Evaluation of heterogeneity statistics as reasonable proxies of the error of precipitation quantile estimation in the Minneapolis-St. Paul region
Highlights
- •
Heterogeneity was tested as a proxy of quantile error for daily rainfall totals.
- •
Monte Carlo estimates were calculated for all possible regions of a gauge network.
- •
Two heterogeneity statistics were found to be reasonable proxies of error.
- •
Previous findings held only one of these to be a reasonable proxy.
Introduction
Rare or extreme precipitation events, which include events classified as natural disasters, have major ecological, economic, and public safety significance. Sample size is often a limiting factor in the estimation of extreme hydrological events; one rule of thumb for flood frequency estimation is that for reliable estimates of a return period T the record length in station-years must exceed (Robson and Reed, 1999). Many statistical hydrologists have followed the index flood method of Dalrymple (1960), in which sample size is increased by grouping gauges, or“sites”, into regions, calculating a regional “growth curve” normalized by an index such as the mean or median of the at-gauge data, and estimating at-site quantiles by multiplying the index flood and the regional growth curve.
Linear moments, analogues of conventional central moments like skewness and kurtosis based on probability weighted moments (Greenwood et al., 1979), are often used in this context. L-moment estimators have lower bias than other common methods of estimation at small sample size (Hosking et al., 1985, Lettenmaier et al., 1987). They are less biased than conventional moment estimators, are not bounded by sample size, and are more robust to outliers. L-moment ratios can be more reliably predicted from a subsample than conventional moment ratios. L-moment ratios, the second through fourth of which are denoted the coefficient of L-variance (L-CV), L-skewness, and L-kurtosis (the first L-moment ratio does not exist), provide greater insight into the underlying distribution of high-skew data than conventional moment ratios. For example, L-moment ratio diagrams are used as decision aides for identifying the underlying distribution of regional data (Hosking, 1990, Vogel and Fennessey, 1993, Hosking and Wallis, 1997, Zafirakou-Koulouris et al., 1998).
L-moment analysis of regions formed according to hydrological characteristics has been conducted in recent decades on streamflow (Vogel et al., 1993, Ouarda et al., 2008, Noto and Loggia, 2009) and precipitation data (Guttman et al., 1993, Werick et al., 1994, Adamowski et al., 1996, Alila, 1999, Smithers and Schulze, 2001, Kyselý et al., 2007, Modarres and Sarhadi, 2011). L-moment ratios for daily data series using “wet-day” (non-zero only) and full datasets have been evaluated across the United States (Hanson and Vogel, 2008). Regional frequency analysis models using fuzzy regions (Jingyi and Hall, 2004, Rao and Srinivas, 2006) and fractional-membership regions of influence (Burn, 1990, Zrinji and Burn, 1994, Gaál et al., 2008) represent alternatives to the strict regional membership model.
The regional pooling mechanism of the index flood method involves the assumption of homogeneity across the sites in a candidate region - at-site differences in L-moment ratios are assumed to be due solely to sampling variability. The degree to which the homogeneity assumption is violated is therefore likely to be related to quantile error. Statistics quantifying the heterogeneity of a region based on Monte Carlo simulation have been proposed which sample from a Generalized extreme-value distribution (Lu and Stedinger, 1992, Alila, 1999).
Hosking and Wallis (1997) define three statistics based on the between-site variation of L-moment ratios, , and . All three H statistics fit the flexible four-parameter Kappa distribution with the average L-moment ratios of the region in question and use Monte Carlo simulation to generate simulated regions from the Kappa. For the real region and for each simulated region a statistic called , or is calculated using the sum of the squared difference between each site’s L-moment ratio values and the regional average. uses only the L-CV, incorporates L-CV and L-skewness, and incorporates L-skewness and L-kurtosis. is calculated when for the real region minus the mean of for simulated regions is divided by the standard deviation of simulated regions’ and are calculated analogously (see Eqs. (6), (7), (8), (9)).
A simulation study is used in Hosking and Wallis (1997) to reject and and to accept . Heterogeneous regions’ RMSEs are divided by their equivalent homogeneous region’s RMSE. This isolates the RMSE increase due to heterogeneity. is shown to have a linear relationship with percent RMSE added due to heterogeneity. Results for and are not reported.
Hosking and Wallis (1997) define thresholds below which regions can be considered “possibly” and “definitely” heterogeneous with reference to a range of percent RMSE added values implied by the -percent RMSE added relationship. is found to indicate a 20–40% increase in RMSE, while is associated with 40–80% increases. These thresholds are also applied to and , which are found to rarely exceed them.
Viglione et al. (2007) investigate and as well as two nonparametric heterogeneity statistics by measuring the fraction of simulated regions that are correctly and incorrectly identified as heterogeneous. The threshold of is used for both and . They confirm the utility of for simulated data with L-skewness below 0.23 and reject . The bootstrap Anderson–Darling test is found to be more powerful than either statistic for data with higher skewness.
