Haneul Lee

Drought is a complex natural hazard that develops across different temporal scales and may propagate from meteorological drought to hydrological drought. In addition, the physical characteristics of drought do not necessarily correspond directly to how drought is represented in the news media. This study analyzed meteorological and hydrological drought characteristics at the level of the 17 first-level administrative divisions (provinces) of South Korea and compared them with the frequency of drought-related news articles. The Standardized Precipitation Index (SPI), Standardized Precipitation Evapotranspiration Index (SPEI), and Streamflow Drought Index (SDI) were calculated at 1-, 3-, 6-, and 12-month timescales using long-term meteorological and streamflow observations. Drought-related news articles published from 2021 to 2025 were retrieved from BIGKinds and processed using explicit cleaning rules, and regions with relatively high frequencies of drought-related news were identified. The spatial agreement between regions with high drought-news frequency and regions with high drought severity derived from each drought index was then evaluated using the top-10 overlap, the Jaccard index, a threshold-based comparison, rank correlation, and permutation tests, and their temporal correspondence was evaluated using monthly rank correlations. In the spatial comparison, the overlap between the top-10 provinces (at most 7 of 10) did not exceed that expected by chance, and this result did not depend on the ranking threshold, the annual aggregation method, the method of counting articles, or normalization by population. In contrast, the monthly number of drought-related articles increased with the number of provinces under drought conditions, most strongly for the 12-month indices (Spearman ρ = 0.42–0.57), and was significantly correlated with the provincial SPI-12 and SPEI-12 values in 8–11 of the 16 provinces. These exploratory results for 2021–2025 suggest that drought-related news coverage follows the temporal evolution of prolonged drought, but that its distribution among provinces cannot be represented by a single drought index or timescale and is only weakly related to the spatial pattern of physical drought severity. News frequency is therefore better interpreted as an indicator of when drought attracts media attention than of where drought is most severe, and its use as complementary information in drought monitoring requires validation against independent drought-impact data.

A statistical model was developed in this study for a quantitative estimation of the damage caused by floods and typhoons in North Korea, and its predictive performance was evaluated. Despite North Korea experiencing recurring damages due to natural disasters such as floods and typhoons, a systematic damage estimation framework is lacking. Therefore, a damage estimation model was developed by treating the entire territory of North Korea as a single analysis unit, and using damage statistics from 1993 to 2018 along with data on precipitation and wind speed from 27 meteorological stations. The principal component regression model showed overall superior performance compared to a generalized linear model, achieving (a) an average NRMSE of 0.1901 and R-squared of 42% for flood damage estimation, and (b) an average NRMSE of 0.3077 and R-squared of 85% for typhoon damage estimation. This quantitative methodology proposed for estimating disaster damage in data-scarce environments can serve as a model and be applied to other countries facing similar disasters and lacking systematic damage estimation frameworks.

The importance of efficient water resource supply has been acknowledged, and it is essential to predict short-term water consumption in the future. Recently, it has become possible to obtain data on water consumption at the household level through smart water meters. The pattern of these data is nonlinear due to various factors related to human activities, such as holidays and weather. However, it is difficult to accurately predict household water consumption with a nonlinear pattern with the autoregressive integrated moving average (ARIMA) model, a traditional time series prediction model. Thus, this study used a deep learning-based long short-term memory (LSTM) approach to develop a water consumption prediction model for each customer. The proposed model considers several variables to learn nonlinear water consumption patterns. We developed an ARIMA model and an LSTM model in the training dataset for customers with four different water-use types (detached houses, apartment, restaurant, and elementary school). The performances of the two models were evaluated using a test dataset that was not used for model learning. The LSTM model outperformed the ARIMA model in all households (correlation coefficient: mean 89% and root mean square error: mean 5.60 m³). Therefore, it is expected that the proposed model can predict customer-specific water consumption at the household level depending on the type of use.