Performance-based comparison of regionalization methods to improve the at-site estimates of daily precipitation

In this article, we compare the performances of three regionalization approaches in improving the at-site estimates of daily precipitation. The first method is built on the idea of conventional RFA (Regional Frequency Analysis) but is based on a fast algorithm that defines distinct homogeneous regions relying on their upper tail similarity. It uses only the precipitation data at hand without the need for any additional covariate. The second is based on the region-of-influence (ROI) approach in which neighborhoods, containing similar sites, are defined for each station. The third is a spatial method that adopts Generalized Additive Model (GAM) forms for the model parameters. In line with our goal of modeling the whole range of positive precipitation, the chosen marginal distribution model is the Extended Generalized Pareto Distribution (EGPD) on which we apply the three methods. We consider a dense network composed of 1176 daily stations located within Switzerland and in neighboring countries. We compute different criteria to assess the models' performances both in the bulk of the distribution as well as in the upper tail. The results show that all the regional methods offered improved robustness over the local EGPD model. While the GAM method is more robust and reliable in the upper tail, the ROI method is better in the bulk of the distribution.

Regional frequency analysis of daily maximum rainfall in Haryana

This paper is sets-out for the regional frequency analysis of daily maximum rainfall from the 27 rain gauge stations in Haryana using L-moments. As the distribution of rainfall varies spatially in Haryana, the 27 rain gauge stations are grouped into three clusters namely, cluster C1, C2 and C3 using Ward’s clustering method and homogeneity of clusters was confirmed using L-moments-based Heterogeneity measure (H). Using goodness-of-fit measure ( DIST Z ) and L-moment ratios diagram, suitable regional frequency distributions were selected among five candidate distributions;Generalized Logistic (GLO), Generalized Extreme Value (GEV),Generalized Normal (GNO), Generalized Pareto (GPA), and Pearson Type-3 (PE3) for each cluster. Results showed that PE3 and GNO were good fitted regional distribution for the cluster C1 and GEV, PE3 and GNO fitted for cluster C2 while for cluster C3; GLO and GEV were good fitted regional distribution. To select a robust distribution among good fitted distributions accuracy measures calculated using Monte Carlo simulations for each cluster. The simulation result showed that PE3 was the best choice for quantile estimation for cluster C1. For cluster C2, PE3 was the best choicefor a large return period and GEV was best for a small return period. For cluster C3, GEV was the most suitable distribution for quantile estimation. Using these robust distributions rainfall quantiles were estimated at each rain gauge station from 2 to 100 year return periods. These estimated rainfall quantiles may be rough guideline for planning and designing hydraulic structures by policy makers and structural engineers.

Uncertainty Analysis of Regional Rainfall Frequency Estimates in Northeast India

Estimation of rainfall quantile is an important step in regional frequency analysis for planning and design of any water resources project. Related evaluations of accuracy and uncertainty help to further assist in enhancing the reliability of design estimates. In this study, therefore, we investigate the accuracy and uncertainty of regional frequency analysis of extreme rainfall computed from genetic algorithm-based clustering. Uncertainty assessment is explored with prediction of quantiles with a new spatial Information Transfer Index (ITI) and Monte Carlo simulation framework. And, accuracy assessment is done with the comparison of regional growth curves to at-site analysis for each homogenous region. Further, uncertainty assessment with the ITI method is compared with Maximum Likelihood estimation (MLE) optimized by a genetic algorithm (GA) to check the suitability of the method. Results obtained suggest the ITI-based uncertainty assessment for regional estimates outperformed those of at-site estimates. The MLE-GA method based on at-site estimates was found to be better than at-site estimates based on L-moments, suggesting the former as a better alternative to compare with regional frequency estimates. Moreover, minimal bias and least deviation of the regional growth curve were obtained in the rainfall regions. The confidence intervals of regional estimates were seen to be well within the bounds of normality assumptions. Doi: 10.28991/cej-2021-03091762 Full Text: PDF

