The Cropland Data Layer (CDL) is a crop-specific land cover data layer created annually for the continental United States using moderate resolution satellite imagery and extensive agricultural ground truth. The CDL is created by the USDA, National Agricultural Statistics Service (NASS), Research and Development Division, Geospatial Information Branch, Spatial Analysis Research Section. For detailed FAQ please visit CropScape and Cropland Data Layers - FAQs. To explore details about the classification accuracies and utility of the data, see state-level omission and commission errors by crop type and year. The asset date is aligned with the calendar year of harvest. For most crops the planted and harvest year are the same. Some exceptions: winter wheat is unique, as it is planted in the prior year. A hay crop like alfalfa could have been planted years prior. For winter wheat the data also have a class called "Double Crop Winter Wheat/Soybeans". Some mid-latitude areas of the US have conditions such that a second crop (usually soybeans) can be planted immediately after the harvest of winter wheat and itself still be harvested within the same year. So for mapping winter wheat areas use both classes (use both values 24 and 26). While the CDL date is aligned with year of harvest, the map itself is more representative of what was planted. In other words, a small percentage of fields on a given year will not be harvested. Some non-agricultural categories are duplicate due to two very different epochs in methodology. The non-ag codes 63-65 and 81-88 are holdovers from the older methodology and will only appear in CDLs from 2007 and earlier. The non-ag codes from 111-195 are from the current methodology which uses the USGS NLCD as non-ag training and will only appear in CDLs 2007 and newer. 2007 was a transition year so there may be both sets of categories in the 2007 national product but will not appear within the same state. Note: The 2024 CDL only has the data band. The cultivated and confidence bands are yet to be released by the provider.

The Cropland Data Layer (CDL), hosted on CropScape, provides a raster, geo-referenced, crop-specific land cover map for the continental United States. The CDL also includes a crop mask layer and planting frequency layers, as well as boundary, water and road layers. The Boundary Layer options provided are County, Agricultural Statistics Districts (ASD), State, and Region. The data is created annually using moderate resolution satellite imagery and extensive agricultural ground truth. Users can select a geographic area of interest or import one, then access acreage statistics for a specific year or view the change from one year to another. The data can be exported or added to the CDL. The information is useful for issues related to agricultural sustainability, biodiversity, and land cover monitoring, especially due to extreme weather events. Resources in this dataset:Resource Title: CropScape and Cropland Data Layer - National Download. File Name: Web Page, url: https://www.nass.usda.gov/Research_and_Science/Cropland/Release/index.php Downloads available as zipped files at https://www.nass.usda.gov/Research_and_Science/Cropland/Release/index.php --

National CDL's -- by year, 2008-2020. Cropland Data Layer provides a raster, geo-referenced, crop-specific land cover map for the continental United States. The CDL also includes a crop mask layer and planting frequency layers, as well as boundary, water and road layers. The Boundary Layer options provided are County, Agricultural Statistics Districts (ASD), State, and Region. National Cultivated Layer -- based on the most recent five years (2013-2020). National Frequency Layer -- the 2017 Crop Frequency Layer identifies crop specific planting frequency and are based on land cover information derived from the 2008 through 2020CDL's. There are currently four individual crop frequency data layers that represent four major crops: corn, cotton, soybeans, and wheat. National Confidence Layer -- the Confidence Layer spatially represents the predicted confidence that is associated with that output pixel, based upon the rule(s) that were used to classify it. Western/Eastern/Central U.S.

Visit https://nassgeodata.gmu.edu/CropScape/ for the interactive map including tutorials and basic instructions. These options include a "Demo Video", "Help", "Developer Guide", and "FAQ".

The USDA National Agricultural Statistics Service (NASS) Cropland Data Layer (CDL) is an annual raster, geo-referenced, crop-specific land cover data layer produced using satellite imagery and extensive agricultural ground reference data. The program began in 1997 with limited coverage and in 2008 forward expanded coverage to the entire Continental United States. Please note that no farmer reported data are derivable from the Cropland Data Layer.

Dataset Card for Multi-Temporal Crop Classification

  Dataset Summary

This dataset contains temporal Harmonized Landsat-Sentinel imagery of diverse land cover and crop type classes across the Contiguous United States for the year 2022. The target labels are derived from USDA's Crop Data Layer (CDL). It's primary purpose is for training segmentation geospatial machine learning models.

  Dataset Structure





  TIFF Files

Each tiff file covers a 224 x 224 pixel… See the full description on the dataset page: https://huggingface.co/datasets/ibm-nasa-geospatial/multi-temporal-crop-classification.

