Utah State (US) 0.6-Meter High-Resolution Greenspace and Vegetated Land Cover Dataset (2021)

Project summary, description or abstract: This project developed a new machine learning-based technology for classifying very high-resolution (0.6-meter) aerial imagery provided through the National Agriculture Imagery Program (NAIP), a program administered by the U.S. Department of Agriculture that acquires leaf-on aerial imagery during the growing season to support agricultural and environmental monitoring. The tool was successfully applied to generate a detailed greenspace land cover map for the entire state of Utah, using 2021 NAIP imagery at 0.6-meter resolution. Brief description of collection and processing of data: Aerial imagery with a spatial resolution of 0.6 meters, collected in 2021 through the National Agriculture Imagery Program (NAIP), was obtained from the USGS EarthExplorer platform. This imagery covers the entire state of Utah. We applied a combination of advanced machine learning methods to classify the imagery into greenspace and non-greenspace land cover. To optimize file size while maintaining high classification accuracy, non-greenspace areas were subsequently recoded as NoData values. Description of files (names, or if too numerous, number of files, file type(s): This dataset consists of six 7z ziped file, which need to be all downloaded and unzip with 7z software. After unzipped, there is a TIFF file with related spatial information files that can be opened with ArcGIS Pro, representing vegetated land cover across Utah in 2021. The spatial resolution of the data is 0.6 meters. Description or definition any other unique information that would help others use your data: This dataset is intended for use in environmental planning, ecological analysis, urban greenspace assessment, and related geospatial studies. The classified vegetated land cover represents areas with visible vegetation during the 2021 growing season, based on aerial imagery captured under leaf-on conditions. Users should note that non-vegetated areas have been set to NoData to reduce file size and improve performance in GIS applications. The dataset is in a projected coordinate system consistent with standard U.S. aerial imagery (UTM Zone 12N, NAD83). It is compatible with most GIS software platforms that support GeoTIFF format. Descriptions of parameters/variables a. Temporal (beginning and end dates of data collection) August – October, 2021b. Instruments used and units of measurements: The spatial resolution of the imagery is 0.6 meters (60 centimeters per pixel).c. Column headings of data files (for tabular data): Not applicable. This dataset is in raster (TIFF) format, not tabular.d. Location/GIS Coverage (if applicable to data):State of Utah, United StatesProjection: UTM Zone 12NDatum: NAD83Resolution: 0.6 meterse. Symbol used for missing data: NoData. Uncertainty, precision, and accuracy of measurements, if known:The classification accuracy of the vegetated land cover map was assessed using a set of ground-truth reference points and achieved an overall accuracy of approximately 88.02%. Quality assurance and quality control that have been applied:Quality assurance included careful selection and preprocessing of the 2021 aerial imagery to ensure leaf-on conditions and minimal cloud cover. The machine learning classification model was trained and validated using a stratified random sample of manually labeled data points. Cross-validation techniques and confusion matrices were used to evaluate model performance. Known problems that limit datas use (quality control, sampling issues, etc.):Classification is most accurate in urban areas, and less so in wild areas. As this is a raster GeoTIFF file representing vegetated land cover, example “records” correspond to pixel values: Pixel value = 1: Vegetated land coverPixel value = NoData: Non-vegetated areas Documentation of the machine learning classification methods and accuracy assessments is available in a manuscript that is currently under review, the PI will update this information as soon as the manuscript get published. Utah state boundary shapefiles or GIS layers from official sources may aid in spatial analysis.