|
|
Conclusions
The modeling system LMF with a part of the parameters from the application to the Patuxent watershed cannot is not transferable to the Parthe watershed. However, there are several points how the simulation of the hydrology in the Parthe watershed can be improved:
Evapotranspiration: To apply the model to the Parthe watershed a new parameterization of evapotranspiration is needed. As changes in simulated evapotranspiration will also influence the simulated runoff and water balance, evapotranspiration should be calibrated before other changes. The evaporation of surface water from the stream probably contributes only to a small part to total evaporation because the Parthe River is rather small and never takes a big part of the area of a cell. Therefor surface water evaporation from stream might be neglected in the application of the model to the Parthe watershed.
Groundwater surface: The simulated groundwater surface differs several meters from measurment values. This can cause difficulties when the groundwater depth is used as an input variable, e.g. the the water availibility for plants is calculated dependent on the groundwater depth. The simulation of the groundwater surface is difficult to improve because a more complex model would also require more input and calibration data.
Runoff: The peaks from snowmelt will be better reproduced if a soil freezing module is introduced. For a further investigation of the seasonal over- and underestimation sensitivity analysises of the soil hydraulic parameters could be conducted and timecourses of groundwater heads ands and soil moisture could be evaluated. The model will probably perform better, when the vertical hydraulic conductivity of the soil is calibrated.
Water balance: In order to simulate the water balance satisfactory it could be another calibration criteria. E.g. evaporation and or subsurface runoff could be calibrated to the water balance.
Because of the long compter times the model could merely be calibrated manually, which is time consuming and subjective. Besides, parameter interactions cannot be systematically investigated. The long calculation time is a general disadvantage of LMF and results the high number of cells and a high level of model complexity within each cell within a grid-based approach. |
|