Hindcasting Water Availability in Semi-Arid Systems

Dryland systems cover one third of the earth’s land surface and are becoming increasingly drier, but existing datasets do not capture all of the types of available water that sustain these systems. In semi-arid environments, small surface water bodies and areas of mesic vegetation (wetlands, wet meadows, riparian zones) function as critical water resources. However, the most commonly-used maps of water availability are derived from the Landsat time series or single date aerial photographs, and are either too coarse spatially or temporally to effectively monitor dynamics of available water. In Chapter 1, I produced a Sentinel Fusion (SF) water resources time series for a semi-arid mountainous region of the western United States, which includes monthly maps of both a) surface water and b) mesic vegetation at 10-m spatial resolution using freely available Earth observation data on an open access platform. I applied random forest classifiers to multispectral data from the Sentinel-2 time series, synthetic aperture radar (SAR) data from the Sentinel-1 time series, and topographic variables. I compared our SF product with three commonly used and publicly available datasets in the western U.S. and demonstrated improvements in mapping both surface water and mesic vegetation. With nine times finer spatial resolution and more frequent image collection, the SF maps characterize intra-annual dynamics of smaller water bodies (< 30-m wide) and mesic vegetation integral to ecosystem functions in semi-arid systems compared to leading Landsat-derived products. Further, this workflow is easily reproducible, uses freely available data on an open access platform, and can be adopted to help guide land use decisions related to water availability by farmers, ranchers, and conservationists in semi-arid environments. Newer earth observation datasets, including those from the Sentinel constellation, provide opportunities to monitor mesic ecosystems at meaningful spatial scales, but are insufficient for measuring decadal-scale changes. Conversely, the Landsat time series has decades of data, but images are spatially coarse relative to many of the mesic ecosystem areas that sustain dryland systems, resulting in classifications with mixed pixels inadequate for effective monitoring. In Chapter 2, I developed a workflow that uses 10-m classifications produced from fusion of the Sentinel-1 and −2 time series (2017–2020) to estimate sub-pixel proportions of Landsat time series observations (2004–2020). Using random forest regression models, I quantified water resource proportions (WRP) of surface water, mesic vegetation, and upland land covers within each 30-m Landsat pixel. I incorporated ancillary covariates to account for varying topographic conditions, land cover, and climate. Results indicate that this approach consistently estimates sub-pixel proportions of Landsat pixels more accurately compared to spectral mixture analysis (SMA). I then demonstrated the ability of this time series to characterize historical water availability at a case study site with a wel-documented restoration history by qualitatively examining the mesic vegetation dynamics time series to identify system responses to restoration efforts. This approach allows analysts to hindcast observations of Sentinel products and measure water resource dynamics with greater precision over larger temporal scales. I envision these WRP data to be useful for measuring the impacts of conservation interventions, disturbance recovery, or land use changes that pre-date the Sentinel time series. Private land protection is essential for achieving biodiversity conservation goals and mitigating the effects of climate change. Methods of protection focus on maximizing acres conserved to achieve these goals, but their ability to do so may be overestimated if they do little to improve key ecosystems in need of protection. Mesic ecosystems are integral to conservation in dryland systems because they sustain wildlife, livelihoods, and landscape connectivity. These areas are known to have disproportionate ecological importance, but are at risk as periods of drought increase in both frequency and duration and extractive land use activities continue to degrade them. In Chapter Three, I used time series maps of mesic vegetation proportion to estimate water availability and analyzed development and wetness dynamics in varying management contexts within the High Divide, a semi-arid region of Idaho and Montana with diverse land tenure characteristics typical in the American West. I used difference in differences panel regressions to estimate likelihood of development and the proportion of mesic vegetation late in the water year indicative of a healthy riverscape on private lands where conservation easements are implemented compared to non-easements. The results emphasize that the current process of easement implementation is effective in targeting undeveloped, healthy riverscapes on private lands worth conserving, but this analysis shows no evidence that they are better able to create or conserve mesic habitats in semi-arid landscapes. Similarly, I did not find strong evidence that lands placed under easement are less likely to be developed than those that are not. These results contribute to the discussion regarding the importance of rigorous monitoring protocols for conservation easements and other conservation tools. This dissertation provides evidence for using freely available satellite imagery for monitoring restoration efforts and varying management contexts. I demonstrate how petabytes of publicly owned data, freely accessible cloud computing platforms, and knowledge shared by stakeholders can be used to improve the way we interact with the Earth and how they can play a vital role in repairing our relationship with the planet that sustains us.