River flow forecasting
Download final report: Satellite Water Monitoring and Flow Forecasting System for the Yellow River Basin (6.9 Mb)
Large Scale Hydrological Model (LSHM)
The LSHM has been developed by UNESCO-IHE. The model is forced by the satellite derived rainfall and evapotranspiration fields. It can assimilate weather forecast scenarios and measured flow data from river discharge stations. The LSHM computes a running water balance on every grid point in the terrain, the two-dimensional accumulation of water towards the river network, and the flow at all points of the river network. The system provides fully dynamic spatial information on the actual state of the river basin water resources. It can also run in forced boundary-flux mode, to provide river flows through part of the network. Thus it is possible to use the system for modeling a non-closed hydrological system. The latter operation mode is also used to provide short-range forecasts of river flows. The LSHM consists of two parts, a land surface flow component and a river flow routing component.
Land surface flow
The land surface flow component takes the EWBMS rainfall and evaporation fields as input and accumulates the resulting surface and subsurface flows through the model area along a two-dimensional gradient, which is broadly related to the topography. The terrain is discretized into distinct grid cells, each of which have a storage potential that is dynamically modified depending on input and output fluxes. The running water balance of the grid cells are described by a two-dimensional continuity equation. The diffusivities are parameterized using a non-linear relationship between storage potential, water deficit and antecedent precipitation. This allows the model to respond more rapidly under increasing saturation conditions while the opposite holds for dry conditions, which is more or less similar to the flow mechanism applied in the TOPMODEL concept (Beven, 2001). The model equations are solved by means of a finite difference technique.
Example of flow simulation of the Upper Yellow River for the years 2000 and 2001 on the basis of EWBMS satellite derived rainfall and evapotranspiration data.
The one-dimensional river flow component is based on the Muskingum-Cunge routing method (Cunge 1969) with lateral inflow. This model routes the flow through a discrete channel network from upstream to downstream points over specified time intervals Dt. The coefficients of the model are a function of the Courant number and the Reynolds number and are parameterized following Ponce (1986). The exchange of water between the land and river components is solved through an iterative procedure. The main advantage of the scheme is that it can simulate diffusion wave dominated flow while requiring only a generalized description of cross-section geometry. As such, the scheme is particularly attractive in large-scale river basin modelling for which detailed geometry data are scarce or lacking. Information on water level may be derived from measured stage-discharge (rating curve) relationships.
The following sequence of pictures shows the derivation of the effective precipitation which may serve as input for runoff forecasting for the Niger river. Upper: rainfall, middle: actual evapotranspiration, lower: effective precipitation = rainfall - actual evapotrtanspiration.