Machine learning has recently emerged as a promising approach for predicting lightning activity by learning patterns from historical data. This study focuses on the prediction of lightning density using machine learning techniques over Northern India, region with high monsoon lightning activity. Models were trained using LIS/OTD 2.5° LRMTS and LRTS lightning data archived at GHRC as the target variable, with K-index, Convective Available Potential Energy (CAPE), Surface Sensible Heat Flux (SSHF), and 2-m air temperature (T2M), were sourced from the ERA5 reanalysis (ECMWF), as predictor variables for the period 2000-2011 and tested for 2012-2013. Three machine learning models, namely Decision Tree, Random Forest, and XGBoost, were developed separately for the monthly dataset using k-fold cross-validation with each year treated as one-fold. Before tuning, Random Forest (R2 = 0.7781) and XGBoost (R2 = 0.7482) outperformed the Decision Tree (R2 = 0.6283). Ensemble performance was greatly enhanced by hyperparameter tuning, with XGBoost achieving R2 = 0.84, followed by Random Forest (R2 = 0.82). Similar hierarchy was seen in baseline performance for the daily dataset, with XGBoost (R2 = 0.72) and Random Forest (R2 = 0.72) significantly outperforming the Decision Tree (R2 = 0.50), while a Long Short-Term Memory (LSTM) model was designed specifically for daily dataset achieving R2 = 0.62. These findings show potential of machine learning, especially ensemble approaches for lightning prediction over Northern India. To improve operational lightning forecasting across India's various climatic regions, future research may use hybrid NWP–ML frameworks, multi-step deep learning architectures like Transformers, and higher-resolution lightning observations.

Atmospheric moisture governs seasonal variability, but its strength and its linkage with rainfall over the complex topography of northeastern India remain unexplored. Precipitable Water (PW), representing the total column atmospheric moisture, is expected to increase with global warming. Using ERA5 reanalysis data (1990–2024), this study examines the climatology, vertical structure and coupling of PW and Total Precipitation (TP) over the Northeastern Region of India (NRI). It also analyzes Vertically Integrated Moisture Transport (VIMT) and links with climate modes. Seasonal variability in PW is pronounced, with minimum values during winter (10–20 mm) and peak values during the monsoon (55–70 mm). Trend analysis shows PW significantly increase during monsoon ( 0.08 mm yr−¹), with insignificant trends in other seasons. Monsoon PW reaches 64–71 mm over the NRI, whereas TP shows strong interannual variability ( 11–22 mm day−¹). Pre-monsoon PW and TP correlations are highest but weaken or become negative during the monsoon over parts of Assam. Seasonal PW-TP coupling is modulated by the El Niño–Southern Oscillation (ENSO) and the Indian Ocean Dipole (IOD), with stronger large-scale control during the monsoon and post-monsoon seasons. These findings provide insights for regional hydrological assessment and support climate adaptation over the NRI.

Lightning represents a significant natural hazard that causes substantial loss of life and property globally, particularly in tropical and subtropical regions experiencing intense convective activity. Accurate prediction of lightning occurrence and intensity remains a critical challenge for operational weather forecasting and hazard mitigation. Recent advances in numerical weather prediction have enabled the development of specialized lightning parameterization schemes that explicitly represent electrification processes within convective clouds. This study evaluates the performance of the WRF-ELEC lightning parameterization scheme alongside the Lightning Potential Index (LPI) diagnostic for predicting lightning activity over Odisha, India, a region highly susceptible to severe convective weather. Model simulations were validated against ground-based observations from the Indian Institute of Tropical Meteorology (IITM) Lightning Location Network for four convective events: 15 June 2019, 24 June 2020, 27 July 2020, and 8 August 2021. Quantitative verification using flash origin density (FOD) from WRF-ELEC revealed superior performance compared to LPI-derived lightning density, with lower aggregate mismatch counts (maximum: 12, mean: 1.06 versus maximum: 19, mean: 2.74), reduced percentage mismatch (maximum: 50

