These data are a species-level classification map of riparian vegetation in the Colorado River riparian corridor in Grand Canyon, Arizona, USA. The classification is derived from 0.2 m pixel resolution multispectral aerial imagery acquired in May 2013. The classification spans the riparian zone of the river corridor between Glen Canyon Dam near Page, Arizona, and Lake Mead at Pearce Ferry, Arizona. The classification is divided into 5 distinct reaches of the river: Glen Canyon, Marble Canyon, Eastern Grand Canyon, Western Grand Canyon upstream of Diamond Creek, and Western Grand Canyon downstream of Diamond Creek. The method used for classification was a combination of supervised Classification And Regression Tree (CART) analysis and unsupervised ISODATA classification. The data release package contains the individual raster data of riparian species vegetation classification data based on USGS 1:24000 quad boundaries, and tabular data of training and validation point locations, confusion matrix counts as accuracy assessment and National Vegetation Classification (NVC) standard and vegetation classification descriptions. Two FGDC metadata files are included; one for the raster classification data and one for the tabular data.

These data are classification maps of total riparian vegetation along the Colorado River in Grand Canyon from Glen Canyon Dam to Pearce Ferry in Arizona. The data are derived from interpretation of multispectral high resolution airborne imagery that was acquired in May 2013. The total vegetation data have the same 0.2-meter ground resolution as the imagery. These data have not undergone a statistical accuracy assessment, but they are based on methods that included image interpretation to exhaustively identify total vegetation which have been shown to produce very high classification accuracies and excellent correlation between maps of total vegetation produced by independent analysts and ground truth. The data represent total vegetation that is primarily green and photosynthetically active at the time of image acquisition, and do not necessarily represent vegetation at various stages of senescence or defoliated/dead vegetation that may still be rooted and standing.

These data are classification maps of total riparian vegetation along the Colorado River in Grand Canyon from Glen Canyon Dam to Pearce Ferry in Arizona. The data are derived from interpretation of multispectral high resolution airborne imagery that was acquired in May 2013. The total vegetation data have the same 0.2-meter ground resolution as the imagery. These data have not undergone a statistical accuracy assessment, but they are based on methods that included image interpretation to exhaustively identify total vegetation which have been shown to produce very high classification accuracies and excellent correlation between maps of total vegetation produced by independent analysts and ground truth. The data represent total vegetation that is primarily green and photosynthetically active at the time of image acquisition, and do not necessarily represent vegetation at various stages of senescence or defoliated/dead vegetation that may still be rooted and standing.

These data were compiled to assess the status and trends of riparian plant communities along the Colorado River between Glen Canyon Dam and Lake Mead, AZ. Three metrics have been proposed to evaluate the "Riparian Vegetation" goal identified in the Glen Canyon Dam Adaptive Management Program's Long Term Experimental and Management Plan (U.S. Department of Interior, 2016). The three metrics are total living plant cover, the proportion of living cover composed of native species, and native species richness. Current policies for Glen Canyon Dam operations result in three longitudinal bands within the riparian area that are flooded at different frequencies. The band, or hydrologic zone, that is most frequently inundated is referred to here as the “active channel” or “AC.” This includes all areas inundated by releases up to 25,000 cubic feet per second (707 m3/s). The “active floodplain” or “AF” is inundated by high flow experiments and includes areas that are inundated by releases between 25,000 cubic feet per second and 45,000 cubic feet per second (1,274 m3/s). The “inactive floodplain” or “IF” is the area along the river that is inundated by releases over 45,000 cubic feet per second, which is not planned under current policies. The metrics are assessed for each of these hydrologic zones. Data from the Grand Canyon Monitoring and Research Center's riparian vegetation monitoring protocol (Palmquist and others, 2018) can be used to evaluate these metrics, which is what is provided here. In short, 80-100 sample sites are randomly selected each year. These sites include debris fans, eddy sandbars, and channel margins. At each randomly selected sample site, ocular cover estimates of each plant species occurring in 1-m2 quadrats spanning the hydrological zones are recorded, along with an estimate of total living plant cover and associated environmental variables. The first metric, total living plant cover, consists of two pieces of data; plant occurrence (a plant is present in the sample frame) and plant cover (proportion of the sample frame covered with living plants). Cover is represented by both an ordinal cover class (1, 2, 3, 4, 5, 6, etc.) and the midpoint of the cover class value (0.01%, 0.5%, 1%, 5%, 10%, 15%, etc). The proportion of native cover is the sum total of native plant cover divided by the sum total of plant cover (native plus nonnative cover) for a sample frame. Native plant richness is the total number of native species rooted inside a sample frame. The total living plant cover data are available for 2016 through 2023. The native cover and richness data are available for 2014 and 2016 through 2023.

