Field-verified irrigated lands data were collected for the Remote Sensing Component of the St. Mary-Milk Rivers Consumptive Use study to aid in the identification and delineation of agricultural fields that are irrigated in 2021 and 2022 in the Milk River basin. This field verification of irrigated fields will provide data that will be used to ground truth evapotranspiration estimates obtained in the Milk River basin using remote sensing methods. This work is part of a larger project aimed at developing a historical database representing monthly actual evapotranspiration (ETa) totals in the Milk River basin from 1985-present using remote sensing. This database will lay the foundation for the establishment of a remote sensing tool with which to objectively estimate ETa in the upper Milk River basin on a monthly interval using high spatial resolution (100 meter) satellite imagery.

In 2015, agricultural irrigation withdrawals accounted for about 42 percent of the total freshwater withdrawals in the United States (Dieter and others, 2018). Consistent and accurate designations of irrigated agricultural lands, irrigation system type, conveyance systems, and water source (groundwater or surface water) are essential for the determination of irrigation water use and ultimately the sound management of our nation’s water resources. Several local, state, and federal agencies compile data (crops, irrigation, irrigation system type, etc.) that can be used to estimate irrigation withdrawals for agricultural. The format of these data varies from data tables, typically compiled at the county level, to spatial Geographic Information System (GIS) polygon layers of agricultural lands. These data sources are often incomplete, out of date, or inconsistently compiled. The USGS and the University of Wisconsin-Madison developed annual Landsat-based Irrigation Dataset (LANID), which consists of irrigation maps, derivative products, and manually collected ground reference data covering the conterminous US (CONUS) for the period of 1997–2017 (Xie and Lark, 2021a). These maps were developed using verified irrigated-lands GIS datasets (i.e. training data) coupled with remotely-sensed, 30-meter resolution Landsat-derived data. The current and future availability of verified field-level data is required to train and validate this and other models.

In 2015, agricultural irrigation withdrawals accounted for about 42 percent of the total freshwater withdrawals in the United States (Dieter and others, 2018). Consistent and accurate designations of irrigated agricultural lands, irrigation system type, conveyance systems, and water source (groundwater or surface water) are essential for the determination of irrigation water use and ultimately the sound management of our nation’s water resources. Several local, state, and federal agencies compile data (crops, irrigation, irrigation system type, etc.) that can be used to estimate irrigation withdrawals for agricultural. The format of these data varies from data tables, typically compiled at the county level, to spatial Geographic Information System (GIS) polygon layers of agricultural lands. These data sources are often incomplete, out of date, or inconsistently compiled. The USGS and the University of Wisconsin-Madison developed annual Landsat-based Irrigation Dataset (LANID), which consists of irrigation maps, derivative products, and manually collected ground reference data covering the conterminous US (CONUS) for the period of 1997–2017 (Xie and Lark, 2021a). These maps were developed using verified irrigated-lands GIS datasets (i.e. training data) coupled with remotely-sensed, 30-meter resolution Landsat-derived data. The current and future availability of verified field-level data is required to train and validate this and other models.

In 2015, agricultural irrigation withdrawals accounted for about 42 percent of the total freshwater withdrawals in the United States (Dieter and others, 2018). Consistent and accurate designations of irrigated agricultural lands, irrigation system type, conveyance systems, and water source (groundwater or surface water) are essential for the determination of irrigation water use and ultimately the sound management of our nation’s water resources. Several local, state, and federal agencies compile data (crops, irrigation, irrigation system type, etc.) that can be used to estimate irrigation withdrawals for agricultural. The format of these data varies from data tables, typically compiled at the county level, to spatial Geographic Information System (GIS) polygon layers of agricultural lands. These data sources are often incomplete, out of date, or inconsistently compiled. The USGS and the University of Wisconsin-Madison developed annual Landsat-based Irrigation Dataset (LANID), which consists of irrigation maps, derivative products, and manually collected ground reference data covering the conterminous US (CONUS) for the period of 1997–2017 (Xie and Lark, 2021a). These maps were developed using verified irrigated-lands GIS datasets (i.e. training data) coupled with remotely-sensed, 30-meter resolution Landsat-derived data. The current and future availability of verified field-level data is required to train and validate this and other models.

In 2015, agricultural irrigation withdrawals accounted for about 42 percent of the total freshwater withdrawals in the United States (Dieter and others, 2018). Consistent and accurate designations of irrigated agricultural lands, irrigation system type, conveyance systems, and water source (groundwater or surface water) are essential for the determination of irrigation water use and ultimately the sound management of our nation’s water resources. Several local, state, and federal agencies compile data (crops, irrigation, irrigation system type, etc.) that can be used to estimate irrigation withdrawals for agricultural. The format of these data varies from data tables, typically compiled at the county level, to spatial Geographic Information System (GIS) polygon layers of agricultural lands. These data sources are often incomplete, out of date, or inconsistently compiled. The USGS and the University of Wisconsin-Madison developed annual Landsat-based Irrigation Dataset (LANID), which consists of irrigation maps, derivative products, and manually collected ground reference data covering the conterminous US (CONUS) for the period of 1997–2017 (Xie and Lark, 2021a). These maps were developed using verified irrigated-lands GIS datasets (i.e. training data) coupled with remotely-sensed, 30-meter resolution Landsat-derived data. The current and future availability of verified field-level data is required to train and validate this and other models.

