Scholarlink

<p>Despite frequent flooding, the monsoon-dominated Nilwala catchment in Sri Lanka is monitored by only two streamflow gauging stations, leaving nearly 70% of the downstream catchment area ungauged. In response to similar challenges observed globally and the limited research efforts addressing them, this study presents a comprehensive flood frequency analysis framework for an ungauged, monsoon-flood-prone catchment, with its application to Akuressa city located in the downstream reach of the Nilwala catchment. Streamflow estimates at Akuressa over a 21-year period (2000–2020) were derived using a hydrological model parameter regionalization method proven appropriate for the Nilwala catchment. Based on the estimated streamflow, flood frequency analysis was performed using linear moments to fit Generalized Extreme Value (GEV), Generalized Logistic (GL), Generalized Pareto (GP), Log-logistic (Log-L), and Log-normal (Log-N) distributions for both Annual Maximum (AM) and Peak Over Threshold (POT) methods. Goodness of fit was evaluated using the Kolmogorov-Smirnov (K-S) test and flood quantiles were interpreted using five distribution visualization techniques. The statistical analysis was conducted using the Hydrologic Engineering Center’s Statistical Software Package (HEC-SSP). GEV, GL, and GP three-parameter distributions performed significantly better than Log-L and Log-N two-parameter distributions in both AM and POT, with GEV, GP, and GL showing superior performance in the POT based on the K-S values of 0.052, 0.054, and 0.056, respectively. GEV, identified as the optimal model for flood frequency analysis at Akuressa using the POT, estimated the extreme flood events in 2003 (1114 m<sup>3</sup>/s) and 2017 (1253 m<sup>3</sup>/s) to have return periods of approximately 250 and 333 years, respectively. Streamflow predictions up to a 10-year return period showed minimal variation, but varied significantly beyond 20 years, emphasizing the need for careful selection of probability distributions for extreme flood estimation. The proposed approach is applicable to other monsoon-dominated catchments with similar data scarcity, subject to validation and careful consideration of local conditions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Flood frequency assessment in the ungauged, flood-prone Akuressa city in a monsoon-dominated catchment in Sri Lanka using SWAT-simulated streamflow

  • 期刊论文

  • 发表日期：

    2026年7月28日

查看全文
  • D. M. K. P. Dissanayake,
  • T. N. Wickramaarachchi

摘要

Despite frequent flooding, the monsoon-dominated Nilwala catchment in Sri Lanka is monitored by only two streamflow gauging stations, leaving nearly 70% of the downstream catchment area ungauged. In response to similar challenges observed globally and the limited research efforts addressing them, this study presents a comprehensive flood frequency analysis framework for an ungauged, monsoon-flood-prone catchment, with its application to Akuressa city located in the downstream reach of the Nilwala catchment. Streamflow estimates at Akuressa over a 21-year period (2000–2020) were derived using a hydrological model parameter regionalization method proven appropriate for the Nilwala catchment. Based on the estimated streamflow, flood frequency analysis was performed using linear moments to fit Generalized Extreme Value (GEV), Generalized Logistic (GL), Generalized Pareto (GP), Log-logistic (Log-L), and Log-normal (Log-N) distributions for both Annual Maximum (AM) and Peak Over Threshold (POT) methods. Goodness of fit was evaluated using the Kolmogorov-Smirnov (K-S) test and flood quantiles were interpreted using five distribution visualization techniques. The statistical analysis was conducted using the Hydrologic Engineering Center’s Statistical Software Package (HEC-SSP). GEV, GL, and GP three-parameter distributions performed significantly better than Log-L and Log-N two-parameter distributions in both AM and POT, with GEV, GP, and GL showing superior performance in the POT based on the K-S values of 0.052, 0.054, and 0.056, respectively. GEV, identified as the optimal model for flood frequency analysis at Akuressa using the POT, estimated the extreme flood events in 2003 (1114 m3/s) and 2017 (1253 m3/s) to have return periods of approximately 250 and 333 years, respectively. Streamflow predictions up to a 10-year return period showed minimal variation, but varied significantly beyond 20 years, emphasizing the need for careful selection of probability distributions for extreme flood estimation. The proposed approach is applicable to other monsoon-dominated catchments with similar data scarcity, subject to validation and careful consideration of local conditions.