Rainfall Forecasts Research Articles

Type - 2 mamdani fuzzy inference system based model for rainfall forecasting

Weather Forecasting is very essential as it is helpful in saving lives and materials by predicting disasters such as cyclonic storms, tsunamis, extreme rainfall, etc. Within the defined range of rainfall rate approximation, this study investigates the application of Fuzzy Logic (FL) to forecast rainfall using the Interval Type –2 Fuzzy Inference System (IT2FIS). Environmental parameters which influence rainfall have been applied in this analysis and every implementation is carried out using MATLAB 9.13. The performance of IT2FIS model is compared with the actual data. Correlation coefficient (R2) and Root Mean-Squared Error (RMSE) have been used to evaluate the performance metrics of the proposed model. The results suggest that the IT2FIS model can capture the dynamic behavior of rainfall data and generate reasonable results, implying that it might be beneficial in long-term rainfall prediction.

Extended-Range Forecast of Winter Rainfall in the Yangtze River Delta Based on Intra-Seasonal Oscillation of Atmospheric Circulations

The Yangtze River Delta (YRD) is an important economic region in China. Heavy winter rainfall may pose serious threats to city operations. To ensure the safe operation of the city, meteorological departments need to provide forecast results for the Spring Festival travel rush weather service. Therefore, the extended-range forecast of winter rainfall is of considerable importance. To solve this problem, based on the analysis of low-frequency rainfall and the intra-seasonal oscillation of atmospheric circulation, an extended-range forecast model for winter rainfall is developed using spatiotemporal projection methods and is applied to a case study from 2020. The results show that: (1) The precipitation in the YRD during the winter has a significant intra-seasonal oscillation (ISO) with a periodicity of 10–30 d. (2) The atmospheric circulations associated with winter rainfall in the YRD have a significant characteristic of low-frequency oscillation. From a 30-day to a 0-day lead, large modifications appear in the low-frequency atmospheric circulations at low, mid, and high latitudes. At low latitudes, strong wet convective activity characterized by a negative OLR combined with a positive RH700 correlation coefficient moves northwestward and covers the entire YRD. Meanwhile, the Western Pacific subtropical high (WPSH) characterized by a positive Z500 anomaly enhances and lifts northward. At mid and high latitudes, the signal of negatively correlated Z500 northwest of Lake Balkhash propagates southeastward, indicating the cold is air moving southward. Multiple circulation factors combine together and lead to the precipitation process in the YRD. (3) Taking the intra-seasonal dynamical evolution process of the atmospheric circulation as the prediction factor, the spatiotemporal method is used to build the model for winter mean extended-range precipitation anomaly tendency in the YRD. The hindcast for the recent 10 years shows that the ensemble model has a higher skill that can reach up to 20 days. In particular, the skill of the eastern part of the YRD can reach 25 days. (4) The rainfall in the 2019/2020 winter has a significant ISO. The ensemble model could forecast the most extreme precipitation for 20 days ahead.

Forecasting monthly rainfall using hybrid time-series models and Monte Carlo simulation amidst security challenges: a case study of five districts from northern Nigeria

AbstractNigeria’s agricultural sector relies heavily on rainfall, but insecurity in various regions poses significant challenges. This study aims to address this issue by identifying secure, rain-rich areas in northern Nigeria to support sustainable agriculture. Two models, one integrating classical statistical methods (polynomial and Fourier series fittings) and another using a hybrid approach (artificial neural networks, polynomial, and Fourier series fittings), were employed to analyze historical rainfall data from 1981 to 2021 in the selected districts (Kano, Zaria, Bida, Nguru, and Yelwa) known for their rainfall levels and security stability. The study demonstrates that the machine learning-classical hybrid model outperforms existing models, including the classical-classical hybrid and benchmark models like Iwok’s (2016) model, Fourier series, and SARIMA models. Multi-step ahead forecasting with this hybrid model reveals potential changes in rainfall patterns. Notably, Kano, Zaria, Bida, and Yelwa are expected to experience increased rainfall from 2022 to 2026, while Nguru may initially witness decreased rainfall, with improvement in the final year (2026). In conclusion, this study introduces an effective approach for rainfall modeling and forecasting, facilitating the identification of secure agricultural regions in northern Nigeria. These findings carry implications for crop production and agricultural development, contributing to climate resilience efforts and assisting stakeholders in strategic decision-making for regional agricultural investments.

