Weighted least squares method-improved estimates when the index flood method is used with annual maximum floods

The index flood method is one approach in flood frequency analysis to obtain flood estimates at a specific site, in which, the at-site index flood is used along with a regional growth curve. The index flood, in this case the mean annual flood, Q̄, is obtained from observed records of streamflow at a nearby gauging station. However, many instances arise where such data are not available. In such ungauged drainage basins, a regression relationship between the mean annual flood and numerically measured drainage basin characteristics such as area, slope, rainfall, etc. is found. A major area of concern in regression relationships has been the presence of heteroscedasticity. This study shows clearly that in such circumstances, based on results from three observed data sets, the use of the weighted least squares method (WLS) with diagonal matrices having their terms equal to inverse coefficient of variation Cv, is superior to the use of the ordinary least squares method (OLS).