Challenges in application of statistical techniques in hydrology: a case study on regional flood modelling in southeast Australia

Regional flood frequency analysis (RFFA) is the most widely used statistical technique to estimate design floods in an ungauged location using flood data recorded at nearby stream gauging stations. The application of statistical techniques in hydrology, particularly in RFFA modelling is challenging in Australia due to a higher degree of non-homogeneity. Statistical tools and techniques involve planning, designing, sampling, data collection, analysing (descriptive and inferential), interpretation, and reporting to conduct statistical hydrologic research. Besides these there are lots of statistical and mathematical terms, concepts, and formulas, which are arithmetically complex and their physical interpretation in relation to rainfall-runoff process is relatively difficult. Hence, students who are not good at maths and hydrology face challenges to overcome these difficulties. Generally exploratory data analysis and visualisation, generalising the study results, scaling up the achieved results, and merging the findings in software engineering are tricky in statistical hydrology. In southeast Australia, the use of statistical techniques in RFFA is challenging, which might be due to highly variability and non-stationarity in hydrological data. This paper presents a case study using data from 88 catchments in New South Wales (NSW) where regression-based methods are used to develop prediction equations that can be applied to ungauged catchments in NSW to estimate design floods. This paper also highlights the learning aspects of statistical hydrology by the first author, which was based on the student-centred learning approach.