An intense thunderstorm accompanied by lightning strikes claimed the lives of more than 11 people in Jaipur, Rajasthan, on 11th July 2021. This study investigates the influence of different Initial Conditions (ICs) on the simulation of this lightning event using the Weather Research and Forecasting (WRF) model. Various datasets Global Forecast System GFS (0.25 degrees x 0.25 degrees), Global Data Assimilation System (GDAS)/ (FINAL) GDAS-FNL (0.25 degrees x 0.25 degrees), FNL (1 degrees x 1 degrees), and ERA5 (0.25 degrees x 0.25 degrees) were used as to assess their impact on model performance. The objectives of this study are to analyses the synoptic conditions associated with the thunderstorm and lightning event, evaluate the sensitivity of lightning simulations to different initial condition, and understand how these conditions influence the spatial and temporal distribution of lightning. The study also aims to assess how ICs affect the accuracy and predictability of lightning simulation. A comprehensive examination of meteorological factors during the onset of the monsoon season highlights the conditions that triggered this severe lightning event. Results show that between 1400 and 1600 UTC, when the event occurred, simulations using ERA5 outperformed FNL, GDAS-FNL, and GFS. ERA5 yielded the highest accuracy (0.67), along with improved scores for Probability of Detection (POD = 0.85), Critical Success Index (CSI = 0.53), and Heidke Skill Score (HSS = 0.36). The simulated lightning flash counts from ERA5 closely aligned with lightning observations over Rajasthan. Furthermore, the study evaluates the strengths and limitations of each ICs in representing key meteorological parameters such as temperature, humidity, precipitation, and lightning activity. Overall, the findings suggest that the ERA5 dataset provides the most reliable input for simulating precipitation and lightning over the region.

Flooding is one of the most catastrophic natural hazards in India, causing significant socio-economic losses annually. Despite the widespread application of GIS-based Analytical Hierarchy Process (AHP) methods for flood hazard assessment, limitations remain in capturing fine-scale hydrological variability. Addressing this gap, the present study advances existing frameworks by integrating additional hydrological indicators specifically the Normalized Difference Water Index (NDWI) and surface roughness to enhance the spatial precision of flood hazard mapping in the Damodar River Basin. . This study employs a GIS-based Analytic Hierarchy Process (AHP) within a multi-criteria decision-making (MCDM) framework to delineate flood hazard zones in the Damodar River Basin by assigning relative weights to key influencing factors. . Parameters such as elevation, slope, roughness, land-cover, drainage density, topographic wetness index (TWI), (NDWI), annual rainfall, and proximity to rivers were analyzed to evaluate the spatial distribution of flood hazard regions. Thematic maps of these parameters were derived using remote sensing data, including SRTM Digital Elevation Model (DEM) and other scientific datasets. The primary purpose of this study is to develop a robust GIS–AHP framework for flood hazard assessment in the Damodar River Basin, providing scientific evidence to guide targeted mitigation and policy interventions. Weighted values for factors were computed, with elevation (0.2099), slope (0.1894), and rainfall (0.1459) being the most critical contributors, while roughness (0.0265) had the least impact. The calculated consistency ratio (CR) of 0.0562, well below the threshold of 0.1, validates the robustness of the weight assignments. The results reveal that approximately 2.42% (1015 km2) of the basin is categorized as low hazard level, 31.01% (12,997 km2) as moderate hazard level, 56.25% (23,571 km2) as high hazard level, and 10.32% (4,326 km2) as very high hazard level. This analysis provides a critical foundation for targeted flood mitigation strategies and resource allocation, emphasizing the need to address high and very high hazard zones to minimize future flood impacts effectively.