La couche de données sur les terres agricoles (CDL) est une couche de données sur la couverture des sols spécifique aux cultures créée chaque année pour les États-Unis continentaux à l'aide d'images satellite à résolution modérée et de données de terrain agricoles étendues. Le CDL est créé par le ministère de l'Agriculture des États-Unis, le Service national des statistiques agricoles (NASS), la division de recherche et développement, la branche d'information géospatiale, l'analyse spatiale, etc.

This collection contains the trained model (.h5) and the training and testing data (.csv or .npy) for the near real-time crop type mapper for Conterminous United States (CONUS) using harmonized Landsat and Sentinel-2 (HLS) dataset with codes on https://github.com/hankui/Real-time-crop-type-mapper

and for the paper

Zhang, H. K., Shen, Y., Zhang, X., Che, X., Yang, Z., et al. (2025), A near real-time crop type mapper for the conterminous United States, In review.

1, The trained model file

v1_70.layer4.METHOD2.BATCH64.LR0.0002.EPOCH30.L20.1.i0.model.h5

The structure of the model please refer to Zhang et al. (2025).

2, Training_and_evaluation.zip file contains training data for generating the above model and evaluation to produce the paper results. The training and testing (evaluation) samples were split as in Zhang et al. (2025) and from different pixel locations.

The training or testing input x (i.e., the HLS reflectance) is stored as a 3D matrix with dimensions n× (176+176+176+176) ×13.

• The first dimension, n, represents the number of training or testing samples.
• The second dimension comprises four segments of 176 values each:
• The first 176 represents the first-year Landsat data, with a maximum of 176 dates.
• The second 176 represents the first-year Sentinel-2 data, with a maximum of 176 dates.
• The third 176 represents the second-year Landsat data, with a maximum of 176 dates.
• The last 176 represents the second-year Sentinel-2 data, with a maximum of 176 dates.
• Time series with fewer than 176 observations were padded with -9999.
• The third dimension corresponds to spectral information, including year, normalized day of year (DOY), and normalized reflectance. Although the year is not used in training or testing, it is included to identify the sample's time. The reflectance bands are ordered as follows: four visible bands, one near-infrared (NIR) band, two shortwave infrared (SWIR) bands, three red-edge bands (Sentinel-2 only), and one broad NIR band (Sentinel-2 only). For Landsat data, the last four bands are filled with -9999. The mean and std normlization file is included in https://github.com/hankui/Real-time-crop-type-mapper

The training or testing output y includes 50 classes (Table 2 in Zhang et al., 2025), with values ranging from 0 to 49. The array in the file ‘inverse_mapping.npy’ can map these values back to the original label values as defined in the CDL keys at https://support.regrid.com/parcel-data/cdl-keys

The training and evaluation files were generated by processing AlignedCONUS_scale60_all_tiles_v1_6.csv and AlignedCONUS_scale60_all_tiles_v1_4.subcol_add_cdls.csv through several steps: combining data into two years, filtering out non-homogeneous pixels, excluding 2013 and 2014 data, applying normalization, discarding records with fewer than four HLS observations over two years, and redefining the labels. The resulting dataset includes 50 classes, comprising 37 crop classes and 13 non-crop classes (Tables 1 and 2 in Zhang et al., 2025).

3, AlignedCONUS_scale60_all_tiles_v1_6.csv contains the original HLS data with day of year, surface reflectance and quality assessment layer. The data was obtained by sampling every 60th 30-meter pixel across 96 systematically distributed tiles covering the CONUS (Fig. 1 in Zhang et al., 2025).

The data was derived for the period 2013 to 2023; however, only the data from 2015 to 2023 was utilized in Zhang et al. (2025) to cover Sentinel-2 data. Only cloud-free observations were stored. Cloud-free observations were defined for those not labelled as snow/ice, cloud, cloud shadow, or adjacent to cloud/shadow in the HLS quality assessment layer.

Each record contained data corresponding to a single pixel location for a specific year.

There are 4233 columns, with 9 columns storing the pixel specific and year information ('tile', 'col', 'row', 'lat', 'lon', 'year', 'total_n', 'tile_id', 'lid').

It contains 176×11 rows for the Landsat time series in a given year, where 176 represents the maximum number of cloud-free observations in a year, and 9 corresponds to the nine Landsat 8/9 bands (day of year, QA, the seven solar reflective bands: four visible, one near-infrared (NIR), and two shortwave infrared (SWIR) bands, and the two thermal bands). Note the two thermal bands are not used in Zhang et al., (2025). If a record contained fewer than 176 observations, missing values were filled with -9999.

