Climate change is a global problem since many countries worldwide are becoming increasingly vulnerable to natural disasters. Numerous climate models in various studies project a decline in agricultural productivity that will mainly be due to excessive heat in tropical and subtropical regions, especially in Southeast Asia. As a Southeast Asian country, Malaysia is no exception to this problem. Hence, the present study aimed to examine the impact of climate change on rice yields in Malaysia. A panel data approach was adopted using data from 1987 to 2017 on eight granary areas in Peninsular Malaysia. The main objectives were to assess the impact of climate variables (i.e., minimum and maximum temperature and precipitation) on rice yield and the variance of the impact during the main season and off-season. Our regression results indicate that precipitation was not statistically significant in all model specifications for both the main and off-season. While the maximum temperature was found to be negatively associated with yield during the off-season, the minimum temperature showed a positive effect in both cropping seasons. We used the HadGEM3-GC31 N512 resolution model based on the high-emission Shared Socioeconomic Pathways 8.5 scenario (SSPs-8.5) from the High-Resolution Model Intercomparison Project (HighResMIP) of the Coupled Model Intercomparison Project Phase 6 (CMIP6) to project future climate change in 2030 and 2040. The projected results indicate that rice yield would show a more positive trend by 2040 when compared to the previous decade, ranging from −0.02 to 19.85% during the main season and −2.77 to 7.41% during the off-season. Although rice yield is likely to increase in certain areas, other areas are projected to experience negative effects. Hence, adaptation at the farm level remains crucial, specifically during the off-season, since climate change could widen the gaps in rice yields between cropping seasons and among granary areas.

Despite the implementation of various initiatives, dengue remains a significant public health concern in Malaysia. Given that dengue has no specific treatment, dengue prediction remains a useful early warning mechanism for timely and effective deployment of public health preventative measures. This study aims to develop a comprehensive approach for forecasting dengue cases in Selangor, Malaysia by incorporating climate variables. An ensemble of Multiple Linear Regression (MLR) model, Long Short-Term Memory (LSTM), and Susceptible-Infected mosquito vectors, Susceptible-Infected-Recovered human hosts (SI-SIR) model were used to establish a relation between climate variables (temperature, humidity, precipitation) and mosquito biting rate. Dengue incidence subject to climate variability can then be projected by SI-SIR model using the forecasted mosquito biting rate. The proposed approach outperformed three alternative approaches and expanded the temporal horizon of dengue prediction for Selangor with the ability to forecast approximately 60 weeks ahead with a Mean Absolute Percentage Error (MAPE) of 13.97 for the chosen prediction window before the implementation of the Movement Control Order (MCO) in Malaysia. Extended validation across subsequent periods also indicates relatively satisfactory forecasting performance (with MAPE ranging from 13.12 to 17.09). This research contributed to the field by introducing a novel framework for the prediction of dengue cases over an extended temporal range.

As climates change globally, water-related disasters increase, causing substantial economic losses and safety risks. During floods, river water levels show unpredictable fluctuations, introducing substantial noise that complicates accurate prediction. A hybrid model that uses eight-dimensional input data from hydrological and meteorological stations is proposed to address these challenges. Initially, the Variational Mode Decomposition preprocesses and denoises water level data, resulting in decomposed Intrinsic Mode Functions (IMFs). Then, the Pearson correlation coefficient between each IMF and input characteristics is computed, and the fluctuation factor for each IMF is defined. IMFs are categorized based on a threshold, leading to a hybrid prediction model. This model integrates convolutional neural networks (CNNs) for spatial information and bidirectional long short-term memory (BiLSTM) networks with an attention mechanism for learning from past and future data points. Comparative evaluations of mean absolute percentage error, root mean square error, mean absolute error, and goodness of fit (R2) show that the proposed model outperforms existing LSTM and CNN–BiLSTM frameworks, reducing RMSE by at least 20% and increasing R2 by approximately 10% on average. The model's practical significance lies in improving the accuracy and efficiency of meteorological forecasting and flood warning systems, contributing substantially to global disaster preparedness and response strategies.

Flood disasters occur worldwide, and flood risk prediction is conducive to protecting human life and property safety. Influenced by topographic changes and rainfall, the water level fluctuates randomly and violently during the flood, introducing many noises and directly increasing the difficulty of flood prediction. A data-driven flood forecasting method is proposed based on data preprocessing and a two-layer BiLSTM-Attention network to improve forecast accuracy. First, the Variational Mode Decomposition (VMD) is used to decompose the data for reducing noise and produce suitable Intrinsic Mode Functions (IMFs); Then, an optimized two-layer attention-based Bidirectional Long Sshort-Term memory (BiLSTM-Attention) network is constructed to predict each IMF. Finally, two optimization algorithms are used to obtain the optimized parameters of VMD and BiLSTM intelligently, increasing the self-adaptability. The inertia factor of particle swarm optimization is improved and then used to optimize the five hyperparameters of BiLSTM. The proposed model reduces storage errors for smaller training sets and can achieve good performance. Three water level data sets from the Yangtze River in China are used for comparative experiments. Numerical results show that the peak height absolute error is within 2 cm, and the relative error of peak time arrival is within 30%. Compared with LSTM, BiLSTM, CNN-BiLSTM-attention, etc., the proposed model reduces the root mean square error by at least 50% and has advantages for high-risk forecasting when the water level exceeds the defense line and fluctuates prominently.