Regional frequency analysis of daily maximum rainfall in Haryana

This paper is sets-out for the regional frequency analysis of daily maximum rainfall from the 27 rain gauge stations in Haryana using L-moments. As the distribution of rainfall varies spatially in Haryana, the 27 rain gauge stations are grouped into three clusters namely, cluster C1, C2 and C3 using Ward’s clustering method and homogeneity of clusters was confirmed using L-moments-based Heterogeneity measure (H). Using goodness-of-fit measure ( ) and L-moment ratios diagram, suitable regional frequency distributions were selected among five candidate distributions; Generalized Logistic (GLO), Generalized Extreme Value (GEV),Generalized Normal (GNO), Generalized Pareto (GPA), and Pearson Type-3 (PE3) for each cluster. Results showed that PE3 and GNO were good fitted regional distribution for the cluster C1 and GEV, PE3 and GNO fitted for cluster C2 while for cluster C3; GLO and GEV were good fitted regional distribution. To select a robust distribution among good fitted distributions accuracy measures calculated using Monte Carlo simulations for each cluster. The simulation result showed that PE3 was the best choice for quantile estimation for cluster C1. For cluster C2, PE3 was the best choicefor a large return period and GEV was best for a small return period. For cluster C3, GEV was the most suitable distribution for quantile estimation. Using these robust distributions rainfall quantiles were estimated at each rain gauge station from 2 to 100 year return periods. These estimated rainfall quantiles may be rough guideline for planning and designing hydraulic structures by policy makers and structural engineers.

Predicting Future Ranked Statistics and Recorded Values for Some Statistical Distributions

Regional frequency analysis (AFR) brings together a variety of statistical methods aimed at predicting the behavior of extreme hydrological variables at ungauged sites. Regression techniques, geostatistical methods and classification are among the statistical tools frequently encountered in the literature. Methodologies based on these tools lead to regional models that offer a simple, but very useful description of the relationship between extreme hydrological variables and physiometeorological characteristics of a site. These regional models then make it possible to predict the behavior of variables of interest at places where no hydrological information is available. These methods are generally based on restrictive theoretical assumptions, including linearity and normality. These do not reflect the reality of natural phenomena. The general objectives of this paper are to identify the methods affected by these hypotheses, evaluate their impacts and propose improvements aimed at obtaining more realistic and fairer representations. Projection pursuit regression is a non-parametric method similar to generalized additive models and artificial neural networks that are considered in AFR to take into account the non-linearity of hydrological processes. In a comparative study, this paper shows that regression with revealing directions makes it possible to obtain more parsimonious models while preserving the same predictive power as the other nonparametric methods. Canonical Correlation Analysis (ACC) is used to create neighborhoods within which a model (e.g. multiple regression) is used to predict hydrologic variables at ungagged sites on the other hand, ACC strongly depends on the assumptions of normality and linearity. A new methodology for delineating neighborhoods is proposed in this paper and uses revealing direction regression to predict a reference point representing hydrological and physiometeorological information that is relevant to these groupings. The results show that the new methodology generalizes that of ACC, improves the homogeneity of neighborhoods and leads to better performance. In AFR, kriging techniques on transformed spaces are suggested in order to predict extreme hydrological variables. However, a transformation is required so that the hydrological variables of interest derive approximately from a multidimensional normal distribution. This transformation introduces a bias and leads to suboptimal predictions. Solutions have been proposed, but have not been tested in AFR. This paper proposes the approach of spatial copulas and shows that this approach provides satisfactory solutions to the problems encountered with kriging techniques. Max-stable processes are a theoretical formalization of spatial extremes and correspond to a more faithful representation of hydrological processes on the other hand; their characterization of extreme dependence poses technical problems which slow down their adoption. In this paper, the approximate Bayesian calculus is examined as a solution. The results of a simulation study show that the approximate Bayesian computation is superior to the standard approach of compound likelihood. In addition, this approach is more appropriate in order to take into account specification errors.

Exploring frequency analysis alternatives on instantaneous peak flow, in the context of flood plain delineation in Southern Québec, Canada.

A flood frequency analysis is conducted using instantaneous peak flow data over a hydrologic sub-region of southern Québec following three distinct methodological frameworks. First, the analysis is conducted locally using available instantaneous peak flow data. Second, the analysis is conducted locally using daily peak flow data processed in order to consider the peak flow effect. Third, a regional frequency analysis is conducted pooling all available instantaneous peak flow data over the study area. Results reveal a notable diversity in the resulting recurrence peak flow estimates and related uncertainties from one analysis to another. Expert judgement appears essential to arbitrate which alternative should be operated considering a specific context of application for flood plain delineation. Pros and cons for each approach are discussed. We finally encourage the use of a diversity of approaches in order to provide a robust assessment of uncertainty affecting peak flow estimates.