It contains 176×13 rows for Sentinel-2 time series in a given year, where 176 represents the maximum number of cloud-free observations in a year, and 13 corresponds to the 13 Sentinel-2 bands (day of year, QA, and the 11 solar reflective bands: four visible, two NIR, three red-edge and two SWIR bands). If a record contained fewer than 176 observations, missing values were filled with -9999.

4, AlignedCONUS_scale60_all_tiles_v1_4.subcol_add_cdls.csv contains the original cdl labels.

It has 10 columns ['tile', 'col', 'row', 'lat', 'lon', 'year', 'tile_id', 'lid', 'cdl', 'cdl_homo'], with the first eight columns specific the pixel-specific and year information that can be linked to the AlignedCONUS_scale60_all_tiles_v1_6.csv file.

The variable ‘cdl’ represents the CDL label; refer to this link https://support.regrid.com/parcel-data/cdl-keys for label definitions. A value of 0 or NaN may indicate that the pixel is located in the ocean or outside the United States.

The ‘cdl_homo’ column indicates whether the label is consistent with all eight neighboring pixels (1 for consistent, 0 for not consistent).

Feature Classes are loaded onto tablet PCs and Field crews are sent to label the crop or land cover type and irrigation method for a subset of select fields or polygons. Each tablet PC is attached to a GPS unit for real-time tracking to continuously update the field crew’s location during the field labeling process.Digitizing is done as Geodatabase feature classes using ArcMap 10.X with NAIP or Google imagery as a background with other layers added for reference. Updates to existing field boundaries of individual agricultural fields, urban areas and more are precisely digitized. Changes in irrigation type and land use are noted during this process.Cropland Data Layer (CDL) rasters from the United States Department of Agriculture (USDA) National Agricultural Statistics Service (NASS) are downloaded for the appropriate year. https://nassgeodata.gmu.edu/CropScape/Zonal Statistics geoprocessing tools are used to attribute the polygons with updated crop types from the CDL. The data is then run through several stages of comparison to historical inventories and quality checking in order to determine and produce the final attributes.2018 marked the first year a comparison could be made using the CDL methodology. The comparison between 2017 and 2018 showed a large change in agricultural land use to other land use. It was determined this shift was due to crop land being allowed to sit fallow for a season and did not represent a shift away from agricultural land. The following code amended the data:***************************************************************************************************************************************####On 02/07/2020 this dataset was amended with the following R script to better reflect agricultural land changes:require(arcgisbinding)arc.check_product()####Bring in layersLU18<-arc.open("Path to data")LU18<-arc.select(LU18)#####Amend dataLU18$Landuse[LU18$Class_Name=='Fallow/Idle Cropland' & LU18$Description== 'Dry Land/Other']<-"Agricultural"LU18$CropGroup[LU18$Class_Name=='Fallow/Idle Cropland' & LU18$Description== 'Dry Land/Other']<-"Fallow/Idle"LU18$IRR_Method[LU18$Class_Name=='Fallow/Idle Cropland' & LU18$Description== 'Dry Land/Other']<-"Dry Crop"arc.write("Path to data", LU18)***************************************************************************************************************************************LUID -Unique ID number for each polygon in the final dataset, matches object.Landuse - Land use type, similar to land cover and represents our own categories of how the land is used.CropGroup - Groupings of broader crop categories to allow easy access to or query of all orchard or grain types etc.Description - Attribute that describes/indicates the various crop types and land use types determined by the GIS process.IRR_Method - Crop Irrigation Methods.Acres - Calculated acreage of the polygon.State - Spatial intersection identifying the State where the polygons are found.County - Spatial intersection identifying the County where the polygons are found.Basin - Spatial intersection identifying the Basin where the polygons are found. Basins, or Utah Hydrologic Basins are large watersheds created by DWRe.SubArea - Spatial intersection identifying the Subarea where the polygons are found. Subareas are subdivisions of the larger hydrologic basins created by DWRe.Label_Class - Combination of Label and Class_Name fields created during processing that indicates specific cover and use types.LABEL - Old shorthand descriptive label for each crop and irrigation type or land use type.Class_Name - Zonal Statistics majority value derived from the USDA CDL Cropscape raster layer, may differ from final crop determination.OldLanduse - This is the old short code found under landuse in past datasets and is kept to maintain connectivity with historical data.LU_Group - These codes represent some in-house groupings that are useful for symbology and other summarizing.SURV_YEAR - Indicates which year/growing season the data represents. Is useful when comparing to past layers.