Background/Question/Methods
Groundwater dependent ecosystems (GDEs) serve as relatively resilient habitats for a variety of flora and fauna. Their health is dependent on many factors, including surface flow, rainfall and groundwater pumping. Monitoring their health is challenging due to the sub-surface nature of a key metric of their health- depth to groundwater (DGW). This study develops a method using climate and remote sensing data to monitor GDEs throughout California.
The California Indicators of Groundwater Dependent Ecosystems (iGDEs) dataset was intersected with annual DGW from USGS and California Statewide Groundwater Elevation Monitoring (CASGEM) wells across the state from 1985 to 2018. This dataset was then summarized with DAYMET climate data, Landsat-based change and seasonality metrics from Google Earth Engine (GEE) over 34 years, and several vegetation and ecoregion-related strata in a Random Forests model to better understand the ability to model DGW with regularly available data.
Results/Conclusions
The Random Forest model had a R2 of 0.74 and RMSE 8.04 ft . When the vegetation and ecoregion-related strata were removed, R2 was 0.69 and RMSE was 8.81 ft. This indicated that the ability to monitor DGW with this modelling approach varied across different ecoregions and vegetation communities. Despite this variability, our results suggest climate and remotely sensed change and seasonality metrics can be used to monitor iGDEs in areas without long term well data. By identifying iGDEs with high model error, this method can also be used as a cost-effective method to identify areas where additional monitoring wells should be installed to more adequately monitor these ecosystems.