We apply a research approach that can inform riparian restoration planning by developing products that show recent trends in vegetation conditions identifying areas potentially more at risk for degradation and the associated relationship between riparian vegetation dynamics and climate conditions. The full suite of data products and a link to the associated publication addressing this analysis can be found on the Parent data release. For this study, the vegetation conditions are characterized using a series of remote sensing vegetation indices developing using satellite imagery, including the Normalized Difference Vegetation Index (NDVI) and the Tasseled Cap (TC) Transformation. The NDVI is a commonly used vegetation index that quantifies relative greenness of the vegetation based on the plant’s photosynthetic activity, measured as a ratio between the Near Infrared (NIR) and Red bands (Tucker, 1979). The NDVI equation follows: NDVI = (NIR band - Red band) / (NIR band + Red band). NDVI has a range of -1 to 1, though green vegetation theoretically ranges from 0 to 1. Dense green vegetation is represented with values closer to 1 while barren soil, rock, and less-dense surface vegetation has values closer to 0. Values below 0 often represent water due to its unique reflective characteristics. The TC transformation is an approached used to transform satellite imagery into a collection of spectral metrics that can quantify various aspects of the vegetation and soil surfaces (Kauth and Thomas, 1976). Specifically, the TC transformation develops 6 separate metrics, though we only assess the three primary metrics: (i) brightness (transformation 1), (ii) greenness (transformation 2), and (iii) wetness (transformation 3). The TC transformation metrics are calculated using a series of coefficients multiplied across reflectance values for the suite of Landsat bands, then summed across each metric. No specific range is identified for the TC transformation metrics, though for both brightness and greenness, positive values represent brighter and greener conditions, respectively, while negative values represent wetter conditions for wetness. Because bandwidths differ slightly between Landsat 4, 5, 7 and Landsat 8, we use two sets of coefficients and complete the calculation separately before combining the collections into a single series of images (DeVries et al., 2016; Zhai et al., 2022). For each index, we calculate the Sen's Slope across a series of climate periods (see associated Child Item - 1) Database of Trends in Vegetation Properties and Climate Adaptation Variables -- Standardized Precipitation Evapotranspiration Index Timeseries for the Upper Gila River Watershed: 1985 to 2021). All raster products were developed using the Google Earth Engine (GEE) cloud computing software program for the Upper Gila River watershed. This is a Child Item for the Parent data release, Mapping Riparian Vegetation Response to Climate Change on the San Carlos Apache Reservation and Upper Gila River Watershed to Inform Restoration Priorities: 1935 to Present - Database of Trends in Vegetation Properties and Climate Adaptation Variables. This Child Item consists of a single XML metadata file and a zipped file containing the raster stacks. The raster stacks are identified by both the season they represent (i.e., spring, late-spring, summer, fall) and the vegetation index (i.e., NDVI, TC brightness, TC greenness, TC wetness). Each band represents Sen's slope value across each climate period, where band 1 represents the Sen's slope across the 1st climate period (i.e., 1985 through 1993), band 2 represents the Sen's slope across the 2nd climate period (i.e., 1993 through 2014), and band 3 represents the Sen's slope across the 3rd climate period (i.e. 2014 through 2021).

