Mesic vegetation persistence: A new approach for monitoring spatial and temporal changes in water availability in dryland regions using cloud computing and the sentinel and Landsat constellations

Nawaraj Shrestha; Nicholas E Kolarik; Jodi S Brandt

doi:10.1016/j.scitotenv.2024.170491

Mesic vegetation persistence: A new approach for monitoring spatial and temporal changes in water availability in dryland regions using cloud computing and the sentinel and Landsat constellations

Sci Total Environ. 2024 Mar 20:917:170491. doi: 10.1016/j.scitotenv.2024.170491. Epub 2024 Feb 1.

Authors

Nawaraj Shrestha¹, Nicholas E Kolarik², Jodi S Brandt²

Affiliations

¹ Human-Environment Systems, Boise State University, 1910 University Dr., Boise, ID 83725, USA; Conservation Survey Division, School of Natural Resources, University of Nebraska-Lincoln, Lincoln, NE 68583, USA. Electronic address: nawaraj.shrestha@huskers.unl.edu.
² Human-Environment Systems, Boise State University, 1910 University Dr., Boise, ID 83725, USA.

PMID: 38301786
DOI: 10.1016/j.scitotenv.2024.170491

Abstract

Climate change and anthropogenic activity pose severe threats to water availability in drylands. A better understanding of water availability response to these threats could improve our ability to adapt and mitigate climate and anthropogenic effects. Here, we present a Mesic Vegetation Persistence (MVP) workflow that takes every usable image in the Sentinel (10-m) and Landsat (30-m) archives to generate a dense time-series of water availability that is continuously updated as new images become available in Google Earth Engine. MVP takes advantage of the fact that mesic vegetation can be used as a proxy of available water in drylands. Our MVP workflow combines a novel moisture-based index (moisture change index - MCI) with a vegetation index (Modified Chlorophyll Absorption Ratio Vegetation Index (MCARI2)). MCI is the difference in soil moisture condition between an individual pixel's state and the dry and wet reference reflectance in the image, derived using 5th and 95th percentiles of the visible and shortwave infra-red drought index (VSDI). We produced and validated our MVP products across drylands of the western U.S., covering a broad range of elevation, land use, and ecoregions. MVP outperforms NDVI, a commonly-employed index for mesic ecosystem health, in both rangeland and forested ecosystems, and in mesic habitats with particularly high and low vegetation cover. We applied our MVP product at case study sites and found that MVP more accurately characterizes differences in mesic persistence, late-season water availability, and restoration success compared to NDVI. MVP could be applied as an indicator of change in a variety of contexts to provide a greater understanding of how water availability changes as a result of climate and management. Our MVP product for the western U.S. is freely available within a Google Earth Engine Web App, and the MVP workflow is replicable for other dryland regions.

Keywords: Mesic vegetation dynamics; Mesic vegetation persistence; Moisture change index; Time series remote sensing; Water availability.