Low-Flow Frequency Analysis with Censored Data

by Chuck Kroll, (S.M.ASCE), Cornell Univ, Ithaca, United States,
Jery Stedinger, (M.ASCE), Cornell Univ, Ithaca, United States,

Document Type: Proceeding Paper

Part of: Water Policy and Management: Solving the Problems


Log-linear regression models are often fit to low-flow quantile estimates from gauged sites to obtain a relationship for estimating low-flow quantiles at ungauged sites. In some regions the low-flow quantile estimates of interest at some gauged sites are zero or below some measurement threshold, and thus are reported as zero. Such values cannot be employed with the traditional log-linear regression methods. A Monte Carlo analysis examined the performance of four models which address this problem. The common practice of adding a constant to all quantiles is inferior to both simple and sophisticated alternatives.

Subject Headings: Low flow | Frequency analysis | River flow | Water flow | Regression analysis | Mathematical models | Monte Carlo method | Probe instruments | Monaco | Europe

Services: Buy this book/Buy this article


Return to search