A. D. Christofides, Ανάπτυξη μοντέλου βροχής απορροής σε Σύστημα Γεωγραφικής Πληροφορίας, Postgraduate Thesis, 112 pages, NTUA, Athens, June 2008.
In this study is developed a new distributed physically-based rainfall-runoff parameter model, which predicts direct runoff hydrographs. The overall goal of this study is to develop a method of estimating the values of peak flow rate and time to peak, and to create a unit hydrograf, based exclusively on the GIS. The basic modeling approach was to use the raster GIS functions of ArcGIS to calculate the travel time from each point in the watershed to the outlet by determining the flow path and the travel time through each cell along the path. The travel time through each individual cell along the flow path was summed to estimate the cumulative travel time to the outlet. The model accounts for differences in overland and channel velocity, slope, and land use. Runoff is routed over the elevation surface. The total travel time to the outlet from each grid cell is estimated based on the runoff pathway and the travel time through each grid cell along the path. The model uses the time-area curve method to calculate synthetic unit hydrograph. Once the cumulative travel time to the outlet is determined for each grid cell, the watershed is divided into 1-hour isochrones. After the watershed is divided into isochrones, a time-area histogram is developed. Then, a unit hydrograph is developed. The unit hydrograph ordinates at each time interval is the incremental area for that time step divided by the time interval. The flow velocity is divided into overland flow velocity and channel flow velocity. The runoff velocity for areas with overland flow can be estimated by using a kinematic wave approximation combined with Manning equation. This calculation is based on the digital elevation model, the landuse and the calibration process. The land use data for the watershed was used to estimate the Manning’s roughness coefficient for the overland cells. In other hand, the channels in the watershed were assumed to be trapezoidal with 1.75:1 side slopes and 2m bottom. An equation was developed to calculate the mean channel flow rate for each cell and another one was developed to estimate the equilibrium depth of flow. The channel flow is calculated using the flow rate and the equilibrium depth of flow for each cell, as well as the geometry of the channel. The model was applied to the sub-basin Areti (bridge Areti) of the basin of river Kalamas. The layer’s resolution was 50 m. This means that the cell’s size was 250 m2. The land use data for the basin Areti was used to estimate the Manning’s roughness coefficient for the overland cells. Due to the fact that many published tables of Manning’s roughness coefficients do not include most of the specific land uses found in basin Areti, a number of assumptions were made to estimate values. The estimated synthetic unit hydrograph was used to create synthetic hydrographs for 4 rainfall episodes. The unit hydrograph combined with the exess rainfall for each episode, which was estimated using the SCS method. The peak flow rate from the synthetic hydrographs was compared with the observed hydrographs, using mainly the relative error and the mean arithmetic relative error (MARE). According to the MARE for the total of the episodes, the model predicted the peak flow rate with grate accuracy (MARE = 0,085). However, the time to peak was predicted successfully only for the half of the episodes. At the end, the sensitivity analysis concluded that the length of the stream network and the intensity of the rainfall (which was used to calculate the overland flow) are the most important parameters for a successful calibration. In addition, a better evaluation for the new model needs more rainfall – runoff data from a watershed which have to be more accurate, too.
Full text (5169 KB)