TOWARDS BUILDING AN INTELLIGENT MOBILE APP FOR RAINFALL PREDICTION USING WEIGHTED AVERAGE ENSEMBLE DEEP SUPERVISED LEARNING

Predicting rainfall accurately and on time is essential for a variety of tasks, including electricity generation, building, flood warnings, aviation operations, and the sustainable use of water resources. Traditionally, rainfall forecasts were based on physical models of the atmosphere, which are imprecise over extended periods of time due to their instability to perturbations and thus, are inaccurate for large periods. Artificial Intelligence (AI) techniques as a means of predicting rainfall have been applied in most developed countries, these have been found to offer many possibilities. AI techniques are more robust to perturbations. Over the past 20 years, rainfall forecasting using AI approaches has demonstrated astounding precision. Rainfall forecasting accuracy could be increased by AI algorithms that use historical weather data elements to reveal hidden trends. This study thereafter proposed the development of an AI mobile app for predicting rainfall that is based on weighted average ensemble deep supervised learning that combines three machine learning algorithms namely; Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), and Gated Recurrent Unit (GRU). . In order to prevent future outbreaks of rainfall hazards, the proposed AI mobile app for rainfall prediction will aid in raising awareness among the public, government officials, planning agencies, disaster management agencies, and related organizations.

Hydroclimatic projection: statistical learning and downscaling model for rainfall and runoff forecasting

Abstract The study is carried out to investigate the surface runoff depth with changing precipitation due to climate change in the study area where sandy loam and loamy soil are dominant. In this study, future rainfall is projected by a statistical downscaling model (SDSM) using a set of predictors derived from a Coupled Model Intercomparison Project Phase 6 (CMIP6) global climate model (GCM) [the Norwegian Earth System Model (NorESM)] with updated scenarios SSP 4.5 and SSP 8.5. Daily rainfall values for the observed period (1981 to 2014) are validated using statistical learning and evaluated with matrices, namely, root mean square error (RMSE), coefficient of correlation, and Nash–Sutcliffe efficiency (NSE), which are found to be valid for further predictions. Rainfall projections show a decrease in rainfall trend of 50% from 2030 to 2040 for scenario SSP 4.5 and an increase of 7% from 2040 to 2050. Predicted rainfall for scenario SSP 8.5 shows a similar trend of decreasing rainfall of 24% for the period 2030–2040 and an increase of 19% in the period 2040–2050. Furthermore, these rainfall values are spatially modelled in a geographic information system (GIS) and rainfall maps are obtained. The obtained rainfall map, land-use map, and soil map are overlaid to compute curve numbers and runoff depths. A similar trend of decrease in runoff is observed for the period 2030–2050. The overall trend of climate change shows a water-stressed scenario.

Coupling Physical Factors for Precipitation Forecast in China With Graph Neural Network

AbstractPrecipitation exerts far‐reaching impacts on both socio‐economic fabric and individual well‐being, necessitating concerted efforts in accurate forecasting. Deep learning (DL) models have increasingly demonstrated their prowess in forecasting meteorological elements. However, traditional DL prediction models often grapple with heavy rainfall forecasting. In this study, we propose physics‐informed localized graph neural network models called ω‐GNN and ω‐EGNN, constrained by the coupling of physical variables and climatological background to predict precipitation in China. These models exhibit notable and robust improvements in identifying heavy rainfall while maintaining excellent performance in forecasting light rain by comparing to numerical weather prediction (NWP) and other DL models with multiple perturbation experiments in different data sets. Surprisingly, within a certain range, even when a DL model utilizes more input variables, GNN can still maintain its advantage. The methods to fuse physics into DL model demonstrated in this study may be promising and call for future studies.

Artificial intelligence predicts normal summer monsoon rainfall for India in 2023

Inaccuracy in the All Indian Summer Monsoon Rainfall (AISMR) forecast has major repercussions for India's economy and people's daily lives. Improving the accuracy of AISMR forecasts remains a challenge. An attempt is made here to address this problem by taking advantage of recent advances in machine learning techniques. The data-driven models trained with historical AISMR data, the Niño3.4 index, and categorical Indian Ocean Dipole values outperform the traditional physical models, and the best-performing model predicts that the 2023 AISMR will be roughly 790 mm, which is typical of a normal monsoon year.

