In arid and semi-arid landscapes, springs are important sources of surface water and soil moisture for wildlife and vegetation. Springs have been suggested as possible climate-change refugia, however, springs that depend on recent precipitation or snowmelt for recharge may be vulnerable to warming and drought intensification. Effective conservation of spring-dependent ecosystems—and the biodiversity they support—requires empirical approaches to assess resilience of springs to water cycle changes. Unfortunately, hydrologic data is unavailable for most springs, and those hydrologic records that do exist typically have inadequate temporal extent and resolution to adequately characterize resilience. To help fill these data gaps, remote sensing can be used to characterize seasonal and inter-annual changes in vegetation condition and moisture availability in spring-dependent ecosystems. In 2016, the U.S. Geological Survey initiated a study to employ Normalized Difference Vegetation Index (NDVI) from 1985-2011 Landsat imagery to (1) develop methods to delineate surface-moisture zones (SMZs) in the vicinity of springs, and (2) characterize the relative resilience or vulnerability of SMZs to inter-annual changes in water availability. For 32 spring clusters in a sage-steppe landscape in southeastern Oregon, USA, 7 NDVI-based indicators of resilience were developed, synthesized, and examined in relation to topographic and hydrologic characteristics.
Preliminary analysis shows that SMZs were generally located immediately downgradient of springs and/or along riparian areas of spring-fed streams. Seven NDVI-based indicators of SMZ resilience to inter-annual changes in water availability produced similar results in situating spring clusters along a resilience-vulnerability gradient: (1) mean, and (2) standard deviation of July NDVI; (3) mean difference in July NDVI and (4) difference in coefficient of variation for July NDVI between each SMZ and its surrounding watershed; (5) strength of the relationship between SMZ July NDVI and 90-day antecedent precipitation, (6) strength of the NDVI relationship with previous winter’s snowpack, and (7) range of NDVI values from an exceptionally wet year followed by 4 dry years. Because all 7 resilience indicators were highly inter-correlated (all P < 0.01), Principal Components Analysis was used to derive an overall metric of SMZ resilience, which accounted for 74% of total variance. This overall resilience metric was positively associated with mean SMZ elevation and mean slope, and was not associated with SMZ size or number of springs in each cluster. This preliminary analysis demonstrates the utility of time-series analysis of remotely-sensed NDVI to evaluate the potential resilience of spring-dependent ecosystems to inter-annual changes in water availability.