A comprehensive analysis of basin dynamics for long-term sustainable water management is essential under changing climate scenarios. This study assesses future streamflow variations in the upper Godavari River Basin (GRB) using the Hydrological Engineering Center's Hydrological Modeling System (HEC-HMS). The Coupled Model Intercomparison Project Phase 6 (CMIP6) General Circulation Models (GCMs) rainfall datasets under Shared Socioeconomic Pathway (SSP) 2-4.5 and SSP5-8.5 serve as the primary drivers of streamflow projections. Model selection, a critical process, was guided by an extensive literature review that identified five optimal GCMs. To mitigate inherent biases in CMIP6 rainfall data, the Quantile Mapping (QM) technique was applied, revealing BCC-CSM2-MR as the most reliable model. Pre-processing of terrain, soil, and land use/land cover (LULC) data facilitated comprehensive characterisation of the watershed. ArcHydro and HEC-GeoHMS tools streamlined the generation of input files for the HEC-HMS model. The ModClark and Initial and Constant methods were adopted for the transform and loss methods, respectively. Model calibration (1988-1990) and validation (1997-1999) demonstrated satisfactory performance, thereby reinforcing the reliability of the projections. Future streamflow projections were derived using calibrated parameters for two temporal windows: near-future (2030-2059) and far-future (2070-2099). Comparative analysis with the historical period (19852014) revealed a noticeable increase in streamflow. Under SSP2-4.5, streamflow is projected to rise by 5.89% to 8.16%, while SSP5-8.5 suggests an increase of 2.78% to 7.63%. These findings underscore the anticipated hydrological shifts driven by climate change. This study enhances adaptation strategies by providing critical insights into streamflow alterations, encouraging informed decision-making for sustainable water resource management in the upper GRB.

This study presents a climatological analysis of hailstorms over India, with a focus on Maharashtra, which recorded the highest number of hail events during the 30-year period from 1994 to 2023. A total of 996 hailstorms were reported across India, with the pre-monsoon season being the most active (637 events), followed by the winter season (214 events), together accounting for 85.6

Natural hazards like thunderstorms or lightning can cause loss to the economy and lives, especially in major urban centers. A long-term analysis of pre-monsoon thunderstorm rainfall and lightning, as well as thermodynamic indices like Convective Available Potential Energy, George’s K-Index, and Total Totals Index, has been performed over the major urban hubs of India, i.e., Delhi, Hyderabad, Chennai, Bangalore, Mumbai, Pune, Kolkata, and Ahmedabad. For this purpose, the study utilized IMDAA and ERA5 datasets correspondingly for rainfall and thermodynamic indices for the period from 1980 to 2020, and TRMM and WGLC datasets for lightning observations for periods from 1999–2009 (PRE10) to 2010–2020 (POST10) respectively. For most cities, the quadri-decadal trend of pre-monsoon precipitation increased except for Kolkata and Ahmedabad, where a decrease was noticed. However, the trend over Chennai was realized to be significant through the Mann-Kendal test. There is an east–west contrast in the amount of pre-monsoon precipitation between the cities located in the eastern and western parts of the country. The spatial distribution indicates that rainfall has considerably increased in the boundary regions of the cities and the associated non-urban neighborhoods in recent times. This urban boundary contrast is more prominent in the case of Hyderabad, Delhi, and Pune, since rainfall shows an increase of > 25 mm in the boundary regions in the recent years. The spatiotemporal analysis of lightning shows an increasing trend in most of the cities, where it is statistically significant over Delhi, Hyderabad, and Kolkata in the POST10 period. For Kolkata, the trend is more profound with a considerable Sen’s slope value ( 0.04) compared to other cities. Most of the flashes are observed in the boundary regions of the cities rather than their center. The analysis also advocates KI and TTI to play a considerable role in the initiation of pre-monsoon thunderstorms compared to CAPE.

Accurate evaluation of cloud microphysical variables is essential for improving cloud parameterization and weather forecasting, yet obtaining high-resolution, spatially and temporally extensive data remains a challenge due to the limitations of in-situ measurements. The present study tries to address this gap by assessing existing equations for estimating vertically integrated liquid water content (VIL, kg/m2) from liquid water content (LWC, g/m3) using C-band dual-polarized Doppler Weather Radar (DWR) data from the India Meteorological Department (IMD) Jaipur station over 78 summer monsoon days in 2020–2022. A long-term climatological analysis (2003–2023) of total column cloud liquid water (TCCLW, kg/m2) from ERA5, liquid water cloud water content (LWCP, kg/m2) from MODIS, and rainfall data from IMD, IMERG, and GPCP datasets has been performed. VIL is computed as the vertical integral of LWC across atmospheric layers using four reflectivity-LWC (Z-LWC) relationships and one reflectivity-differential reflectivity (Z, ZDR-LWC) relationship from existing literature. The performance of empirically radar derived VIL has been evaluated by comparing with satellite-derived (MODIS) cloud liquid water path (LWP, kg/m2) and TCCLW. The results show that VIL values increase with rainfall intensity, leading to higher estimation errors. Among all relations tested, the hybrid equation (which includes Z and ZDR) consistently demonstrated superior performance, particularly during high-intensity rainfall events, with lower root mean square error (RMSE) and mean absolute error (MAE) values. The method also captured more detailed spatial patterns of liquid water distribution with reduced bias, making it the most reliable estimator. Despite limitations such as beam blockage and slight spatial shifts due to interpolation, the current study may provide a foundation for improving real-time precipitation forecasts and understanding cloud microphysics by incorporating polarimetric radar products. The future work may aim at refining the methodology through enhanced cloud-type-specific estimators.

