Estimating river low flows statistics in ungauged sites

Knowledge of low flow events frequency is required for water resources planning and management and their definition is necessary for several purposes, including water supply planning and irrigation systems running and moreover to maintain amount and quality of water. Low flow regime is tightly dependent on the catchment hydrogeological feature and a detailed surface and groundwater catchment analysis is necessary for an accurate characterization. However on a practical perspective, although scientifically proven, statistical analysis is widely apply to derive indices to characterize low flow regimes and as a measure for environmental minimum flow. With these indices it is possible to identify the occurrence, the extent and magnitude of the hydrological droughts, the ones that affect mainly water supply systems. Low flow indices are commonly evaluated at gauged sites from observed streamflow time series. Their reliability can be affected by the lack of observed streamflow data, a diffuse problem in the real world. In order to overcome these problems and to estimate low flow statistics in ungauged sites it is possible to refer to a regional frequency analysis, widely used since long time and in different disciplines. It consists in inferring data in ungauged stations using hydrological and statistical methods applied over a more or less wide area, a region. The methods employ catchment and climatic characteristics, supposed to be measured over the investigated area, as independent variables, and data from other catchments where stream flow data are recorded. The analysis of low flow indices is carried out on the flow data of 65 consistent hydrometric stations located in Tuscany region, in Central Italy, recorded from 1949 to 2008. The area is subdivided into different regions using the L-moments method applied to indices derived from the flow duration curve (Q70 annual series), to the 7-day annual minimum (Q7,T) series and to the annual SQI, Standardized Discharge Index. The division into subregions is validated using discordancy and heterogeneity tests. Several subdivisions are tested, starting from previous studies on different hydrological extreme values and introducing some hydrological features. For every river section of interest the catchment area is identified and an appropriate set of catchment physiographic and climatic characteristics is defined. A physiographical space-based method is used to relate the duration and annual minimum indices of low flow to the rivers basins characteristics. The new space is built as a power correlation of the catchment geomorphologic and climatic characteristics. In this space several interpolation techniques, either deterministic or geostatistical, such as Inverse Distance, Thiessen polygon methods and Kriging, are applied. The results are valuated using the jack-knife method. Different error measurement (mean square error, mean relative error…) are also assessed to compare the results, to quantify the accuracy of the different techniques and to define the most suitable procedure for low flow regionalization.