The application of data mining techniques for the regionalisation of hydrological variables

Abstract

Flood quantile estimation for ungauged catchment areas continues to be a routine problem faced by the practising Engineering Hydrologist, yet the hydrometric networks in many countries are reducing rather than expanding. The result is an increasing reliance on methods for regionalising hydrological variables. Among the most widely applied techniques is the Method of Residuals, an iterative method of classifying catchment areas by their geographical proximity based upon the application of Multiple Linear Regression Analysis (MLRA). Alternative classification techniques, such as cluster analysis, have also been applied but not on a routine basis. However, hydrological regionalisation can also be regarded as a problem in data mining -a search for useful knowledge and models embedded within large data sets. In particular, Artificial Neural Networks (ANNs) can be applied both to classify catchments according to their geomorphological and climatic characteristics and to relate flow quantiles to those characteristics. This approach has been applied to three data sets from the south-west of England and Wales; to England, Wales and Scotland (EWS); and to the islands of Java and Sumatra in Indonesia. The results demonstrated that hydrologically plausible clusters can be obtained under contrasting conditions of climate. The four classes of catchment found in the EWS data set were found to be compatible with the three classes identified in the earlier study of a smaller data set from south-west England and Wales. Relationships for the parameters of the at-site distribution of annual floods can be developed that are superior to those based upon MLRA in terms of root mean square errors of validation data sets. Indeed, the results from Java and Sumatra demonstrate a clear advantage in reduced root mean square error of the dependent flow variable through recognising the presence of three classes of catchment. Wider evaluation of this methodology is recommended.

Estimation of flood warning runoff thresholds in ungauged basins with asymmetric error functions

“…Successful applications of ANNs for the estimation of index floods or flood quantiles at ungauged sites are reported in Muttiah et al (1997), Hall et al (2002), Dawson et al (2006), Shu and Burn (2004), Shu and Ouarda (2008), Singh et al (2010), Simor et al (2012) and Aziz et al (2013).…”

“…In the recent years, SOMs were also successfully applied for catchments classification either based on geo-morphoclimatic descriptors (Hall and Minns, 1999;Hall et al, 2002;Srinivas et al, 2008;Di Prinzio et al, 2011) or based on hydrological signatures (Chang et al, 2008;Ley et al, 2011;Toth, 2013); however, it is important to underline that the clustering is not carried out here in order to identify a pooling group of similar catchments for developing a region- specific model, but for the optimal division of the available data for the parameterisation and independent testing of a single model to be applied over the entire study area.…”

Catchment classification based on characterisation of streamflow and precipitation time series

“…In the recent years, also non-supervised neural networks, and in particular of the SOM (self-organising mapping) type, were successfully applied (and sometimes compared with other methods such as K-means or Fuzzy C-means) for catchments classification purposes (Hall and Minns, 1999;Hall et al, 2002;Jingyi and Hall, 2004;Chang et al, 2008;Srinivas et al, 2008;Di Prinzio et al, 2011;Ley et al, 2011). SOM-type neural networks learn to cluster the input data by recognizing different patterns organising the data on the basis of their similarity, quantified by means of a distance measure (in the present case, like in the majority of applications, the Euclidean distance).…”

Regional flood frequency analysis using kernel‐based fuzzy clustering approach

“…Traditionally hard clusters were formed for RFFA by hydrologists using algorithms such as partitional [e.g., Wiltshire, 1986;Burn, 1989;Bhaskar and O'Connor, 1989;Burn and Goel, 2000], hierarchical [e.g., Mosley, 1981;Tasker, 1982;Nathan and McMahon, 1990;Burn et al, 1997], hybrid of partitional and hierarchical clustering [Hosking and Wallis, 1997;Rao and Srinivas, 2006a], and self-organizing feature maps (SOFMs) [e.g., Hall and Minns, 1999;Hall et al, 2002;Jingyi and Hall, 2004]. In a few studies [e.g., Bargaoui et al, 1998;Hall and Minns, 1999;Jingyi and Hall, 2004;Rao and Srinivas, 2006b], fuzzy clusters were formed.…”