The spatial extents of verified irrigated lands were compiled from various federal and state sources across the nation and combined into a single Geographic Information System (GIS) geodatabase for the purpose of model training and validation. In cooperation with U.S. Geological Survey (USGS), researchers at the University of Wisconsin (UW) generated a nation-wide map of irrigated lands using remote-sensing techniques that will be incorporated into future irrigation water-use models. The verified spatial data varies in scope, accuracy, and time period represented, but in general represents GIS coverages (polygons) of agricultural land irrigated for at least some period during 2002–17. Data from 14 states were provided to UW (Arizona, California, Colorado, Florida, Georgia, Idaho, Illinois, Mississippi, Montana, New Mexico, Texas, Utah, Washington, and Wyoming). It is important to validate that the remote sensing techniques correctly identify both irrigated and non-irrigated land. Varying data sources prevent this approach from being applied throughout the United States, but most datasets used for validation include at least some “non irrigated” land identification.

The spatial extents of verified irrigated lands were compiled from various federal and state sources across the nation and combined into a single Geographic Information System (GIS) geodatabase for the purpose of model training and validation. In cooperation with U.S. Geological Survey (USGS), researchers at the University of Wisconsin (UW) generated a nation-wide map of irrigated lands using remote-sensing techniques that will be incorporated into future irrigation water-use models. The verified spatial data varies in scope, accuracy, and time period represented, but in general represents GIS coverages (polygons) of agricultural land irrigated for at least some period during 2002–17. Data from 14 states were provided to UW (Arizona, California, Colorado, Florida, Georgia, Idaho, Illinois, Mississippi, Montana, New Mexico, Texas, Utah, Washington, and Wyoming). It is important to validate that the remote sensing techniques correctly identify both irrigated and non-irrigated land. Varying data sources prevent this approach from being applied throughout the United States, but most datasets used for validation include at least some “non irrigated” land identification.

This dataset represents the extent and spatial distribution of irrigated agricultural lands in the Upper Colorado River Basin for 2007-10. The boundaries in this dataset were modified from data developed by state and local agencies in Colorado, New Mexico, Utah, and Wyoming. The data contain information about the irrigation method used to water the fields and an estimate of the irrigation status of the field for the summer growing seasons between 2007 and 2010. Irrigation method was determined from examination of 1-meter aerial imagery. Irrigation status was estimated from Landsat 5 Thematic Mapper satellite imagery and land cover classification methods.

This dataset represents the extent and spatial distribution of irrigated agricultural lands in the Upper Colorado River Basin for 2007-10. The boundaries in this dataset were modified from data developed by state and local agencies in Colorado, New Mexico, Utah, and Wyoming. The data contain information about the irrigation method used to water the fields and an estimate of the irrigation status of the field for the summer growing seasons between 2007 and 2010. Irrigation method was determined from examination of 1-meter aerial imagery. Irrigation status was estimated from Landsat 5 Thematic Mapper satellite imagery and land cover classification methods.

The Upper Missouri River headwaters (UMH) basin (36 400 km2 ) depends on its river corridors to support irrigated agriculture and world-class trout fisheries. We evaluated trends (1984–2016) in riparian wetness, an indicator of the riparian condition, in peak irrigation months (June, July and August) for 158 km2 of riparian area across the basin using the Landsat normalized difference wetness index (NDWI). We found that 8 of the 19 riparian reaches across the basin showed a significant drying trend over this period, including all three basin outlet reaches along the Jefferson, Madison and Gallatin rivers. The influence of upstream climate was quantified using per reach random forest regressions. Much of the interannual variability in the NDWI was explained by climate, especially by drought indices and annual precipitation, but the significant temporal drying trends persisted in the NDWI–climate model residuals, indicating that trends were not entirely attributable to climate. Over the same period we documented a basin-wide shift from 9 % of agriculture irrigated with center-pivot irrigation to 50 % irrigated with center-pivot irrigation. Riparian reaches with a drying trend had a greater increase in the total area with center-pivot irrigation (within reach and upstream from the reach) relative to riparian reaches without such a trend (p < 0.05). The drying trend, however, did not extend to river discharge. Over the same period, stream gages (n = 7) showed a positive correlation with riparian wetness (p < 0.05) but no trend in summer river discharge, suggesting that riparian areas may be more sensitive to changes in irrigation return flows relative to river discharge. Identifying trends in riparian vegetation is a critical precursor for enhancing the resiliency of river systems and associated riparian corridors.