Two approaches are used in this study to quantify the power of , and as proxies of error due to heterogeneity. The original Hosking and Wallis (1997) simulation study is recapitulated and results for and are presented alongside those for . Thresholds for are found using its linear relationship to quantile error, not ’s. An enumeration study is also conducted, estimating , and quantile error for all possible regionalizations of a small daily precipitation gauge dataset. Components of error unrelated to heterogeneity are preserved in this study, allowing the heterogeneity statistics’ relationships with total estimated quantile error to be compared to the ideal case represented in the simulation experiment.
The remainder of this paper is structured as follows. Daily precipitation gauge data from which all possible regions are to be enumerated are presented in the following section, Section 2. Section 3 introduces the equations and methods used to calculate linear moments of the data, estimate regional heterogeneity, assign a regional distribution, estimate the RMSE of regional quantile estimates, and perform a simulation experiment analogous to that presented in Hosking and Wallis (1997). Section 4 presents the results first of the simulation experiment, then of the enumeration experiment, which compares heterogeneity and error estimates across all possible regions formed from the selected precipitation gauges. Section 5 discusses the results; Section 6 summarizes the paper and offers conclusions and potential avenues of future research.
Section snippets
Data
Mean annual precipitation in Minnesota ranges from the low teens to above 30 in., with the mean annual precipitation generally increasing from the northwest to the southeast. Moist air carried from the Gulf of Mexico is an important source of precipitation in Minnesota. Almost half of yearly precipitation occurs in June, July, and August (Baker et al., 1967). Rainfall quantiles have been estimated for Minnesota using the Generalized extreme-value distribution on the annual maximum series (
Linear moments
The Hosking (1990) approach begins from the probability weighted moments (PWM) of Greenwood et al. (1979), defined in Eq. (1):where is the cumulative distribution function (cdf) of is the inverse cdf or quantile function of X for probability F, and is the rth-order PWM ( is equal to the mean ). Hosking (1990) defines the L-moments in Eq. (2):with calculated according to Eq. (3):
is the mean of a
Results of simulation experiment
The simulation experiment described in Section 3.5 outputted similar results to those described in Table 4.1 of Hosking and Wallis (1997) for . Averages of the H statistics are taken across 100 simulations for each set of L-moment ratios. In Fig. 3, depicting the linear relationship between each H statistic and percent RMSE added due to heterogeneity for a non-exceedance frequency of 0.01, results for are similar to Fig. 4.2 in Hosking and Wallis (1997). Results for and are not
Discussion
In the simulation experiment the performance gap between and is relatively narrow, indicating that L-skewness offers a useful amount of heterogeneity information in the presence of L-CV variation. is consistently a slightly less faithful proxy for error than across the simulated regions, but like it can also be plotted against percent RMSE added due to heterogeneity and threshold values can be described as equivalent to a range of percent RMSE added. Analogously to the process
Summary and conclusions
Hosking and Wallis (1997) find that has power as a proxy of error while and do not, but simulation and enumeration studies conducted here paint a more nuanced picture. remains the favored heterogeneity statistic across simulated and real-world datasets across a wide range of skewness, but is nearly as effective. The efficacy of has been obscured through the application of thresholds constructed with reference to the linear relationship between and percent RMSE added, which
Acknowledgements
Data were provided by the State Climatology Office, Minnesota Department of Natural Resources–Division of Ecological and Water Resources. This work also used the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by National Science Foundation Grant No. OCI-1053575. Financial support provided by the George Mason University Presidential Scholarship is gratefully acknowledged. The authors also wish to thank Jason Giovannettone for his assistance.
References (34)
- et al.
Regional rainfall distribution for Canada
Atmospheric Research
(1996) - et al.
Using a high-density rain gauge network to estimate extreme rainfall frequencies in Minnesota
Appl. Geograp.
(2011) - et al.
Regional flood frequency analysis for the Gan-Ming River basin in China
J. Hydrol.
(2004) - et al.
Sampling variance of normalized GEV/PWM quantile estimators and a regional homogeneity test
J. Hydrol.
(1992) - et al.
Statistically-based regionalization of rainfall climates of Iran
Global Planet. Change
(2011) - et al.
Intercomparison of regional flood frequency estimation methods at ungauged sites for a Mexican case study
J. Hydrol.
(2008) - et al.
Regionalization of watersheds by fuzzy cluster analysis
J. Hydrol.
(2006) - et al.
A methodology for the estimation of short duration design storms in South Africa using a regional approach based on L-moments
J. Hydrol.
(2001) - et al.
Flood frequency analysis for ungauged sites using a region of influence approach
J. Hydrol.
(1994) A hierarchical aproach for the regionalization of precipitation annual maxima in Canada
Journal of Geophysical Research
(1999)