We apply a research approach that can inform riparian restoration planning by developing products that show recent trends in vegetation conditions identifying areas potentially more at risk for degradation and the associated relationship between riparian vegetation dynamics and climate conditions. The full suite of data products and a link to the associated publication addressing this analysis can be found on the Parent data release. For this study, the vegetation conditions are characterized using a series of remote sensing vegetation indices developing using satellite imagery, including the Normalized Difference Vegetation Index (NDVI) and the Tasseled Cap (TC) Transformation. The NDVI is a commonly used vegetation index that quantifies relative greenness of the vegetation based on the plant’s photosynthetic activity, measured as a ratio between the Near Infrared (NIR) and Red bands (Tucker, 1979). The NDVI equation follows: NDVI = (NIR band - Red band) / (NIR band + Red band). NDVI has a range of -1 to 1, though green vegetation theoretically ranges from 0 to 1. Dense green vegetation is represented with values closer to 1 while barren soil, rock, and less-dense surface vegetation has values closer to 0. Values below 0 often represent water due to its unique reflective characteristics. The TC transformation is an approached used to transform satellite imagery into a collection of spectral metrics that can quantify various aspects of the vegetation and soil surfaces (Kauth and Thomas, 1976). Specifically, the TC transformation develops 6 separate metrics, though we only assess the three primary metrics: (i) brightness (transformation 1), (ii) greenness (transformation 2), and (iii) wetness (transformation 3). No specific range is identified for the TC transformation metrics, though for both brightness and greenness, positive values represent brighter and greener conditions, respectively, while negative values represent wetter conditions for wetness. The TC transformation metrics are calculated using a series of coefficients multiplied across reflectance values for the suite of Landsat bands, then summed across each metric. Because bandwidths differ slightly between Landsat 4, 5, 7 and Landsat 8, we use two sets of coefficients and complete the calculation separately before combining the collections into a single series of images (DeVries et al., 2016; Zhai et al., 2022). All raster products were developed using the Google Earth Engine (GEE) cloud computing software program for the Upper Gila River watershed. This is a Child Item for the Parent data release, Mapping Riparian Vegetation Response to Climate Change on the San Carlos Apache Reservation and Upper Gila River Watershed to Inform Restoration Priorities: 1935 to Present - Database of Trends in Vegetation Properties and Climate Adaptation Variables. This Child Item consists of a single XML metadata file and four zipped files containing the raster stacks, where each zipped file contains the rasters for each vegetation index, respectively (i.e., NDVI, TC brightness, TC greenness, TC wetness). The raster stacks within each index-specific zipped folder are identified by the season they represent (i.e., spring, late-spring, summer, fall). Each band within the separate raster stacks (n=37) represents a year from 1985 through 2021 (i.e., band 1 is 1985 and band 37 is 2021).

We apply a research approach that can inform riparian restoration planning by developing products that show recent trends in vegetation conditions identifying areas potentially more at risk for degradation and the associated relationship between riparian vegetation dynamics and climate conditions. The full suite of data products and a link to the associated publication addressing this analysis can be found on the Parent data release. For this study, the vegetation conditions are characterized using a series of remote sensing vegetation indices developing using satellite imagery, including the Normalized Difference Vegetation Index (NDVI) and the Tasseled Cap (TC) Transformation. The NDVI is a commonly used vegetation index that quantifies relative greenness of the vegetation based on the plant’s photosynthetic activity, measured as a ratio between the Near Infrared (NIR) and Red bands (Tucker, 1979). The NDVI equation follows: NDVI = (NIR band - Red band) / (NIR band + Red band). NDVI has a range of -1 to 1, though green vegetation theoretically ranges from 0 to 1. Dense green vegetation is represented with values closer to 1 while barren soil, rock, and less-dense surface vegetation has values closer to 0. Values below 0 often represent water due to its unique reflective characteristics. The TC transformation is an approached used to transform satellite imagery into a collection of spectral metrics that can quantify various aspects of the vegetation and soil surfaces (Kauth and Thomas, 1976). Specifically, the TC transformation develops 6 separate metrics, though we only assess the three primary metrics: (i) brightness (transformation 1), (ii) greenness (transformation 2), and (iii) wetness (transformation 3). The TC transformation metrics are calculated using a series of coefficients multiplied across reflectance values for the suite of Landsat bands, then summed across each metric. No specific range is identified for the TC transformation metrics, though for both brightness and greenness, positive values represent brighter and greener conditions, respectively, while negative values represent wetter conditions for wetness. Because bandwidths differ slightly between Landsat 4, 5, 7 and Landsat 8, we use two sets of coefficients and complete the calculation separately before combining the collections into a single series of images (DeVries et al., 2016; Zhai et al., 2022). All raster products were developed using the Google Earth Engine (GEE) cloud computing software program for the Upper Gila River watershed. This is a Child Item for the Parent data release, Mapping Riparian Vegetation Response to Climate Change on the San Carlos Apache Reservation and Upper Gila River Watershed to Inform Restoration Priorities: 1935 to Present - Database of Trends in Vegetation Properties and Climate Adaptation Variables. This Child Item consists of a two multi-band raster stacks. The raster stacks are identified by the vegetation index they represent (i.e., NDVI, TC greenness). Each band within the separate raster stacks represents a month from January 2014 through December 2020 (i.e., band 1 is January 2014 and band 84 is December 2020). Though the full study extends until December 2021, we chose to remove data from 2021 from the phenological analysis due to increased fire activity during 2021. Additionally, we chose not to include the TC transformation metrics of brightness and wetness to focus on metrics to simply measure vegetation greenness.