Interpretable spatial-temporal attention convolutional network for rainfall forecasting

The interpretability of rainfall forecasting models is a major challenge in the field of artificial intelligence. Its importance is equal to the evaluation of model accuracy. Owing to the uncertainty and nonlinearity of the rainfall forecasting process, existing hydrological rainfall forecasting models often have low prediction robustness, and the machine learning method ignores the influence of physical factors in the surrounding areas; thus, the physical interpretability of the model is inferior. This study proposes an interpretable spatial-temporal attention convolution network for rainfall interval forecasting (IDSTA-TCN). To analyze and mine the uncertainty of input data more effectively, we enhanced a variational mode decomposition (EBO-VMD) algorithm based on Bayesian optimization to decompose the input data. To improve the performance of the original temporal convolution network (TCN) unit in rainfall forecasting and to effectively use the spatial-temporal information of rainfall forecasting data, we improved the spatial-temporal attention mechanism, dynamically combined the geographical information of the meteorological stations in the surrounding area to assist the network unit in better mining the nonlinear impact of historical input information on current rainfall forecasting, and improved the prediction performance of the IDSTA-TCN model. Finally, we designed a more reasonable error factor adjustment interval output strategy to better quantify the impact of prediction errors on the rainfall interval prediction results. Moreover, we conducted a series of experiments to compare and analyze the accuracy of each model according to the most commonly used evaluation indicators in the field of rainfall forecasting. The results demonstrate that the proposed IDSTA-TCN model is superior to the baseline models in most cases. We preliminarily analyzed the spatial-temporal attention weights, which explained the physical rationality of our model.

Time series analysis and development of simulation model for monthly rainfall using ARIMA model

Rainfall holds critical significance for water resource applications, particularly in rainfed agricultural systems. This study employs the Autoregre ssive Integrated Moving Average (ARIMA) technique, a data mining approach commonly used for time series analysis and future forecasting. Given the increasing importance of climate change forecasting in averting unexpected natural hazards such as floods, frost, forest fires, and droughts, accurate weather data forecasting becomes imperative. The objective of this study was to develop a Seasonal Auto-Regressive Integrative Moving Average (SARIMA) model for forecasting monthly rainfall in Junagadh Station, Gujarat. Utilizing 50 years of historical data (1968 to 2016), the SARIMA model predicts weekly rainfall for the subsequent five years (2018 to 2022). Through comprehensive evaluation using ACF and PACF plots, AIC, SBC, MAPE, and MAE values, the study identifies SARIMA (1,0,0)(3,1,1)12 as the optimal model, offering the most accurate prediction. The robust results affirm that the SARIMA model provides reliable and satisfactory weekly rainfall predictions. This research contributes valuable insights into the precision and efficacy of SARIMA models for rainfall forecasting, aiding in strategic water resource management in the Junagadh region.

A Comparative Analysis of ARIMA and other Statistical Techniques in Rainfall Forecasting: A Case Study in Kolkata (KMC), West Bengal

Rainfall forecasting in urban areas is a significant consideration for city planners due to its connection with urban water management. In this study, the ARIMA (auto-regressive integrated moving average) model, as well as several regression approaches such as simple linear and second to sixth-degree polynomial regression equations, have been used to forecast the annual rainfall based on 120 years of monthly and annual rainfall from 1901 to 2020 in Kolkata Municipal Corporation (KMC), West Bengal. This study compares the performance of ARIMA and other regression techniques in forecasting rainfall using the metrics of R-squared and root mean square error (RMSE). The ARIMA model has been implemented using machine learning techniques in the Python programming language, while additional regression equations have been computed and analyzed using Microsoft Excel 2019. In order to employ the ARIMA model, all assumptions were assessed, and the optimal model order was established using the import auto-Arima package from the pmdarima.arima library. The stepwise model.aic function yielded 0,1,1 as the most suitable order for the model. The findings indicate that, out of all the regression methods employed for rainfall prediction, the fifth-degree polynomial equation exhibits the lowest root mean square error (RMSE), establishing it as the most effective model for rainfall forecasting in this study.

Long-term and short-term rainfall forecasting using deep neural network optimized with flamingo search optimization algorithm

Rainfall forecasting is essential because heavy and irregular rainfall creates many impacts like destruction of crops and farms. Here, the occurrence of rainfall is highly related to atmospheric parameters. Thus, a better forecasting model is essential for an early warning that can minimize risks and manage the agricultural farms in a better way. In this manuscript, Deep Neural Network (DNN) optimized with Flamingo Search Optimization Algorithm (FSOA) is proposed for Long-term and Short-term Rainfall forecasting. Here, the rainfall data is obtained from the standard dataset as Sudheerachary India Rainfall Analysis (IRA). Moreover, the Morphological filtering and Extended Empirical wavelet transformation (MFEEWT) approach is utilized for pre-processing process. Also, the deep neural network is utilized for performing rainfall prediction and classification. Additionally, the parameters of the DNN model is optimizing by Flamingo Search Optimization Algorithm. Finally, the proposed MFEEWT-DNN- FSOA approach has effectively predict the rainfall in different locations around India. The proposed model is implemented in Python tool and the performance metrics are calculated. The proposed MFEEWT-DNN- FSOA approach has achieved 25%, 26%, 25.5% high accuracy and 35.8%, 24.7%, 15.9% lower error rate for forecasting rainfall in Cannur at Kerala than the existing Map-Reduce based Exponential Smoothing Technology for rainfall prediction (MR-EST-RP), modular artificial neural networks with support vector regression for rainfall prediction (MANN-SVR-RP), and biogeography-based extreme learning machine (BBO-ELM) (BBO-ELM-RP) methods respectively.

