2023 Irrigated Lands for the Treasure Valley Aquifer: Machine Learning Generated

TV Irrigated Lands 2023 was created for use in water budget studies within the TV study boundary. The area of interest was determined by Hydrology Section staff at IDWR, and a study boundary was given to GIS staff and used to clip the model output. The random forest (RF) model is a type of supervised machine learning algorithm requiring GIS staff to provide manually labeled training data. GIS staff also provide the RF model with several input features, typically raster datasets that help distinguish characteristics of irrigated lands. TV Irrigated Lands 2023 used the following as input features: • Harmonized Landsat 8, 9 OLI and Sentinel 2A, 2B satellites (HLS-2 Landsat Operational Land Imager Surface Reflectance and TOA Brightness Daily Global 30m [2], HLS Sentinel-2 Multi-spectral Instrument Surface Reflectance Daily Global 30m [3] ; bands: SWIR 2, NIR, Blue, and calculated NDVI) • 10-meter digital elevation model [4] (including slope and aspect) • Height Above Nearest Drainage (HAND) [5] • OpenET eeMETRIC monthly evapotranspiration [6] • PRISM 800m Monthly Precipitation [7] For additional information on the imagery averaging process for Landsat and Sentinel-2 imagery, please see below. Additional datasets used only for labeling training data include Mapping EvapoTranspiration at high Resolution with Internalized Calibration (METRIC) [8], IDWR-provided Active Water Rights Place of Use, the Cropland Data Layer [9] for 2023, and the National Agriculture Imagery Program (NAIP) imagery [10] for Idaho 2023. The accuracy of the TV Irrigated Lands 2023 dataset was verified by several methods. Firstly, a validation test is done by withholding a subset of the training data to evaluate how well the model classifies unseen information. Second, GIS staff will run several iterations of the model with variations of training data, with the goal of improving classification for areas consistently misclassified. This process requires GIS staff knowledge, aided by supplementary datasets, to review the area and make decisions. Once a model iteration is determined as ‘final’, a manual mask is created to correct any remaining misclassification in the dataset. Consistent areas of misclassification for TV Irrigated Lands 2023 include the area near major rivers and riparian areas (i.e., the Boise near Garden City and Nyssa, and the Payette near Emmett), and in urban/rural interfaces with mixed irrigation and impervious surfaces. Misclassification largely occurred in areas with active water rights, but no visible irrigation or irrigation infrastructure in the satellite and aerial imagery available. Final decisions made for these areas were conservative, relying heavily on both an active water right and clear indication of diverted water and artificial application of water to a given field. References: [1] https://developers.google.com/earth-engine/apidocs/ee-classifier-smilerandomforest [2] https://developers.google.com/earth-engine/datasets/catalog/NASA_HLS_HLSS30_v002 [3] https://developers.google.com/earth-engine/datasets/catalog/NASA_HLS_HLSL30_v002 [4] https://developers.google.com/earth-engine/datasets/catalog/USGS_3DEP_10m [5] Donchyts, G., Winsemius, H., Schellekens, J., Erickson, T., Gao, H., Savenije, H., & van de Giesen, N. (2016). Global 30m height above the nearest drainage (HAND). Geophysical Research Abstracts, 18, EGU2016-17445-3. EGU General Assembly 2016. [6] https://developers.google.com/earth-engine/datasets/catalog/OpenET_EEMETRIC_CONUS_GRIDMET_MONTHLY_v2_0 [7] https://gee-community-catalog.org/projects/prism_daily/?h=prism#key-methodological-features [8] https://data-idwr.hub.arcgis.com/documents/776cfc545e0944fc89a75d4777031bb4/about [9] https://developers.google.com/earth-engine/datasets/catalog/USDA_NASS_CDL [10] https://developers.google.com/earth-engine/datasets/catalog/USDA_NAIP_DOQQ