A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Discriminatory Power of Heterogeneity Statistics with Respect to Error of Precipitation Quantile Estimation
AbstractAt low sample size, sampling error may be reduced by pooling multiple gauge records. This creates an error component due to heterogeneity, the degree to which the pooled regional data’s quantile estimates are different from the true at-site quantiles. Heterogeneity statistics attempt to quantify the degree to which error is added due to regional heterogeneity. They are justified through elucidation of a so-called reasonable proxy relationship with error caused by heterogeneity and through the ability of heterogeneity thresholds to detect heterogeneous regions. In this paper, previous findings regarding three heterogeneity statistics H1–H3 are revisited; a previous finding that H1 is superior to H2 and H3 is amended based on simulation experiments and upon enumeration of all possible regionalizations of a small gauge dataset across time scales from daily to monthly. Thresholds defined based on H1 are shown to be 4× too high for application to H2 and new thresholds are derived for H2. Two nonparametric heterogeneity statistics are tested and found to achieve only the unsatisfactory performance level of H3.
Discriminatory Power of Heterogeneity Statistics with Respect to Error of Precipitation Quantile Estimation
AbstractAt low sample size, sampling error may be reduced by pooling multiple gauge records. This creates an error component due to heterogeneity, the degree to which the pooled regional data’s quantile estimates are different from the true at-site quantiles. Heterogeneity statistics attempt to quantify the degree to which error is added due to regional heterogeneity. They are justified through elucidation of a so-called reasonable proxy relationship with error caused by heterogeneity and through the ability of heterogeneity thresholds to detect heterogeneous regions. In this paper, previous findings regarding three heterogeneity statistics H1–H3 are revisited; a previous finding that H1 is superior to H2 and H3 is amended based on simulation experiments and upon enumeration of all possible regionalizations of a small gauge dataset across time scales from daily to monthly. Thresholds defined based on H1 are shown to be 4× too high for application to H2 and new thresholds are derived for H2. Two nonparametric heterogeneity statistics are tested and found to achieve only the unsatisfactory performance level of H3.
Discriminatory Power of Heterogeneity Statistics with Respect to Error of Precipitation Quantile Estimation
Houck, Mark. H (author) / Ferreira, Celso M / Wright, Michael J
2015
Article (Journal)
English
| British Library Online Contents | 2015
Estimation Issues for Precipitation Extreme Quantile Determination
| British Library Conference Proceedings | 2013
Wiley | 1997
Discriminatory Rate Disallowed
Wiley | 1985