grid to the 40-m resolution, determine the total number 40-m grid cells coded "1" (numeric code for "row crops") within each 1-by-1 km area, and store this total in the cell value of an interim 1-km resolution grid. The resampling process was necessary as in interim step prior to the creation of the 1-km resolution grid to ensure multiple undivided grid cells would fit into each 1-by-1 km area. Though 25-m resolution grid cells also fit undivided into 1-km grid cells, resampling at the 25-m resolution would create a raster dataset that contained more detail than that captured in the original data. b. Resample the 30-m resolution grid to the 40-m resolution, but this time, determine the total number of 40-m grid cells within each 1- by 1- km area that have data (or are valid). In most cases, the total number of valid 40-m cells within a 1-km grid cell was 625, but along the shoreline the divisor was not always 625 because some of the grid cells representing the oceans did not have cell values (were coded null). c. Generate the final 1-km resolution percentage output grid for the crop class, by dividing the grid produced in step "a" by the grid produced in step "b". The percentages were stored in the grid cells as floating numbers. 7. The 30-m 1990 county boundaries grid (Jo Ann M. Gronberg, U.S. Geological Survey, written commun., 2005) was re-gridded to a resolution of 1 kilometer, preserving the unique numeric county code, the Federal Information Processing Standards code, or "FIPS" code, in the 1-km resolution grid. The resampling technique that was used was the nearest-neighbor assignment. 8. The county area of land associated with each crop class was determined from the overlay of the 1-km grids of crop classes (generated in step 6) with the 1-km grid of county boundaries (see step 7). The results were stored in an INFO table. A single INFO data table was created consisting of the following fields: FIPS = numeric county code CGF_SQKM = total square kilometers of the land classified as "row crops", "small grains", or "fallow" in the county PAST_SQKM = total square kilometers of the land classified as "pasture/hay" in the county ORCH_SQKM = total square kilometers of the land classified as "orchards/vineyards/other" or "LULC orchards/vineyards/other in the county **From this point on, the procedures are conducted multiple times, once for each of the 219 pesticides in this dataset.** 9. The rate of pesticide application by county for each of the 3 crop classes was calculated for each pesticide. The rate was computed by dividing the county mass of pesticide applied to crops associated with the crop class (computed in step 1), by the county area of land associated with the crop class (computed in step 8). An INFO table was created for each pesticide consisting of FIPS code and application rate by crop class- the new rate fields were called CGF_RATE, PAST_RATE, and ORCH_RATE: A single INFO data table was created consisting of the following fields: FIPS (numeric county code) CGF_RATE (county application rate to CGF, = CGF_KG/CGF_SQKM) PAST_RATE (county application rate to PAST, = PAST_KG/PAST_SQKM) ORCH_RATE (county application rate to ORCH, = ORCH_KG/ORCH_SQKM) 10. The INFO data table containing the county rate of pesticide application (output of step 9) was merged with the value attribute table (also in INFO format) of the 1-km resolution grid of the 1990 county boundaries (output of step 7) by linking the common key field, the unique numeric code for a county (FIPS). 11. Interim 1-km resolution grids of the county rate of pesticide application for each crop class were created: (grid)CGF_RATE, (grid)PAST_RATE, (grid)ORCH_RATE. The grid cell values in a single county have the same county application rate values throughout the county. So for instance, all CGF grid cells in county A were assigned the CGF_RATE calculated for county A (in step 9). 12. The 1-km resolution grids of pesticide use by crop class were created by multiplying the grid of county rate of pesticide application by crop class (created in step 11) by the grid of percentage land pertaining to the respective crop class (created in step 6) divided by 100. (grid)CGF_KG = [(grid)CGF_RATE * (grid)CGF_PCT/100] (grid)PAST_KG = [(grid)PAST_RATE * (grid)PAST_PCT/100] (grid)ORCH_KG = [(grid)ORCH_RATE * (grid)ORCH_PCT/100] 13. The final 1-km resolution grid of total pesticide application for all crops was created by summing the 3 individual pesticide use grids by crop class created in step 12. (grid)KGxxxx = (grid)CGF_KG + (grid)PAST_KG + (grid)ORCH_KG where xxxx = numeric code for the pesticide (see "Supplemental_Information" for list of pesticide code and names 2006 - 2007 Raster Grid Cell 2940 4645 1 Albers Conical Equal Area 29.5 45.5 -96.0 23.0 0.00000 0.00000 row and column 1000.0 1000.0 Meters North American Datum of 1983 Geodectic Reference System 80 6378137.0 298.257222 The only attribute table available for each grid is the "statistics" file, which contain the following statistics for all the cells in the grid: KGxxxx.STA: (where xxxx represents the numeric code for a pesticide): COLUMN ITEM NAME WIDTH OUTPUT TYPE N.DEC ALTERNATE NAME 1 MIN 8 15 F 3 9 MAX 8 15 F 3 17 MEAN 8 15 F 3 25 STDV 8 15 F 3 where MIN = minimum value of grid cells MAX = maximum value of grid cells MEAN = mean of grid cell values STDV = standard deviation of grid cell values These statistics are not an accurate representation of the pesticide use cell values because the zero cell values in non-agricultural areas skew these numbers. Please also note: - The pesticide use grids have cell values stored as floating numbers, and in the ArcInfo grid model, floating grids do not have a value attribute table. - The values in each grid cell represent the estimated kilograms of the pesticide used in the 1-by-1 km area. - There is no distinction between a grid cell with zero pesticide use on agricultural land and a grid cell that has no agricultural land (and thus no pesticide use). - Some data values may appear to be zero if the display width of the number is not carried out far enough. For instance, the estimated amount of total abamectin in Alachua county in Florida is 0.24 kg, and this amount is apportioned out to the 1-km grid cells that have agricultural land within Alachua county. Therefore, the resulting data values in the abamectin grid in this county are extremely small. - The grid cells outside the U.S. boundary and shoreline along the oceans and major rivers draining to the oceans, have a "no data" cell value in all the pesticide grids. - In the ASCII text file, "no data" is indicated as -9999. None U.S. Geological Survey Ask USGS - Water Webserver Team mailing

445 National Center

Reston VA 20192 1-888-275-8747 (1-888-ASK-USGS) http://water.usgs.gov/user_feedback_form.html Although these data have been used by the U.S. Geological Survey, U.S. Department of the Interior, no warranty expressed or implied is made by the U.S. Geological Survey as to the accuracy of the data and related materials. The act of distribution shall not constitute any such warranty, and no responsibility is assumed by the U.S. Geological Survey in the use of this data, software, or related materials. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government. Data Catalog Web page to download pesticide grid 31 KB http://water.usgs.gov/GIS/dsdl/agpest97grd/index97.html None. This dataset is provided by USGS as a public service. 200708 U.S. Geological Survey Ask USGS -- Water Webserver Team mailing

445 National Center

Reston VA 20192 1-888-275-8747 (1-888-ASK-USGS) http://answers.usgs.gov/cgi-bin/gsanswers?pemail=h2oteam&subject=GIS+Dataset+agpest97grd FGDC Content Standards for Digital Geospatial Metadata FGDC-STD-001-1998