This study presents a statistical approach for assessing meteorological hazards based on trends and abrupt changes in precipitation characteristics. Daily rainfall data from 64 stations in South Korea (SK) and 27 stations in North Korea (NK) were used to identify temporal patterns in the rainfall characteristics of both regions using seven rainfall indices, such as the total annual rainfall and annual number of wet days. This study suggests the use of three steps in identifying meteorological hazards based on two statistical analyses. In step 1, we conducted a trend analysis of a 10-year moving average of the rainfall index using the Mann-Kendall (MK) trend test. Most stations (65.6 %) in SK exhibit clear increasing trends in five indices, whereas far fewer have data indicating any trends in five of the indices in NK (25.9 %). In step 2, abrupt changes in all rainfall indices were identified using a Bayesian Change Point (BCP) approach. The results contradict those from the MK trend analysis. The proportion of stations in NK where trends were identified is much higher than that in SK. In step 3, the results from the two previous steps were integrated to identify the meteorological hazards based on the identified trend and change point. The BCP approach can be used to identify meteorological hazards that MK cannot, as the former approach focuses on the change point during the entire period. As a result, meteorological stability at the sites of weather stations can be identified, and then the meteorological hazards across the entire Korean peninsula can be spatially interpolated. Although SK and NK are located on the same peninsula, distinct differences in the trends were observed.
All Science Journal Classification (ASJC) codes
- Atmospheric Science