A systematic quantitative review on the performance of some of the recent short-term rainfall forecasting techniques

Abstract

Rainfall forecasting is a high-priority research problem due to the complex interplay of multiple factors. Despite extensive studies, a systematic quantitative review of recent developments in rainfall forecasting is lacking in the literature. This study conducted a systematic quantitative review of statistical, numerical weather prediction (NWP) and machine learning (ML) techniques for rainfall forecasting. The review adopted the preferred reporting items for systematic reviews and meta-analyses (PRISMA) technique for screening keywords and abstracts, leading to 110 qualified papers from multiple databases. The impact of rainfall threshold, meteorological parameters, topography, algorithm techniques, geographic location, the horizontal resolution of the model, and lead time on rainfall forecast was examined. The review shows the importance of precipitable water vapor (PWV) along with other meteorological parameters for accurate nowcasting in coastal and mountainous regions. An increase in rainfall forecast uncertainty with an increase in the lead time makes the NWP model less popular for the short-term forecast. The pre-processing techniques increased the accuracy of ML techniques by considering extreme values and detecting the irregularly distributed multi-scale features of rainfall in space and time. Future research can focus on hybrid models with improved accuracy for nowcasting. The output from the hybrid model serves as input for the decision support system required for urban flood risk management.

Landslide initiation thresholds in data sparse regions: Application to landslide early warning criteria in Sitka, Alaska, USA

“…Uncertainty in the forecasted precipitation is added to uncertainty of the model and decision threshold. As storms approach and precipitation forecasts become more constrained, the precipitation uncertainty will be reduced (Fabry and Seed, 2009;Ashok and Pekkat, 2022). Thus, landslide predictions for the future (days in 725 advance) become more accurate as the storm approaches (hours in advance).…”

Can Denoising Enhance Prediction Accuracy of Learning Models? A Case of Wavelet Decomposition Approach

“…In this section, we begin with a brief review on selected works that have proposed benchmark forecasting techniques and hybrid models integrating wavelet analysis for the purpose of price forecasting. Extensive research has been conducted in the realm of time series forecasting, leading to the proposal and evaluation of numerous modeling techniques [8,9]. The autoregressive integrated moving average (ARIMA) methodology has emerged as the most widely employed linear technique in time series analysis [10].…”

WF-UNet: Weather Fusion UNet for Precipitation Nowcasting

“…In addition, forecasting with NWP may be sensitive to noise present in the measurements of weather variables [11,12]. NWP models may take hours to run and are also less accurate than persistence-based forecasts on less than 4 hour predictions [13,14]. In recent years, the enormous amount of ever-increasing weather data has stimulated research interest in data-driven machine learning techniques for nowcasting tasks [15][16][17][18][19][20].…”

“…Instead, they analyze and learn from historical weather data such as past wind speed and precipitation maps to predict the future. By taking advantage of available historical data, data-driven approaches have shown better performance than classical ones in many forecasting tasks [13,14].…”