Field Boundary Tools data guide

Version 1.0.0

Only two datasets are required: a Census county boundary and USDA Crop Sequence Boundaries. All other datasets are optional and are skipped when unavailable. Extract downloaded ZIP files before selecting the source-data folder in QGIS. The USDA 2025 Cultivated Layer is strongly recommended. It removes predominantly non-cultivated polygons and identifies cultivated CSB gaps where CLU or FTW coverage is missing. The raster scores vector polygons; its 30-meter pixel edges are not used as final field boundaries.

How to run the plugin

  1. Place the inputs in one folder or its subfolders.
  2. Open Processing > Toolbox in QGIS.
  3. Open Field Boundaries > Create county field layer.
  4. Select the state, county, and data folder.
  5. Review the distances, cultivated threshold, and minimum field area.
  6. Select Run and review the log.
  7. Open the returned review link to compare source and final layers.
  8. Open the desired final GeoPackage for subsequent QGIS work.
  9. Use input_inventory.json and run_summary.txt to verify the run.

Exact downloads

RoleDownloadExact local input
County boundary - required 2025 TIGER/Line County and Equivalent nationwide ZIP Extract the ZIP. Keep tl_2025_us_county.shp together with its .dbf, .shx, and .prj files. The tool uses GEOID to select the county chosen in the dialog.
Crop Sequence Boundaries - required USDA NASS NationalCSB 2018-2025 rev23 Extract the ZIP and keep the complete NationalCSB_2018-2025_rev23.gdb directory. USDA publishes this as a large CONUS archive, not as a county file. The tool extracts and caches only the selected county before geometry repair.
USDA Cultivated Layer - recommended Download the 2025 National Cultivated Layer ZIP (about 300 MB)
USDA NASS national download page
2025 metadata and class definitions 		Extract the ZIP and keep 2025_cultivated_layer.tif in the selected folder or a subfolder. The tool recognizes GeoTIFF names containing cultivated. Class 2 is Cultivated, class 1 is Non-Cultivated, and 0 is background. By default a vector polygon must be at least 50% class 2. The layer is based on the 2021-2025 CDLs and has 30-meter cells.
2025 and 2024 Cropland Data Layers - field attributes 2025 National 10-meter CDL ZIP (about 9.8 GB)
2024 National 10-meter CDL ZIP (about 9.0 GB)
USDA NASS national CDL download page 		Extract both ZIPs and keep 2025_10m_cdls.tif and 2024_10m_cdls.tif in the selected folder or subfolders. The tool discovers them automatically; no separate raster selection is needed. The tool reads the official raster attribute table when present and otherwise uses its built-in USDA CDL class-code table. This means the crop attributes still work if only the .tif files were copied and the download's external .aux sidecar was not. It stores the top three named crop classes for each final field, and reports each crop's percent of all non-background classified cells in that field. Non-crop land-cover and fallow/idle classes are not ranked as crops. The official 30-meter resampled downloads are also recognized if disk space or speed is more important than 10-meter detail.
LGRIP30 L3 V002 irrigation - recommended
Create a free NASA Earthdata Login Official LGRIP30 L3 V002 catalog Official Version 2 user guide
LP DAAC distributes the CONUS product as 17 overlapping 10-degree GeoTIFF tiles, not one nationwide TIFF. Sign into Earthdata once, choose a state, and use every tile link shown. Keep the original LGRIP30_L3_...tif filenames together in the selected data folder. The tool automatically discovers and mosaics the tile or tiles overlapping the chosen county. L3 classes are 0 water, 1 non-cropland, 2 irrigated cropland, and 3 rainfed cropland. The output irr_pct is class-2 cells divided by all valid class 0-3 cells inside the field. The older local CONUS Irrigation Data Layer.tif remains supported as a binary fallback, but LGRIP30 L3 is preferred when both are present.
TIGER roads
Loading county roads... Browse the 2025 Census All Roads archive
Download and extract tl_2025_<five-digit-county-FIPS>_roads.zip. This is the All Roads product. Do not choose the statewide prisecroads product unless omitting local roads is intentional.
Fields of the World
Download Indiana FTW GeoParquet FTW product information
The green button downloads US_<state-abbreviation>.parquet directly. Keep the original filename in the source-data folder; QGIS reads it without extraction or conversion. Source Cooperative currently provides all 50 states, but not D.C. or Puerto Rico. FTW is used automatically only after higher-priority CLU coverage.
Indiana metadata note (July 20, 2026): the current upstream US_IN.parquet contains Indiana coordinates but omits its CRS and advertises a stale Calgary-area dataset extent. Version 0.5.4 assigns EPSG:4326 and scans actual feature geometry for the county, then caches the corrected subset.
Hydrography
Download Indiana NHD GeoPackage ZIP USGS hydrography product information
The green button starts the official state-level NHD GeoPackage download directly. Extract the ZIP and keep NHD_H_<State>_State_GPKG.gpkg in the source-data folder; the tool finds its NHDFlowline layer. These archives can be very large. Indiana is approximately 1.75 GB. Enter the flowline exclusion distance in feet; the tool converts it to meters internally.
Building footprints Microsoft state building-footprint downloads Download and extract the selected-state GeoJSON ZIP. Version 0.5.4 recognizes both building_footprints.geojson and the original state filename, such as Indiana.geojson. Enter the building exclusion distance in feet; the tool converts it to meters internally.
Wind turbine bases - recommended where present Download nationwide USWTDB GeoJSON ZIP Nationwide USWTDB Shapefile ZIP
Official USWTDB data and metadata page 		Download the complete national dataset once and extract it into the selected data folder. Names containing uswtdb or wind turbine are discovered automatically and clipped to the selected county. By default the tool removes a 60-foot-diameter circle (30-foot radius) centered on each mapped turbine base and provides the points as an HTML reference overlay. USWTDB does not contain access-drive geometry. When both USWTDB and the county TIGER All Roads layer are available, version 0.8.0 queries OpenStreetMap for county service/track ways, traces only routes that begin near a turbine and connect to the TIGER network, and carves them at a default 24-foot total width. The raw Overpass response is cached for 30 days under .field_boundary_cache, so repeat county tests do not download it again. High-confidence routes are preserved separately as turbine_driveway_candidates.gpkg and as a yellow dashed HTML overlay. If the network or Overpass is unavailable, the run continues with turbine-base carving only. Review candidates against imagery: OpenStreetMap coverage and tags are community-maintained.
Industrial solar - USPVDB Download nationwide USPVDB Shapefile ZIP Nationwide GeoJSON ZIP
Official USPVDB downloads, metadata, and web services 		Download the current nationwide polygon database and extract the complete Shapefile set, including .shp, .dbf, .shx, and .prj. Version 4.0 is named similar to uspvdb_v4_0_20260414.shp. The database covers ground-mounted photovoltaic facilities of at least 1 MW and maps array boundaries. The tool clips it to the county, adds an Industrial Solar HTML checkbox, and writes solar_pct as USPVDB overlap area divided by final-field area. It does not carve solar automatically.
PAD-US lands
Loading PAD-US catalog... USGS PAD-US product information
Download the selected-state geodatabase. Use the PADUS4_1Fee polygon feature class. Exporting that layer as padus_fee_managers.gpkg is the most predictable input. The green button opens the exact PADUS4_1_State_<state>_GDB_KMZ.zip file in the official USGS ScienceBase download service. Extract the ZIP before use.
Parcels County or state government GIS/open-data portal Use a current parcel polygon layer named tax_parcels.gpkg, or include parcel or cadastral in the file or layer name. There is no single national public parcel download. Parcels are displayed in the HTML comparison but never split final field geometry because property lines may divide one managed field.
Common Land Units USDA CLU access explanation Use only an authorized CLU polygon dataset, named common_land_units.gpkg. USDA/FSA does not make CLU geometry publicly downloadable.