We apply a research approach that can inform riparian restoration planning by developing products that show recent trends in vegetation conditions identifying areas potentially more at risk for degradation and the associated relationship between riparian vegetation dynamics and climate conditions. The full suite of data products and a link to the associated publication addressing this analysis can be found on the Parent data release. To characterize the climate conditions across the study period, we use the Standardized Precipitation Evapotranspiration Index (SPEI). The SPEI is a water balance index which includes both precipitation and evapotranspiration in its calculation. Conditions from the prior n months, generally ranging from 1 to 60, are compared to the same respective period over the prior years to identify the index value (Vicente-Serrano et al., 2010). Values generally range from -3 to 3, where values less than 0 suggest drought conditions while values greater than 0 suggest wetter than normal conditions. For this study, we are using the 12-month, or 1-year, SPEI to compare annual conditions within the larger Upper Gila River watershed. The SPEI data was extracted into a CSV spreadsheet using data from the Gridded Surface Meteorological (GRIDMET) dataset, which provides a spatially explicit SPEI product in Google Earth Engine (GEE) at a 5-day interval and a spatial resolution of 4-km (Abatzoglou, 2013). In GEE, we quantify overall mean values of SPEI across each 5-day period for the watershed from January 1980 to December 2021. Using R software, we reduced the 5-day values to represent monthly mean values and constrained the analysis to water year 1980 (i.e., October 1980) through water year 2021 (i.e., October 2021). Using the monthly timeseries, we completed the breakpoint analysis in R to identify breaks within the SPEI time series. The algorithm identifies a seasonal pattern within the timeseries. When the seasonal pattern deviates, a breakpoint is then detected. These breaks can be used to pinpoint unique climate periods in the time series. This is a Child Item for the Parent data release, Mapping Riparian Vegetation Response to Climate Change on the San Carlos Apache Reservation and Upper Gila River Watershed to Inform Restoration Priorities: 1935 to Present - Database of Trends in Vegetation Properties and Climate Adaptation Variables. The spreadsheet attached to this Child Item consists of 5 columns, including the (i) month from January 1985 through October 2021, (ii) the 1-year SPEI monthly time series, (iii) the dates identified as breaks within the breakpoint algorithm, (iv) the breakpoint trend identified within the breakpoint algorithm, and (v) the dates that were used as the climate period breaks in this study. The climate periods identified in this spreadsheet using the SPEI data were used as the climate periods in our riparian study.

