CSDMS 2015 annual meeting poster CrystalNg
Integrating data and models with an ensemble Kalman filter to examine Mojave Desert ecohydrology
Abstract:
Ecohydrological models that couple soil moisture and plant growth dynamics are needed to examine the vulnerability of desert ecosystems to climate change. However, models can be hindered by structural and parameter errors, and representing deserts is particularly challenging due to unique plant adaptations and marginal moisture conditions. Incorporating field observations with the model is critical for ensuring that simulations produce realistic dynamics. I present a coupled ecohydrological model based on NCAR's CLM4.0-CN [Olesen et al., 2010] that relies on hydrological and ecological data from a Mojave Desert study area to address structural and parameter uncertainty. A new multi-site implementation of the ensemble Kalman filter is developed to condition estimates of plant and soil parameters that are characteristic of the study area. A 50-year hindcast with the model makes it possible to extend the 3.5-year data record in time to examine multi-year relationship between rainfall and plant growth. The simulations also make it possible extend soil moisture data from two points to the full root zone to consider depth-dependent interactions. By combining data and model, we found that the perennial shrub Larrea tridentata (creosote bush) is uniquely adapted to desert conditions. Simulations for the study area show its growth is strongly controlled by ~3-year rainfall totals, which it mostly maintains by immediate uptake of moisture in the upper 80 cm of the soil, but also importantly by using more persistent moisture at 80 to 100 cm depth when available. These results highlighting the potential ecohydrological impact of interannual climate change would not have been possible with either the model or data alone.
* Please acknowledge the original contributors when you are using this material. If there are any copyright issues, please let us know and we will respond as soon as possible.
