Regionalisation of Maximum Annual Runoff Using Hierarchical and Trellis Methods with Topographic Information
2013
Chérif, Rim | Bargaoui, Zoubeida
Regional frequency approaches are frequently proposed in order to estimate runoff quantiles for non-gauged catchments. Partitioning methods such as cluster analysis are often applied in order to regionalize catchments using topography, soil and hydroclimatological characteristics. This paper aims to construct a mean regional frequency curve for annual maximum runoffs, using topographic descriptors for cluster analysis. Both trellis and hierarchical classifications partitioning methods are performed using basin area; basin perimeter; characteristic length; global slope index; compaction index; specific gradient slope and geodesic coordinates as attributes. To build the distance measures, various multidimensional spaces are considered with pairs or triplets of attributes. Resulting clusters were checked for hydrological homogeneity using the test of Hosking and Wallis based on L-moments estimates. A sample of 40 Tunisian gauged basins covering a range of areas from 56 to 16483 km² has been considered to achieve these purposes. The classification and the test of Hosking and Wallis concluded for separating the gauged basins in two hydrological homogeneous regions. Also, the basin global slope index is found as the main discriminating classification factor. Further, regional quantiles of the standardized maximum annual flood (index flood) were estimated using GEV distribution. The two regional curves are distinguishable for extremes events, suggesting that the second region with high slope index displays more variability in the extremes. However, comparisons of RMSE results using two regions against one single pooled region suggest that estimation of standardized quantiles is more accurate in the case of one single region for non extreme events.
Показать больше [+] Меньше [-]Ключевые слова АГРОВОК
Библиографическая информация
Эту запись предоставил National Agricultural Library