Context: The Democratic People’s Republic of Korea (DPRK) is vulnerable to drought due to monsoon-driven variable precipitation that affects agricultural production and water security. Reliable multi-timescale drought forecasting aids risk mitigation; however, nonlinear non-stationary drought signals are hard to capture indata-scarce regions. Aims: This study developed and validated a Complete Ensemble Empirical Mode Decomposition with Adaptive Noise-Long Short-Term Memory (CEEMDAN-LSTM) hybrid Standardised Precipitation Evapotranspiration Index (SPEI) drought forecasting model. Its performance is systematically compared against a standalone LSTM for multiple SPEI timescales under DPRK climatic conditions. Methods: Monthly meteorological records from 10 stations collected over a period of 60 years were used to derive SPEI time series. CEEMDAN decomposed SPEI into intrinsic mode functions for noise reduction and trend extraction as LSTM inputs; hyperparameters were optimised by Bayesian optimisation. Model performance was assessed with Nash-Sutcliffe Efficiency (NSE), Kling-Gupta Efficiency (KGE), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). Key results: CEEMDAN-LSTM generally outperformed standalone LSTM with timescale-dependent skill. Short-term SPEI predictions had inherent accuracy limits (NSE < 0.5) yet reduced relative errors. Medium- to long-term SPEI yielded large accuracy gains (max NSE = 0.95 for SPEI-12), with residual-distribution analysis showing effective suppression of extreme prediction errors. Spatial heterogeneity existed; complex terrain at stations such as Chongjin limited decomposition-driven improvements. Conclusions: CEEMDAN-LSTM improves forecast skill by processing non-stationary drought signals. Without complex external predictors, this hybrid method enables reliable medium-long-term drought monitoring and extreme-event detection. Implications: This single-variable time-series method enables data-efficient drought early warning for data-scarce regions, supporting drought response and adaptive water management without large-scale remote-sensing or global datasets.
13 April 2025 in Journal of Atmospheric Science Research
The El Niño-Southern Oscillation (ENSO) is a dominant interannual climate mode influencing global weather patterns. This study investigates the evolving relationship between ENSO and seasonal climate over the Democratic People's Republic of Korea (DPRK) from 1950 to 2024. Using the Niño3.4 index alongside homogenized monthly temperature and precipitation records from 37 stations, we applied continuous wavelet analysis, Mann-Kendall abrupt change detection, and 21-year sliding correlation techniques. To isolate interannual ENSO signals, linear trends associated with global warming were removed from temperature data. Results indicate that 25 El Niño and 20 La Niña events occurred, with the Niño3.4 index exhibiting a dominant 1–5.7-year periodicity and a statistically significant regime shift in the late 1970s. Following detrending, the ENSO–winter temperature teleconnection weakened markedly after the late 1980s. Specifically, El Niño-induced DJF warming intensified by +0.45 °C (p < 0.01) in the 1991–2024 period compared to 1950–1990, whereas La Niña winters transitioned from anomalously warm to cold. Summer climate responses also shifted significantly: El Niño-related JJA precipitation decreased by 73.1 mm (from +20.7 to −52.4 mm, p < 0.05) in the DPRK, while La Niña summers exhibited opposite trends. These nonstationary relationships, driven by decadal reorganizations of Pacific-Asian atmospheric circulations, provide critical insights for improving seasonal climate prediction and informing regional climate adaptation strategies in the DPRK.
10 April 2026 in Environmental Science: Atmospheres
Summer precipitation over the northern part of the Korean Peninsula (SP‐NPKP) is critical for water resources, agriculture, and disaster prevention. This study aims to detect suitable atmospheric circulation indices for annual prediction of SP‐NPKP and to evaluate their predictive skill. We used 77 years of data from 1948 to 2024, including NCEP/NCAR reanalysis variables and observed summer precipitation from 37 stations. The study is based on the finding that 1‐year lag correlations between selected indices and SP‐NPKP generally exceed concurrent correlations. We analyzed linear trends of SP‐NPKP, sea level pressure over Asia, 700‐hPa vorticity anomalies, and Arctic Oscillation indices. Using the ‘area shift’ experiment, we identified optimal domains for sea level pressure anomalies over Asia and the North Pacific, yielding effective predictors: the SLP anomaly index over central Eurasia (SLPAI‐1), that over the Okhotsk Sea (SLPAI‐2), the 700‐hPa vorticity anomaly index (VORAI‐700), the 500‐hPa temperature anomaly index (TAI‐500), and the leading SLP principal component (SLP‐PC1). Annual predictions were performed using principal component regression (PCR) and backpropagation neural network (BPNN) models. Based on 5‐fold cross‐validation, PCR showed limited skill with R2 = 0.1628, RMSE = 151.57 mm, and MAE = 124.22 mm, while BPNN demonstrated significantly superior performance with R2 = 0.4031, RMSE = 119.81 mm, and MAE = 103.15 mm. This confirms that neural networks better capture the nonlinear dynamics of regional precipitation. Our study provides a novel, data‐driven framework for identifying region‐specific predictors, offering valuable insights for improving operational seasonal prediction systems in East Asia.