PRAKIRAAN INDEKS KEKERINGAN MENGGUNAKAN METODE SEASONAL AUTOREGRESSIVE INTEGRATED MOVING AVERAGE (SARIMA) BERDASARKAN DATA STANDARDIZED PRECIPITATION INDEX (SPI) KOTA BANJARBARU

Drought is a disaster that has a bad impact, especially in the city of Banjarbaru. There are various ways to reduce the impact of drought in the future, one of which is by looking for information regarding the predicted drought index for the following year. The data used in this research to find the drought index value is Banjarbaru City rainfall data for 2007-2022. Seasonal Autoregressive Integrated Moving Average (SARIMA) method is a calculated method for predicting rainfall data and the data obtained is a forecast of rainfall in the city of Banjarbaru for the next 12 periods, namely SARIMA (0,2,3) (0,1,1)12. This model is a model that is suitable for use because it has fulfilled several tests, namely stationarity of variance and mean, significance test, white noise test and normality test with an AIC value of 1022,60 and an equation model obtained from SARIMA (0,2,3) (0,1,1)12 is (1-B)2 (1-B12) Zt=(1+1,77B-0,54B2+ 0,23B3 )(1-0,96B12 )εt. After obtaining forecast rainfall data for the next 12 periods. Rainfall data for 2007-2022 and forecast results for 2023 were used to find the drought index value using the Standardized Precipitation Index (SPI) method. It was found that the highest negative drought index value occurred in January, namely -1,774, including the dry category and the drought index had a positive value The highest occurred in June, namely 0,582, including the normal category. The calculation results of this drought index forecast are used to provide additional information to anticipate drought disasters in the future. Keywords: Drought Index, Rainfall, SPI Method, SARIMA Method, AIC

Process-Informed Subsampling Improves Subseasonal Rainfall Forecasts in Central America.

Subseasonal rainfall forecast skill is critical to support preparedness for hydrometeorological extremes. We assess how a process-informed evaluation, which subsamples forecasting model members based on their ability to represent potential predictors of rainfall, can improve monthly rainfall forecasts within Central America in the following month, using Costa Rica and Guatemala as test cases. We generate a constrained ensemble mean by subsampling 130 members from five dynamic forecasting models in the C3S multimodel ensemble based on their representation of both (a) zonal wind direction and (b) Pacific and Atlantic sea surface temperatures (SSTs), at the time of initialization. Our results show in multiple months and locations increased mean squared error skill by 0.4 and improved detection rates of rainfall extremes. This method is transferrable to other regions driven by slowly-changing processes. Process-informed subsampling is successful because it identifies members that fail to represent the entire rainfall distribution when wind/SST error increases.

Short-term rainfall prédiction using prédictive analytics: A case study in Telangana

Rainfall forecasting is critical because heavy rains can bring a full of tragedies. This prediction aids people in requiring preventive steps, and it should be correct. There are two types of rainfall forecasting: short-term forecasting and long-term forecasting. Predictions are typically correct for short-term, but creating a model for future rainfall prediction is the most difficult task. Because it is strongly tied to economy and human lives, heavy precipitation forecast could be a severe disadvantage for natural science departments. It’s the cause of natural disasters like floods and droughts that affect people all around the world every year. For countries like India, where agriculture is the primary source of income, the accuracy of rainfall estimates is critical. Regression may be used in the prediction of precipitation utilizing machine learning approaches. The goal of this work is to provide non-experts with easy access to the techniques and approaches used in the field of precipitation prediction, as well as a comparison of the various machine learning algorithms.

