Log-Exponential Reservoir Operating Rules for Global And Regional Hydrological Modeling


Many hydrological models simulate both runoff (Water Balance Model) and discharge (Water Routing) over given river networks (STN, DRT, HydroSHEDS, etc.). But water infrastructure development (dams, inter-basin water transfer lines, irrigation canal networks, etc.) in industrial and post-industrial time frames impose real challenges to the modeling of water routing and prediction of river discharge, especially for large-scale regional and global geographic extents where detailed information about operating rules for such hydro-infrastructure units often do not exist. The global and regional river dam databases used in water routing simulations (e.g. GRanD and NID) provide some limited information on dam construction dates and purposes (e.g. hydropower, irrigation, water supply, flood control, etc.), but do not indicate how these are being operated over the given hydrological year cycle and over extreme low/high in-flow regimes. So the formulation of generic and use-specific reservoir operating rules for regional and global hydrologic simulations are still debated issues for the hydrology modeling community. In our network independent WBM-TrANS model (Water Balance Model-Transport from Anthropogenic and Natural Systems) we have formulated and tested a new Log-Exponential OPerAting Rule for Dams (LEOPARD) that can be readily parameterized for a generic and/or specific dam purpose. The key features of the LEOPARD formulation include a combination of adjustable logarithmic and exponential functions describing the release of water from the reservoirs and other adjustable parameters for minimum storage and two exponent curvature coefficients (one each for logarithmic and exponential functions). In the LEOPARD model the dam discharge/release calculations are normalized to Average Annual Discharge (AAD), which, in turn, is taken as a running average of the past 5 years. The latter is critical to simulate dam fill-up periods and shifts in the hydrological cycle over long-term climate variability (e.g. climate change and direct human modification). Here we present testing of the LEOPARD rules in both regional and global model runs. Results indicate a very good match of observed vs. modeled reservoir release flows for a number of large, medium, and small relative-capacity dams. We also found that LEOPARD produces satisfactory results for dam fill-up periods following construction, which can take several years for large dams. Dam removal flow can also be simulated by linear reduction of reservoir capacity to zero over the residence time of the reservoir.

Publication Date


Journal Title

Fall Meeting, American Geophysical Union (AGU)


American Geophysical Union Publications

Document Type

Conference Proceeding