Rainfall input

For rainfall input the developed model uses a catchment-wide grid of rainfall intensities [mm hr-1]. The spatial resolution of the grid as well as the timestep can be freely chosen and adopted to the available data. Inside a GIS ARC/INFO framework this information may be gained from the interpolation of a dense network of recording raingauges or, as in the present application, from calibrated rainfall radar.

Runoff generation

According to hydrologically relevant surface characteristics the catchment is spatially disaggregated into different terrain types. Mapping may be governed by airphoto analysis together with thorough ground truthing. On a grid base overland flow generation accounts for the two dominant processes in arid and semi-arid zones: Hortonian and saturation excess runoff. On each terrain type runoff generation is parameterised independently using the same time step as the rainfall input. The soil is conceptualised as a soil storage [mm] which emptied by constant rates of evaporation [mm/h] (occurring only during times of no rain) and deep infiltration [mm/h]. The filling of the soil storage is calculated according to a infiltration function which drops down from an initial value [mm/h] to a final value [mm/h] with a predefined temporal decay. Rather than following a mathematical function, this decay is directly derived from rainfall simulator experiments, as are the values for initial loss, initial/final infiltration rate and deep infiltration. Once the soil storage is entirely filled, saturation excess runoff takes place, Hortonian runoff is calculated as rain that falls in excess of the infiltration function. The only remaining calibration parameter in the volume of the soil storage which corresponds to pre-wetting of the terrain when the model is run in a single-event mode.

Runoff concentration

Using a constant space increment the channel network is divided into segments which are adjoined by small tributary catchments (model elements) delimited by topography. The information on amount of runoff generated in every time step is aggregated onto these separate model elements. Lateral runoff delivery from the model elements to the channel segments follows after a hydrologic timelag [min] which accounts for slope length and gradient.

Channel flow and transmission losses

Inside the channel network the MVPMC4-method [Ponce and Chaganti, 1994] of the Muskingum-Cunge technique is used for streamflow routing. The channels are represented by a compound section consisting of inner channels and bars. At the beginning of a flow only the percentage of the cross section occupied by inner channels is covered by the flood. Then, at the bankfull stage, the complete cross section (including inner channels and all bars) is inundated. In between the width is linearly interpolated. Parameters to be determined for each channel segment are: channel length [m], percentage occupied by inner channels, channel width [m], bankfull stage [m] and Manning n. To account for transmission losses different constant infiltration rates [mm/h] are assigned to inner channels and bars, based on field experiments. After the depth of the active alluvium [m] is saturated the subtraction of losses is stopped. Additionally surface reservoirs of given capacities can be introduced into the channel network.

Application of the ZIN-model

The ZIN-model in its present form only accounts for direct runoff components and has only been applied to simulate single events. Applications include: Nahal Zin (1400 km2, arid Negev) Nahal Yael (0.5 km2, hyper-arid, southern Negev, Israel) Löchernbach-Catchment (1.8 km2 with direct runoff from paved roads, southern Germany) Glasbach-Catchment (0.17 km2, urban, southern Germany) City of Ramallah (0.3 – 0.9 km2, urban, West Bank) Wadi Natuf (250 km2, West Bank)

Coupling of TRAIN and ZIN

Design by noname designs.