Dust storms accompanied by lightning are a critical natural hazard that can result in significant loss of life and property. This study employs the Weather Research and Forecasting model coupled with Chemistry (WRF-CHEM) to investigate the interactions between dust storms and lightning over a 48-hour simulation period, using both dust-inclusive and dust-exclusive scenarios over Uttar Pradesh on 23rd May 2022. The WRF model without dust effectively simulated lightning but underestimated its intensity, while the WRF-CHEM model with dust-inclusive simulation captured both lightning intensity and the path and strength of the dust storm. However, it struggled to reproduce the spatial distribution of lightning events accurately. WRF-CHEM predicts over 110 lightning strokes under dust conditions with visibility of 1-2 km, while WRF underpredicts by nearly 60 % due to missing dust electrification; in extreme dust conditions (visibility <1 km), WRF-CHEM simulates 130 strokes, whereas WRF without dust predicts only about 60. The simulated lightning datasets were validated using observations from the WWLLN (World Wide Lightning Location Network) and the IITM Lightning Location Network. Additionally, model-derived parameters such as temperature, relative humidity, and Convective Available Potential Energy (CAPE) were validated against ECMWF ERA5 reanalysis data. The NOAA HYSPLIT model was also utilized to analyze the dynamics, trajectory of the dust storm and corroborate the WRF-CHEM simulations. The WRF-CHEM with dust ON outperformed the WRF without dust. This study highlights the intricate interactions between dust and lightning, contributing valuable insights into predicting thunderstorm and lightning events associated with dust storms, which is crucial for mitigating their adverse impacts.

Climate change poses severe challenges to agriculture, water resources, infrastructure, and livelihoods, increasing the frequency of hydrological extremes at the basin level. The Ramganga basin, a vital water source for millions, is highly vulnerable to these events. This study assesses CMIP6 models’ ability to simulate precipitation- and temperature-related hydrological extremes over the basin. Using 34 years of observed data (1981–2014) and CMIP6 Shared Socioeconomic Pathways (SSP370) precipitation scenarios, the study assesses hydrological extremes for the mid-future (2041–2070). Bias correction was performed using the Quantile Mapping (QM) ensemble method on the four best-performing CMIP6 models. Trends across 51 grid points were assessed using the Mann–Kendall test and Sen’s slope, while droughts were characterized using the Standardized Precipitation Index (SPI) at multiple timescales and run theory. Findings highlight an increase in consecutive wet days (CWD) historically, shifting to a statistically non-significant (p > 0.05) future trend. The PRCPTOT index indicates a statistically significant (p < 0.05) 90

Severe weather events such as thunderstorms are responsible for deaths and asset damage. Lightning discharge is the most common phenomenon in thunderstorms, which causes death and losses. This review shows the research that has been done on lightning over the Indian region. The severe events are analyzed and simulated using the Weather Research Forecasting model (WRF). The different microphysical schemes, such as WSM-6, NSSL-2, Morrison, and Thomson, have been mainly used for analyzing the events. The NSSL-2 performed well in simulating the lightning events for the lightning potential index (LPI) but has limitations. It is not able to produce cloud-to-ground and intercloud lightning as compared to observations. The lightning parameterization scheme works efficiently in providing CG and IC with some discrepancies. The model time step is critical for understanding the events that are integrated into microphysical schemes. The model simulated results have been validated using reanalysis datasets such as ERA5 and IMDAA. The lightning flash rates simulated by the model are validated using ground observational datasets such as IITM–LDN. The thunderstorm indices with optimal threshold can be used to predict the thunderstorm. Model skill scores like the true statistics score and Heidke skill score are used to assess a model’s accuracy. Recent studies on applications of AI/ML on lightning prediction show promising results.

