Estimating urban tree recovery after drought using an eco-hydrologic model parameterized by remote sensing data

Urban trees provide several ecosystem services including stormwater filtering and reducing the urban heat island, but these services depend on tree health and may change in response to climate extremes such as drought. Quantifying how tree health and ecosystem services change over time is challenging due to lack of data and the spatial heterogeneity of urban layouts. Remote sensing can capture the spatial distribution of tree canopies at fine resolutions, but is limited to singular moments in time. To use this data to make predictions of how vegetation will change with climate, we combined LAI data produced from hyperspectral imagery and waveform lidar with a distributed eco-hydrologic model. The model, Regional Hydro-Ecologic Simulation System (RHESSys), simulates water, carbon, and energy fluxes, which depend on several vegetation parameters that affect how plants respond to water stress. It was initialized with maps of LAI and carbon data to capture pre-drought conditions. The model was then used to simulate evapotranspiration and net primary production (NPP), as proxies of tree resilience to water stress. Simulations estimated plant responses both prior to, during, and following a drought (2012-2016) for Santa Barbara, a medium sized urban area in semi-arid southern California. The simulations were run at 3 spatial scales: a single patch with one species of tree, an urban park with a stand of mixed vegetation, and a small hillslope with a mix of vegetation and impervious area. By running the model at different spatial levels, we were able to explore the various pathways through which water contributes to vegetation drought vulnerability and post-drought vegetation recovery. All three spatial levels were ran with and without irrigation. Irrigation rates were derived from long-term Santa Barbara water use data sets and were assumed to be constant throughout the simulation period. The single patch scale and urban park scale have only direct precipitation or irrigation as local water inputs. At the hillslope scale, both upslope lateral groundwater subsidy and additional runoff from impervious area contribute to tree water availability. The results show that recovery depends substantially on tree type, and show the extent to which drought response and recovery time decreases with irrigation.