Partitioning evapotranspiration in drylands using eddy covariance fluxes and ECOSTRESS data
Thursday, August 5, 2021
Link To Share This Poster: https://cdmcd.co/dEyW55 Live Discussion Link: https://cdmcd.co/83bK4k
Emma G. Reich, Kimberly E. Samuels-Crow and Kiona Ogle, School of Informatics, Computing, and Cyber Systems, Northern Arizona University, Flagstaff, AZ, John B. Bradford and Daniel R. Schlaepfer, Southwest Biological Science Center, U.S. Geological Survey, Flagstaff, AZ, Marcy Litvak, Department of Biology, University of New Mexico, Albuquerque, NM, Daniel R. Schlaepfer, Center for Adaptable Western Landscapes, Northern Arizona University, Flagstaff, AZ, Kiona Ogle, Center for Ecosystem Science & Society, Northern Arizona University, Flagstaff, AZ
Presenting Author(s)
Emma G. Reich
School of Informatics, Computing, and Cyber Systems, Northern Arizona University Flagstaff, AZ, USA
Background/Question/Methods Drylands are experiencing greater drought, making it increasingly important to understand how water fluxes (e.g., evapotranspiration [ET]) respond to changing climate. Partitioning ET into evaporation (E) and transpiration (T) can inform how and when climatic variables drive ET across different water-limited biomes and how ecosystems could respond to climate change. Many drylands experience episodic environmental events (notably monsoon rains), so partitioning ET over short time periods is critical to explore the processes leading to temporal variation in these fluxes. Our objectives are to partition ET and evaluate how climatic factors influence physical (E) and biologically driven (T) processes seasonally and across biome types. To address our objectives, we developed a semi-mechanistic model of ET, informed by eddy covariance flux tower data and remotely sensed (ECOSTRESS) water-use efficiency (WUE), to partition ET into E and T at a sliding weekly time-step, across seven sites along an elevation and aridity gradient in New Mexico. The model describes the relationship between T and gross primary productivity (GPP), constrained by WUE and simple physical equations for soil and intercepted E. The model was fit to multiple data sources (ET, GPP, WUE) within a Bayesian framework to partition ET on a weekly timescale over a 10-year period, allowing us to explore the temporal dynamics of E, T, T/ET, and how they are governed by episodic climate events. Results/Conclusions Overall, T dominates ET across the elevation gradient, and T and E become temporally decoupled around episodic precipitation events. Weekly growing season T/ET was similar across the elevation gradient (about 0.80 ± 0.15 [mean ± standard deviation]) despite higher ET at the highest elevations. Across all sites, E was most strongly correlated with precipitation (R2 ≈ 0.64), suggesting a strong relationship between E and water inputs. Across all sites, T was most strongly correlated with air temperature (R2 ≈ 0.39), particularly at higher elevations (R2 ≈ 0.58) where there is more seasonal temperature variability, suggesting that seasonality of plant activity is more meaningful than episodic storms for T. E and T covaried more tightly at lower (R2 ≈ 0.33) than higher (R2 ≈ 0.07) elevations. However, during the monsoon season, there is approximately an 8-day lag between peak E and subsequent peak T. These results point to the importance of episodic climate events, seasonality, and biome type in governing E, T, and E/T dynamics in drylands.