HTML boundary review output

Every successful run creates field_boundary_review.html and a companion field_boundary_review_data folder. Keep them together. Both corner selectors can display hydrography-carved fields, fields without hydrography carving, the Cultivated Layer, 2025 CDL, 2024 CDL, irrigation, CLU, FTW, Raw CSB, or Parcels. Independent checkboxes toggle NHD hydrography, TIGER roads, USWTDB wind turbines, USPVDB industrial solar, and detected turbine driveways; missing optional layers are disabled. The upper-right information panel is collapsible, and the basemap selector offers Hybrid, Satellite, and Streets. Its boundary vectors open directly from disk; Leaflet and Esri Hybrid imagery require an internet connection. Use the returned REVIEW link to open it after processing.

Fast repeat testing

Keep Reuse cached county input layers enabled. Prepared county GeoPackages and a discovery index are stored under .field_boundary_cache. Unchanged repeat runs avoid reopening statewide and national data. Cultivated coverage scores are cached with the county vector subsets and are recalculated only when the source raster, source vector, or threshold changes. CLU is the recommended default boundary source; All recognized optional datasets are used automatically. Parcels remain a comparison layer and never split fields. County caching is always enabled. Results are not loaded automatically; use the returned OUTPUT link when ready.

Clinton County, Indiana example

Find other five-digit county codes in the Census national county-code list.

Recognized containers and names

The tool scans GeoTIFF cultivated rasters plus GeoPackage, GeoParquet, Shapefile, GeoJSON, FlatGeobuf, SQLite, KML, and Esri File Geodatabase vector inputs. GeoParquet requires a QGIS/GDAL build with the Parquet driver.

Helpful names include county_boundary, crop_sequence_boundaries, ftw, common_land_units, tax_parcels, tiger_roads, nhd_flowline, building_footprints, and padus. GeoPackages and geodatabases may contain multiple layers; layer names are also inspected.

Generated boundaries require review against current imagery and local knowledge. They are agronomic planning features, not legal surveys or property-boundary products.