Color: [130, 130, 130, 255] Background Color: N/A Outline Color: N/A Vertical Alignment: bottom Horizontal Alignment: left Right to Left: false Angle: 0 XOffset: 0 YOffset: 0 Size: 7 Font Family: Arial Font Style: normal Font Weight: normal Font Decoration: none
Description: The amount of suitable habitat for poplar increases dramtically if irrigation is available. Researchers were interested in determining whether or not specific landowners are able to irrigate their property as a part of a suitability analysis. The ability to irrigate depends on landowners having water rights. Water rights information is not available for the majority of the AHB-NW five state region. Another method needed to be used to predict the likelyhood that a landowner has water rights.The presence of photosynthetically active vegetation in otherwise dry agricultural areas during the summer suggests that the area is irrigated and therefore that the landowner has water rights.A data product called the Global Land Survey (GLS) is produced from Landsat data by the United States Geological Survey (USGS) and the National Aeronautics and Space Administration (NASA). GLS are global, minimal-cloud cover, orthorectified data products used for assessments of land-cover, land cover-change, and ecosystem dynamics such as disturbance and vegetation health. GLS datasets are built from Landsat scenes from many dates centered around a specfic year (epoch). A GLS data product was created for the 2010 epoch with imagery from Landsat 7 Enhanced Thematic Mapper Plus (ETM+) and Landsat 5 Thematic Mapper (TM) collected on dates ranging from 2009 to 2011. Photosynthetically active vegetation is unique in that it absorbs red wavelength light and reflects near-infrared (NIR) wavelength light. Landsat sensors collect electromagnetic radiation in multiple wavelengths (bands). By using the red and NIR bands from the 2010 GLS, a metric called Normalized Difference Vegetation Index (NDVI) was calculated. NDVI = (NIR - red)/(NIR + red). High NDVI values indicate low red values and high NIR values, while low NDVI values indicate the opposite. A high NDVI value therefore indicates the likely presence of photosynthetically active vegetation, while a low NDVI value indicates its absence.Environmental Systems Research Institute (ESRI) provides NDVI change products as image services through ArcGIS Online that compare GLS NDVI datasets from two dates showing vegetation changes over time. For this project, NDVI Change 2000-2010 was used. The 2010 data (Band 1) was used as the source NDVI data.The NDVI data was provided by ESRI in the GCS_WGS_84 (EPSG:4326 -- http://spatialreference.org/ref/epsg/4326/) projection with 0.00026949439 degree by 0.00026949459 degree pixels (approximately 30 meters depending on latitude). It was projected to the AHB-NW Pacific Northwest Albers projection (http://spatialreference.org/ref/sr-org/7260/) and resampled from 30 meter to 32 meter pixels using the nearest neighbor method. 32 meter pixels fit within the nested grid of rasters developed by UC Davis.NDVI values were converted into four classes of irrigation likelihood: highly likely, likely, unlikely, and highly unlikely. This conversion was done through comparison with National Agriculture Imagery Program (NAIP) imagery provided by the USGS as an ArcGIS service (http://isse.cr.usgs.gov/ArcGIS/rest/services/Orthoimagery/USGS_EDC_Ortho_NAIP). A set of 100 sample points were randomly located in areas manually identified as irrigated agricultural lands in the NAIP imagery. Each point may have been located in an irrigated field, an unirrigated field, or in unirrigated areas around fields. Each point was manually classified as irrigated or unirrigated based on its appearance in the NAIP imagery. The NDVI values for each sample point were determined.A script was written to determine the threshold NDVI value dividing irrigated and unirrigated lands. Each NDVI value from the minimum to the maximum was iteratively selected as the threshold between irrigated and unirrigated (NDVI values below the threshold are considered unirrigated, NDVI values above the threshold are considered irrigated) and the manual classification from the NAIP imagery was examined. If a point's NDVI value was below the threshold it was considered unirrigated. If that point was manually classified in the NAIP imagery as irrigated, it was mis-classified at the given NDVI threshold. If a point's NDVI value was above the threshold it was considered irrigated. If that point was manually classified in the NAIP imagery as unirrigated, it was also mis-classified at the given NDVI threshold. The NDVI threshold value that minimized the number of mis-classified points was selected as the threshold dividing irrigated and unirrigated lands.Irrigated lands were further divided into two classes, highly likely irrigated and likely irrigated, at the NDVI value halfway between the NDVI threshold and the NDVI value at which all of the manually classified irrigated points were mis-classified.Unirrigated lands were further divided into two classes, highly unlikely irrigated and unlikely irrigated, at the NDVI value halfway between the NDVI threshold and the NDVI value at which all of the manually classified unirrigated points were mis-classified.The minimum raster cell size used in the AHB-NW poplar suitability analysis is 256 meters. The NDVI data and the irrigation likelihood data were both 32 meter cells. There are 64 32 meter cells in each 256 meter cell (8 cells by 8 cells). The count of the 32 meter cells in each irrigation likelihood class was calculated, and this count was divided by 64 to get the percent of each 256 meter cell in each likelihood class.The following formula was used to calcaulated a weighted average likelihood value for each cell: ((Class 4 Count/64) * 0) + ((Class 3 Count/64) * 0.33333333333333) + ((Class 2 Count/64) * 0.66666666666667) + ((Class 1 Count/64) * 1)
Copyright Text: University of Washington School of Environmental and Forest Sciences, ESRI, USGS, NASA