Monthly Rainfall Forecasting Using High Order Singh's Fuzzy Time Series Based on Interval Ratio Methods: Case Study Semarang City, Indonesia

Monthly Rainfall Forecasting Using High Order Singh's Fuzzy Time Series Based on Interval Ratio Methods: Case Study Semarang City, Indonesia

Prediction of Streamflow in River Basin-Using ANN

In addition to the flood level predictions, our system provides valuable insights into future rainfall patterns. With the data set we have gathered, we can determine the expected amount of rainfall in the upcoming months. By combining the flood level predictions with the rainfall data, we can better understand the overall flood risk and take proactive measures to mitigate its impact. Our system equips us with the necessary information to make informed decisions and enhance flood preparedness strategies. The main difference is that we focus on predicting flood levels using a combination of current water level data and real-time weather data. This allows us to have a more accurate understanding of potential flood events. Additionally, our paper also incorporates rainfall data to assess the risk of flooding in the coming months. By considering multiple factors, we aim to provide a more holistic understanding of flood risks and enhance preparedness strategies. By combining the flood level predictions with rainfall forecasts, we can assess the flood risk in the coming months and take preventive actions, such as implementing early warning systems or strengthening infrastructure, to minimize the impact of potential flood events.

Long-range forecasting of daily rainfall using machine learning techniques

Long-range forecasting of daily rainfall using machine learning techniques

Different Initial Condition Perturbation Methods for Convection-Permitting Ensemble Forecasting over South China during the Rainy Season

Abstract In this study, downscaling, ensemble data assimilation, time lagging, and their combination were used to generate initial condition (IC) perturbations for 12-h convection-permitting ensemble forecasting for heavy-rainfall events over South China during the rainy season in 2013–20. These events were classified as weak- and strong-forcing cases based on synoptic-scale forcing during the presummer rainy season and as landfalling tropical cyclone (TC) cases. This study investigated the impacts of various IC perturbation methods on multiscale characteristics of perturbations and the forecast performance for both nonprecipitation and precipitation variables. These perturbation methods represented different source IC uncertainties and thus differed in multiscale characteristics of perturbations in vertical structures, horizontal distributions, and time evolution. The combination of various IC perturbation methods evidently increased perturbations or spreads of precipitation in both magnitude and location and thus improved the forecast-error estimation. Such an improvement was most and least evident for TC cases during the early and late forecasts, respectively, and was more evident for strong- than weak-forcing cases beyond 6 h. The combination of various IC perturbation methods generally improved both the ensemble-mean and probabilistic forecasts with case-dependent improvements. For heavy rainfall forecasting, 1–6-h improvements were most prominent for TC cases in terms of discrimination and accuracy, while 7–12-h improvements were least prominent for weak-forcing cases in terms of reliability and accuracy. In particular, the improvements in predicting weak-forcing cases increased with spatial errors. In contrast, for strong-forcing cases, the improvements were least and most prominent before and beyond 6 h, respectively. Significance Statement Precipitation forecasting for heavy-rainfall events over South China in the rainy season is still challenging due to large uncertainties. Convection-permitting ensemble forecasting is expected to address such uncertainties to improve forecasts of heavy rainfall. However, it is not yet clear how to optimally design convection-permitting ensembles by implementing perturbations in initial conditions (ICs). This study investigates the impacts of various IC perturbation methods on convection-permitting ensemble forecasting over South China in the rainy season. Various IC perturbation methods show discrepant multiscale characteristics of perturbations, which generally complement each other when these perturbations are combined. The added values of combining various IC perturbation methods in forecasting are confirmed for most variables. However, such values are case dependent, with the largest values for tropical cyclone cases during the early forecasts and for the presummer rainy season cases with strong synoptic-scale forcing during late forecasts. Thus, it is still essential to further improve the combination of various types of IC perturbation methods.

Implementing advanced techniques for urban mountain torrent surveillance and early warning using rainfall predictive analysis

Urban flood forecasting and early warning play a pivotal role in ensuring efficient flood mitigation and management. The unpredictability in precipitation's intensity, temporal patterns, and spatial distribution introduces considerable variability into the basin's flow dynamics. This, in turn, escalates the uncertainty surrounding hydrological predictions, complicating the task of flood forecasting and early warning. To address these challenges, this research introduces a method that refines rainfall forecasts using a Wavelet Neural Network (WNN). By establishing a benchmark for area rainfall, we've developed a comprehensive disaster prevention and early warning system that synergizes real-time precipitation data, area rainfall, and flood peak predictions. Specifically tailored for urban terrains prone to mountain torrents, the WNN-based monitoring and pre-alarm model offers a sound and practical forecasting tool. Its relevance is accentuated by its potential to spearhead urban flood control initiatives. Our findings validate the model's adaptability and efficacy, particularly within urban mountainous watersheds, heralding a fresh paradigm in mountain flood disaster forecasting and early warnings.

A Novel Linear Rainfall Forecast Model Based on GNSS Observations and CAPE

A Novel Linear Rainfall Forecast Model Based on GNSS Observations and CAPE