We apply a research approach that can inform riparian restoration planning by developing products that show recent trends in vegetation conditions identifying areas potentially more at risk for degradation and the associated relationship between riparian vegetation dynamics and climate conditions. The full suite of data products and a link to the associated publication addressing this analysis can be found on the Parent data release. To characterize the climate conditions across the study period, we use the Standardized Precipitation Evapotranspiration Index (SPEI). The SPEI is a water balance index which includes both precipitation and evapotranspiration in its calculation. Conditions from the prior n months, generally ranging from 1 to 60, are compared to the same respective period over the prior years to identify the index value (Vicente-Serrano et al., 2010). Values generally range from -3 to 3, where values less than 0 suggest drought conditions while values greater than 0 suggest wetter than normal conditions. For this study, we are using the 12-month, or 1-year, SPEI to compare annual conditions within the larger Upper Gila River watershed. The SPEI data was extracted into a CSV spreadsheet using data from the Gridded Surface Meteorological (GRIDMET) dataset, which provides a spatially explicit SPEI product in Google Earth Engine (GEE) at a 5-day interval and a spatial resolution of 4-km (Abatzoglou, 2013). In GEE, we quantify overall mean values of SPEI across each 5-day period for the watershed from January 1980 to December 2021. Using R software, we reduced the 5-day values to represent monthly mean values and constrained the analysis to water year 1980 (i.e., October 1980) through water year 2021 (i.e., October 2021). Using the monthly timeseries, we completed the breakpoint analysis in R to identify breaks within the SPEI time series. The algorithm identifies a seasonal pattern within the timeseries. When the seasonal pattern deviates, a breakpoint is then detected. These breaks can be used to pinpoint unique climate periods in the time series. This is a Child Item for the Parent data release, Mapping Riparian Vegetation Response to Climate Change on the San Carlos Apache Reservation and Upper Gila River Watershed to Inform Restoration Priorities: 1935 to Present - Database of Trends in Vegetation Properties and Climate Adaptation Variables. The spreadsheet attached to this Child Item consists of 5 columns, including the (i) month from January 1985 through October 2021, (ii) the 1-year SPEI monthly time series, (iii) the dates identified as breaks within the breakpoint algorithm, (iv) the breakpoint trend identified within the breakpoint algorithm, and (v) the dates that were used as the climate period breaks in this study. The climate periods identified in this spreadsheet using the SPEI data were used as the climate periods in our riparian study.

We apply a research approach that can inform riparian restoration planning by developing products that show recent trends in vegetation conditions identifying areas potentially more at risk for degradation and the associated relationship between riparian vegetation dynamics and climate conditions. The vegetation is characterized using a series of remote sensing vegetation indices developing using satellite imagery, including the Normalized Difference Vegetation Index (NDVI) and Tasseled Cap (TC) Transformation metrics of brightness, greenness, and wetness. Each of these remote sensing vegetation indices provides a unique characterization of the vegetation properties. For example, NDVI provides a general overview of vegetation condition while the TC Transformation metrics are multi-dimensional and interrelated and can provide distinctive information on conditions within a vegetation ecosystem. The climate conditions are characterized by defining climate periods based on a timeseries of 1-year Standardized Precipitation Evapotranspiration Index (SPEI) values averaged across the Upper Gila River watershed and divided into climate periods (i.e., 1985 through 1993; 1993 through 2014; 2014 through 2021) using the breakpoints algorithm in R Software. Finally, the remote sensing products were developed based on a seasonal framework to obtain more information on the vegetation dynamics. The seasonal framework is defined by averaging multiple images across the following two-month periods: (i) spring (March/April), (ii) late-spring (May/June), (iii) summer (July/August), and (iv) fall (September/October). This Parent data release consists of five Child Items. The first Child Item includes the SPEI timeseries that was used to identify the climate periods applied in this analysis. The second Child Item consists of 16 rasters with 37 bands apiece, where each raster in this Child Item is identified by a combination of the index (i.e., NDVI, TC brightness, TC greenness, TC wetness) and the season (i.e., spring, late-spring, summer, fall), and each band represents a yearly mean value for all years from 1985 through 2021. The third Child Item also consists of 16 rasters, though with only 3 bands apiece. Similarly, for this Child Item, each raster is identified by a combination of the index (i.e., NDVI, TC brightness, TC greenness, TC wetness) and the season (i.e., spring, late-spring, summer, fall). However, each band represents the linear Sen's slope trend across a different climate period, where band 1 represents the Sen's slope trend across the 1st climate period (i.e., 1985 through 1993), band 2 represents the Sen's slope trend across the 2nd climate period (i.e., 1993 through 2014), and band 3 represents the Sen's slope trend across the 3rd climate period (i.e. 2014 through 2021). The fourth and fifth Child Items both include a series of monthly images, stacked into single rasters, for NDVI and TC greenness covering the length of the 3rd climate period (i.e., 2014 through 2021) that were produced to address a series of case studies more directly. Specifically, the fourth Child Item address a phenological analysis while the fifth Child Item addresses a fire-based case study. All raster products were developed using the Google Earth Engine (GEE) cloud computing software program.

