Assessing future rainfall intensity–duration–frequency characteristics across taiwan using the k‐nearest neighbor method

Pei Yuan Chen, Ching Pin Tung, Jung Hsuan Tsao, Chia Jeng Chen

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

This study analyzes the changes in rainfall intensities across Taiwan using the k‐Nearest Neighbor method. Biases are corrected according to the identified discrepancy between the probability distribution of the model run and that of the observed data in the historical period. The projections of 21 weather stations in Taiwan under 10 (2RCP x 5GCM) scenarios for the near‐(2021– 2040) and far‐future (2081–2100) are derived. The frequently occurred short‐duration storm events in some regions decrease, but they are still vulnerable to flood considering the existing drainage capacities. More specifically, the land‐subsidence region in the central, the landslide‐sensitive mountainous region in the north and central, the pluvial‐ and fluvial‐flood prone region in the north, and the eastern regions with vulnerable infrastructures should be especially aware of longduration extreme events. Associations of the rainfall intensity with the different return period as well as the duration are further analyzed. The short‐duration extreme events will become stronger, especially for 1‐h events in the northern region and 1 or 2‐h events in both the southern and eastern regions. In addition, places without experiences of long‐lasting events may experience rainfall amounts exceeding 500 mm should be alert. Adaptation measures such as establishing distributed drainage system or renewing hydrological infrastructures in the eastern region are suggested considering the near future projection, and in the central and the southern regions for far future as well. Our findings can assist adaptation‐related decision‐making for more detailed stormwater/water resource management.

Original languageEnglish
Article number1521
JournalWater (Switzerland)
Volume13
Issue number11
DOIs
StatePublished - 1 Jun 2021

Keywords

  • Durations
  • K‐Nearest Neighbors (k‐NN) method
  • Rainfall‐intensity projection
  • Return periods
  • Temporal downscaling

Fingerprint

Dive into the research topics of 'Assessing future rainfall intensity–duration–frequency characteristics across taiwan using the k‐nearest neighbor method'. Together they form a unique fingerprint.

Cite this