North Florida Forest Water Yield Geospatial Tool

Tool overview and LAI

This ArcGIS Pro tool predicts water yields resulting from adjustments to forest structure. Specifically, it predicts water yields associated with target Leaf Area Index (LAI) values for areas of interest. The tool embeds a model (Acharya et al. 2022) that predicts water yield from LAI, Aridity Index (potential evapotranspiration [PET] divided by precipitation), and a binary variable of mean groundwater depth ("shallow" as <2.5 meter depth and "deep" as >5 meter depth). We computed baseline LAI in Google Earth Engine (GEE) Javascript from Landsat 8 satellite data using an algorithm for southeastern US pine forests (Blinn et al. 2019). LAI was aggregated over the past 5 years; for this tool version, this is 2017-2022.

The water yield equation: 

Water Yield = 197.6 – 137.2*Aridity Index – 9.7*LAI + 16.8*Groundwater Depth.

Contents of ForestWaterYieldTool.zip

The tool package includes an ArcGIS Pro document. Open this document to run the tool and to view rasters of baseline LAI and baseline water yield. Both rasters are located in ForestWaterYieldTool.gdb (a file geodatabase). The folder SampleAOI_Shapefile contains a shapefile with sample Areas of Interest, to allow easy trial of the tool.

Values in the water yield raster have been multiplied by 100 and values in the LAI raster are multiplied by 1000 -- this was done to reduce file size (because integer files are smaller than float type). For instance, a pixel value of 2842 indicates a water yield of 28.42 cm/year. For LAI, a pixel value of 3249 means an LAI of 3.249. The tool adjusts for this, but you must adjust accordingly if doing your own analysis of the datasets.

LAI and water yield have non-null pixel values at valid land cover types (Deciduous Forest, Evergreen Forest, Mixed Forest, Shrub/Scrub, Grassland/Herbaceous, and Woody Wetlands), based on 2019 NLCD. The Woody Wetlands category was screened using a remote sensing water index (Lefebvre et al., 2019), to remove areas wetter than typical upland forest in the region. 

Tips for running (including troubleshooting)

Even a successful tool run will yield a "completed with warnings" message from ArcGIS -- this is fine.

Troubleshooting: If the tool fails, try these steps, re-running after each: 1) if writing to the same Excel file as a previous run, ensure that Excel file is closed, 2) ensure the Maximum LAI is higher than (or equal to) the Minimum LAI, 3) click the refresh button in ArcGIS Pro, 3) close and re-open the ArcGIS Pro map, and 4) re-start your computer.

Some users have reported non-specific errors when running from cloud or network folders. Try moving the package to a local folder (like your Desktop).

The tool is password-protected as a barrier to unintentional modification. To edit the tool or to view tool mechanics, the password is “password.” 

Tool output

The output of the tool is an Excel file with:

• CurrentLAI: Mean current/baseline LAI, averaged over all valid pixels within the AOI (based on pixel centroid).

• CurrentWY: Mean current/baseline water yields, averaged over all valid pixels within the AOI.

• TotalAcres: The summed acreage of all pixels within the AOI, regardless of land cover type.

• ValidAcres: The summed acreage of all valid pixels (those of valid land cover type) within the AOI, for which water yield is calculated.

• PctExcluded: Calculated as 1-ValidAcres/TotalAcres, the % of pixels excluded from the calculation because of invalid land cover type (e.g., developed land, open water, cultivated, emergent herbaceous wetlands, or woody wetland not meeting the screening criteria).

• WY_atLAI_min… WY_atLAI_max (one or more fields): Mean new water yields for each target LAI. Each target LAI value is appended to the field name. For instance, if the user sets the tool to calculate water yield for LAIs of 2 through 4 at an increment of 0.2, there will be 11 WY_atLAI fields spanning from WY_atLAI_2, WY_atLAI_2.2, … , WY_atLAI_4. 

Contents of OtherVariables_Supplemental_NotNeededforTool.zip

The two other variable datasets (Aridity Index and groundwater depth category) used to calculate baseline water yield are not necessary for running the tool. We provide them here as contextual information.

Aridity Index was calculated in GEE using PRISM ~4x4 km resolution precipitation and temperature (https://prism.oregonstate.edu/recent/). We estimated PET from temperature using Hargreaves-Samani (Hargreaves & Samani, 1982, Samani, 2000). We calibrated Aridity Index using field-based Aridity Index from the Acharya et al. study. Groundwater category is generated primarily using a Depth to Water Table raster (Baker, 2022), with 0-2.5 meter depths set to "shallow" or 0, and depths >5 meters to "deep" or 1. For the 2.5-5 meter range, we used a shapefile of aquifer confinement (Williams & Kuniansky, 2015), setting confined or thinly confined areas to "shallow" and unconfined areas to "deep."

In the Acharya et al. study, water table depths at field plots had a bimodal distribution: confined sites had 0-2.5 meter depth versus unconfined sites with >5 meter depth. However, this pattern does not hold in the Big Bend coastal region, a low-elevation, unconfined area with shallow water tables. Accordingly, we devised a new methodology rather than relying solely on a map of confinement.