We apply a research approach that can inform riparian restoration planning by developing products that show recent trends in vegetation conditions identifying areas potentially more at risk for degradation and the associated relationship between riparian vegetation dynamics and climate conditions. The full suite of data products and a link to the associated publication addressing this analysis can be found on the Parent data release. For this study, the vegetation conditions are characterized using a series of remote sensing vegetation indices developed using satellite imagery, including the Normalized Difference Vegetation Index (NDVI). The NDVI is a commonly used vegetation index that quantifies relative greenness of the vegetation based on the plant’s photosynthetic activity, measured as a ratio between the Near Infrared (NIR) and Red bands (Tucker, 1979). The NDVI equation follows: NDVI = (NIR band - Red band) / (NIR band + Red band). NDVI has a range of -1 to 1, though green vegetation theoretically ranges from 0 to 1. Dense green vegetation is represented with values closer to 1 while barren soil, rock, and less-dense surface vegetation has values closer to 0. Values below 0 often represent water due to its unique reflective characteristics. All raster products were developed using the Google Earth Engine (GEE) cloud computing software program for the Upper Gila River watershed. This is a Child Item for the Parent data release, Mapping Riparian Vegetation Response to Climate Change on the San Carlos Apache Reservation and Upper Gila River Watershed to Inform Restoration Priorities: 1935 to Present - Database of Trends in Vegetation Properties and Climate Adaptation Variables. This Child Item consists of a multi-band raster stack of monthly NDVI images from January 2014 through July 2022 covering the area of the Bylas Fire. We included data from 2022, contrasting the full study which only includes data through 2021, to include additional data regarding our post-fire vegetation response analysis. Each band within the raster stack represents a month from 2014 through 2022 (i.e., band 1 is January 2014 and band 103 is July 2022).

We apply a research approach that can inform riparian restoration planning by developing products that show recent trends in vegetation conditions identifying areas potentially more at risk for degradation and the associated relationship between riparian vegetation dynamics and climate conditions. The full suite of data products and a link to the associated publication addressing this analysis can be found on the Parent data release. For this study, the vegetation conditions are characterized using a series of remote sensing vegetation indices developed using satellite imagery, including the Normalized Difference Vegetation Index (NDVI). The NDVI is a commonly used vegetation index that quantifies relative greenness of the vegetation based on the plant’s photosynthetic activity, measured as a ratio between the Near Infrared (NIR) and Red bands (Tucker, 1979). The NDVI equation follows: NDVI = (NIR band - Red band) / (NIR band + Red band). NDVI has a range of -1 to 1, though green vegetation theoretically ranges from 0 to 1. Dense green vegetation is represented with values closer to 1 while barren soil, rock, and less-dense surface vegetation has values closer to 0. Values below 0 often represent water due to its unique reflective characteristics. All raster products were developed using the Google Earth Engine (GEE) cloud computing software program for the Upper Gila River watershed. This is a Child Item for the Parent data release, Mapping Riparian Vegetation Response to Climate Change on the San Carlos Apache Reservation and Upper Gila River Watershed to Inform Restoration Priorities: 1935 to Present - Database of Trends in Vegetation Properties and Climate Adaptation Variables. This Child Item consists of a multi-band raster stack of monthly NDVI images from January 2014 through July 2022 covering the area of the Bylas Fire. We included data from 2022, contrasting the full study which only includes data through 2021, to include additional data regarding our post-fire vegetation response analysis. Each band within the raster stack represents a month from 2014 through 2022 (i.e., band 1 is January 2014 and band 103 is July 2022).

