Translating the Interconnections between Ecological and Hydrological Processes in a Small Watershed into Process Networks using Information Theory

Wednesday, 17 December 2014
Joon Kim1, Nam C Woo2, Sanghyun Kim3, Juyeol Yun1, Sehee Kim1, Minseok Kang1, Chun-Ho Cho4 and Jung-Hwa Chun5, (1)Seoul National University, Seoul, South Korea, (2)Yonsei University, Seoul, South Korea, (3)Pusan National University, Environmental Engineering, Busan, South Korea, (4)National Institute of Meteorological Research, Seogwipo-si, South Korea, (5)KFRI Korea Forest Research Institute of the Korea Forest Service, Seoul, South Korea
We demonstrate how field measurements can inform the selection of model frameworks in small watershed applications. Based on the assumption that ecohydrological systems are open and complex, we employ the process network analysis to identify the system state and the subsystems architecture with changing environment conditions. Ecohydrological and biogeochemical processes in a watershed can be viewed as a network of processes of a wide range of scales involving various feedback loops and time delay. Using the KoFlux tower-based measurements of energy, water and CO2 flux time series along with those representing the soil-plant-atmospheric continuum; we evaluated statistical measures of characterizing the organization of the information flows in the system. We used Shannon’s information entropy and calculated the mutual information and transfer entropy, following Ruddell and Kumar (2009). Transfer entropy can measure the relative strength and time scale of couplings between the variables. In this analysis, we selected 15 variables associated with ecohydrological processes, which are groundwater table height, water temperature, specific conductivity, soil moisture contents at three depths, ecosystem respiration, gross primary productivity, sensible heat flux, latent heat flux, precipitation, air temperature, vapor pressure deficit, atmospheric pressure, and solar radiation. The data-driven nature of this investigation may shed a light on reconciling model parsimony with equifinality in small watershed applications. (Acknowledgment: This work and the data used in the study were funded by the Korea Meteorological Administration Research and Development Program under Grant Weather Information Service Engine (WISE) project,153-3100-3133-302-350 and Grant CATER 2014-3030, respectively. The KoFlux site was supported by the Long-term Ecological Study and Monitoring of Forest Ecosystem Project of Korea Forest Research Institute.)