الفهرس 
Cyclonic (cyclonic precipitation)Only 14 pages are availabe for public view
 |  | from | 217 |  |  |
| | | from | 217 | | |
Abstract
In hydrological modeling, there are several sources of uncertainties associated with runoff estimation. Rainfall, which is the main input for these models, has a large impact on the runoff uncertainties. The objective of this research is to develop a monitoring rainfall network design method to reduce uncertainties associated with runoff estimation by reducing the uncertainty in rainfall data. This method depends on measuring the variations of the mean areal precipitation over any watershed of interest. Two sources of rainfall data were used; data recorded by existing rain gauges networks and data retrieved from GSMaP satellite product. This work includes two main parts; the assessment of reducing the number of rain gauges of predefined locations and developing the method for designing monitoring rainfall network. The differences between deterministic and geostatistical interpolation methods were investigated. Various statistical performance measures were used for achieving this assessment and selecting the optimal group(s) of rain gauges. Python scripts were developed and then used wherever needed throughout the work to complete the computations and to develop the method. Results reveal that the flexibility of the developed method is suitable for application in any ungauged watershed of interest and for any number of rain gauges according to the budget. The optimal rain gauge network design was achieved for various rain gauges number. Results also illustrate that the interpolation method does not affect the estimation of the mean areal precipitation for catchments having small areas. Additionally, both interpolation methods give similar results when the number of rain gauges increases over the watershed. The critical value of error increases by decreasing the probability of exceedance. By increasing the number of rain gauges in the investigated scenario, the values of resultant errors significantly decrease.