We developed this spatial database of historic and current floodplains to analyze trends in vegetation conditions, identify areas more at risk of degradation, and assess the relationship between riparian vegetation dynamics and climate conditions.Our study area is the riparian areas along the San Carlos River and Gila River within the San Carlos Apache Reservation and the Upper Gila River Level-4 Hydrologic Unit Code (HUC) watershed. The product was developed by digitizing the active channel and primary riparian floodplain using historic 1935 aerial imagery (Arizona State University GeoData, 2021) and National Agriculture Imagery Program (NAIP) aerial imagery from 2019 (U.S. Department of Agriculture, 2021). This data was used in the development of the following data release: Petrakis, R.E., Norman, L.M., Middleton, B.R., 2023, Mapping Riparian Vegetation Response to Climate Change on the San Carlos Apache Reservation and Upper Gila River Watershed to Inform Restoration Priorities: 1935 to Present - Database of Trends in Vegetation Properties and Climate Adaptation Variables: U.S. Geological Survey data release, https://doi.org/10.5066/P9HL0N5T. And in support of the following publication: Petrakis, R.E., Norman, L.M., Middleton, B.R., 2023. Riparian Vegetation Response amid Variable Climate Conditions across the Upper Gila River Watershed: Informing Tribal Restoration Priorities. Frontiers in Environmental Science. https://doi.org/10.3389/fenvs.2023.1179328. This Parent data release consists of three shapefiles, or Child Items. The first two shapefiles show the extent of the historic (i.e., 1935) and current (i.e., 2019) riparian floodplain, including the active channel and primary riparian floodplain portions, along the San Carlos River and Gila River within the San Carlos Apache Reservation. The first shapefile - 1935 Riparian Floodplain and Active Channel Boundaries for the San Carlos River and Gila River - was produced for 1935, representing historic conditions, by digitizing the active channel and primary riparian floodplain using 1935 aerial imagery. The second shapefile - 2019 Riparian Floodplain and Active Channel Boundaries for the San Carlos River and Gila River - was produced for 2019, representing current conditions, by digitizing the active channel and primary riparian floodplain using National Agriculture Imagery Program (NAIP) aerial imagery from 2019. The final shapefile - 2019 Riparian Floodplain for the San Carlos River and Gila River within the Upper Gila River Level-4 Hydrologic Unit Code Watershed - shows the boundary of the riparian floodplain for the Gila River and San Carlos River across the entire Upper Gila River Level-4 HUC.

We mapped potential climate change refugia for riparian areas in the central and western USA for 2040-2069 and 2070-2099. Riparian refugia are existing riparian areas that are projected to maintain riparian vegetation and associated ecological function under plausible future climates. Four input variables were included in the riparian refugia index: two landscape variables that represent where existing riparian areas may be more resilient to climatic changes (riparian connectedness and landscape diversity) and two climate variables that reflect projected exposure to climate change (runoff and warm days). For the climate variables, we considered two global circulation models: moderately hot and wet (CNRM-CM5) and hot and dry (IPSL-CM5A-MR) under RCP 8.5. The climate variables represented the projected change from a historical baseline (1971-2000) for two future 30-year time periods, mid-century (2040-2069) and late century (2070-2099). The four input variables of uniform pixel size were assigned equal weights and layered together using ArcGIS Pro's Suitability Modeler to create an index for riparian refugial quality. Here we provide raster layers for the riparian refugia index and three of the four input variables including riparian connectedness, runoff, and warm days. The fourth input variable, landscape diversity, was produced by The Nature Conservancy and is available online at The Nature Conservancy's Resilient and Connected Network. The four climate scenarios (CNRM-CM5 2040-2069, CNRM-CM5 2070-2099, IPSL-CM5A-MR 2040-2069, and IPSL-CM5A-MR 2070-209) are included as individual rasters for the riparian refugia index, runoff, and warm days, and are zipped into each base folder. We also provide a geodatabase that contains all the data (riparian refugia index, riparian connectedness, runoff, and warm days).