These data were collected as part of a methodological comparison for collecting riparian vegetation data. Two common methods for collecting vegetation data were used: line-point intercept and 1m2 ocular quadrats (visual cover estimates). At each site and transect, both methods were used to collect cover and composition data by four different observers. The same transects and quadrats were utilized for both methods and all observers. Field data collected included percent cover for total living foliar cover, each plant species encountered, litter, dead plant material that is still standing, and ground cover features (biological soil crust, rock, sand, and fine soil particles). Line-point intercept data were collected at 25 cm intervals along each transect and at four points along the edge of each 1m2 quadrat. Since transects varied in length, the number of data points collected along each transect also varied. A pin flag was dropped vertically to the ground at 25 cm intervals and every plant species and ground cover element that touched the pin flag was recorded in the order it touched the pin flag from top to bottom, including any species that would touch the pin flag if it continued upward indefinitely. Each species was only recorded once at each point. Ocular quadrat data were collected at each of the 1 m2 quadrats. Cover estimates were recorded to the nearest 5% other than those estimates under 5% which were recorded as either 1% or “trace”. Observers calibrated their ocular estimates at the beginning of sampling and when a new observer started sampling. Observers were given reference cards illustrating multiple levels of percent cover (1 – 95%), which were used during calibration and throughout data collection. Five observers with three levels of experience participated in this study. Two observers had extensive experience with identification of plant species in the study area, as well as with the methods used. One observer was familiar with the methods as well as riparian plant identification, but had not previously worked in this study area. Two observers had not worked in this system or with these methods before, but had experience conducting vegetation surveys. All observers received on-site training. At each site, four observers sampled the entire site using both field methods.

These data were collected as part of a methodological comparison for collecting riparian vegetation data. Two common methods for collecting vegetation data were used: line-point intercept and 1m2 ocular quadrats (visual cover estimates). At each site and transect, both methods were used to collect cover and composition data by four different observers. The same transects and quadrats were utilized for both methods and all observers. Field data collected included percent cover for total living foliar cover, each plant species encountered, litter, dead plant material that is still standing, and ground cover features (biological soil crust, rock, sand, and fine soil particles). Line-point intercept data were collected at 25 cm intervals along each transect and at four points along the edge of each 1m2 quadrat. Since transects varied in length, the number of data points collected along each transect also varied. A pin flag was dropped vertically to the ground at 25 cm intervals and every plant species and ground cover element that touched the pin flag was recorded in the order it touched the pin flag from top to bottom, including any species that would touch the pin flag if it continued upward indefinitely. Each species was only recorded once at each point. Ocular quadrat data were collected at each of the 1 m2 quadrats. Cover estimates were recorded to the nearest 5% other than those estimates under 5% which were recorded as either 1% or “trace”. Observers calibrated their ocular estimates at the beginning of sampling and when a new observer started sampling. Observers were given reference cards illustrating multiple levels of percent cover (1 – 95%), which were used during calibration and throughout data collection. Five observers with three levels of experience participated in this study. Two observers had extensive experience with identification of plant species in the study area, as well as with the methods used. One observer was familiar with the methods as well as riparian plant identification, but had not previously worked in this study area. Two observers had not worked in this system or with these methods before, but had experience conducting vegetation surveys. All observers received on-site training. At each site, four observers sampled the entire site using both field methods.

