loading page

Analyzing Uncertainty in Probable Maximum Precipitation Estimation with a Large Ensemble Climate Simulation Data
  • Youngkyu Kim,
  • Sunmin Kim,
  • Yasuto Tachikawa
Youngkyu Kim
Graduate School of Engineering, Kyoto university

Corresponding Author:kim.youngkyu.44m@st.kyoto-u.ac.jp

Author Profile
Sunmin Kim
Graduate School of Engineering, Kyoto university
Author Profile
Yasuto Tachikawa
Graduate School of Engineering, Kyoto University
Author Profile


Probable maximum precipitation (PMP) is one of the most important key indices in hydrological design, and it should be estimated without the risk of overestimation or underestimation. However, observed extreme rainfall is insufficient to evaluate the estimated PMP in most regions, and there are no conclusive criteria to evaluate the possibility of PMP over- or underestimation. Our study aims to evaluate the reasonability of PMP estimation using a moisture-maximization method by comparing estimated PMPs based on the conventional method using surface dew points (SDP) with values based on actual precipitable water obtained from upper-air data (UAD). Furthermore, the estimated PMPs were evaluated with extreme-scale reference precipitation estimated from large ensemble climate simulation data (d4PDF) to check the possibility of PMP over- and underestimation. This reference value corresponds to a 3,000-year return period. The UAD-based estimation showed a reasonable PMP with a low deviation to the reference precipitation in most target areas, and the SDP-based estimation showed overestimated PMPs to the reference precipitation in areas with low SDPs. To control PMP overestimation of the SDP-based approach, the upper bound of the moisture maximizing ratio (MMR) in the SDP-based approach was limited to 2.0. Consequently, the SDP-based approach, which limits the upper bound of MMR, decreased deviations between estimated PMPs and reference values in areas with low SDPs. The SDP approach could reduce the possibility of PMP overestimation by limiting the upper bound of MMR. This evaluation could be conducted from extreme-scale reference precipitation under sufficient extreme events by utilizing the d4PDF.