These data represent total vegetation and surface water along approximately 12 kilometers of the Paria River upstream from the confluence of the Colorado River at Lees Ferry, Arizona. They are derived from airborne, multispectral imagery obtained in late May 2009, 2013, and 2021, collected with a push-broom sensor with 4 spectral bands depicting Blue, Green, Red and Near-Infrared wavelengths at a spatial resolution of 20 centimeters. The vegetation classification data were created using a supervised classification algorithm provided by Harris Geospatial in ENVI version 5.6.3 (Exelis Visual Information Solutions, Boulder, Colorado). The water data were created using a Green Normalized Difference Vegetation Index (gNDVI) threshold of gNDVI <= 0.2. Each classification dataset contains unclassified pixels (0) and vegetation or surface water pixels (1) and are published as Cloud Optimized GeoTIFF raster datasets.

These data represent total vegetation and surface water along approximately 12 kilometers of the Paria River upstream from the confluence of the Colorado River at Lees Ferry, Arizona. They are derived from airborne, multispectral imagery obtained in late May 2009, 2013, and 2021, collected with a push-broom sensor with 4 spectral bands depicting Blue, Green, Red and Near-Infrared wavelengths at a spatial resolution of 20 centimeters. The vegetation classification data were created using a supervised classification algorithm provided by Harris Geospatial in ENVI version 5.6.3 (Exelis Visual Information Solutions, Boulder, Colorado). The water data were created using a Green Normalized Difference Vegetation Index (gNDVI) threshold of gNDVI <= 0.2. Each classification dataset contains unclassified pixels (0) and vegetation or surface water pixels (1) and are published as Cloud Optimized GeoTIFF raster datasets.

These data area classified maps of water in the Colorado River at a discharge of approximately 227 meters squared/second in Grand Canyon from Glen Canyon Dam to Pearce Ferry in Arizona. The data are derived from interpretation of multispectral high resolution airborne imagery that was acquired in May 2013. The water classification data have the same 0.2-meter ground resolution as the imagery. These data have not undergone a statistical accuracy assessment, but they are based on methods that included image interpretation to exhaustively identify water which have been shown to produce very high classification accuracies and excellent correlation between maps of total vegetation produced by independent analysts and ground truth. When developing these data from the native raster format we also considered the differences in water origin, and differentiated between water in the Colorado River mainstem as opposed to within tributary channels. Backwaters with fluid connection to the mainstem river channel were categorized as mainstem water. Backwaters completely disconnected from the mainstem were grouped with the tributary water. We created a water classification dataset from multispectral high resolution imagery. All processing steps were completed in ENVI + IDL 5.3 a product of Harris Geospatial Solutions (copyright 2017 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation) and ArcGIS 10.3 a product of ESRI (copyright 2017).

These data area classified maps of water in the Colorado River at a discharge of approximately 227 meters squared/second in Grand Canyon from Glen Canyon Dam to Pearce Ferry in Arizona. The data are derived from interpretation of multispectral high resolution airborne imagery that was acquired in May 2013. The water classification data have the same 0.2-meter ground resolution as the imagery. These data have not undergone a statistical accuracy assessment, but they are based on methods that included image interpretation to exhaustively identify water which have been shown to produce very high classification accuracies and excellent correlation between maps of total vegetation produced by independent analysts and ground truth. When developing these data from the native raster format we also considered the differences in water origin, and differentiated between water in the Colorado River mainstem as opposed to within tributary channels. Backwaters with fluid connection to the mainstem river channel were categorized as mainstem water. Backwaters completely disconnected from the mainstem were grouped with the tributary water. We created a water classification dataset from multispectral high resolution imagery. All processing steps were completed in ENVI + IDL 5.3 a product of Harris Geospatial Solutions (copyright 2017 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation) and ArcGIS 10.3 a product of ESRI (copyright 2017).