The rising frequency of heat waves in India presents significant risks to public health, agriculture, and the economy. In March 2022, temperatures reached a record-breaking 33.10 degrees C, the highest in 122 years resulting in two major heat wave events: March 11-21 and March 26-31, which claimed 33 lives. This study delves into the impact of anthropogenic emission/aerosols and meteorological data assimilation on model-predicted surface meteorological variables using "Weather Research and Forecasting (WRF) model" coupled with Chemistry (WRFChem). Four distinct simulation scenarios namely, WRF, WRFDA (WRF with meteorological Data Assimilation), WRF-Chem, and WRF-ChemDA (WRF-Chem with meteorological Data Assimilation) were executed across the Haryana domain to assess the sensitivity of model outputs. Analyses within the WRFDA and WRF-ChemDA frameworks utilized a 6-hourly cyclic 3-Dimensional Variational (3DVAR) Data Assimilation (DA) of NCEP ADP Surface Observational Fields. Most critically, the incorporation of aerosols and DA techniques markedly improved forecasts of key meteorological variables, including 2 m Temperature (T2), 2 m Relative Humidity (RH2), Planetary Boundary Layer Height (PBLH), and Outgoing Longwave Radiation (OLR). Skill assessment metrics, including the Heidke Skill Score (HSS similar to 0.5), Accuracy (ACC >0.9), and Probability of Detection (POD similar to 1), demonstrate that WRF-ChemDA outperformed other models, especially during heat wave events. Conclusively, this study advocates for the meticulous selection of modeling approaches to accurately simulate heat wave events, ensuring that selected models adeptly capture the intricate dynamics and complexities of extreme temperature phenomena.

This study delves into the recent decline in rainfall observed in Northeast India (NEI). It investigates the factors driving this decrease, emphasizing the influence of climatic phenomena such as the El Niño-Southern Oscillation (ENSO) and the Indian Ocean Dipole (IOD), along with the broader impacts of climate change on regional precipitation patterns. The study aims to analyze temporal variations in Precipitation Indices (PI) over NEI, employing the ETCCDI precipitation indices like consecutive dry days (CDD), consecutive wet days (CWD), and total precipitation (PRCPTOT). PI are calculated using the ClimPACT2 software, facilitating the analysis of temporal variations. Furthermore, temporal trends experienced by the indices throughout the study period are comprehended through the application of the Mann-Kendall test. Findings indicate a significant increase in CDD alongside significant decreases in CWD and PRCPTOT, underscoring shifting precipitation dynamics in NEI.

Rainfall-runoff modeling requires a selection of a suitable hydrologic model for the determination of an accurate quantity. A better understanding of rainfall-runoff processes is important for stream flow generation within a river basin. This study investigate the long-term spatial-temporal trend of rainfall and streamflow using the hydrologic engineering center-hydrologic modeling system (HEC-HMS) model to understand the rainfall-runoff dynamics in the Banas River basin. Rainfall variability was analyzed for two periods: 1971-1995 and 1996-2020. Continuous wavelet transform was used to analyze the periodicity and the inter-seasonal relationship in rainfall. HEC-GeoHMS was used to generate the various inputs for the hydrologic modeling inputs for the HEC-HMS model. Initial constant loss and ModClark transform method were used to set up the HEC-HMS model for the study area. Calibration and validation of the model were performed using 3 years of observed data 2010-2012 and 2013-2015, respectively. Statistical model efficiency was checked using the coefficient of determination (R2), Nash-Sutcliffe efficiency, and root mean squared error to evaluate the performance of the HEC-HMS model. The finding indicated that the R2 values are 0.80 for calibration and 0.82 for validation periods, respectively, which are in good arrangement with the measured values. This study analyzes the relationship between rainfall and runoff, which will aid in proper and efficient water resource management. Rainfall-runoff modeling requires a suitable hydrologic model for the determination of an accurate quantity. In this study, spatial-temporal trends of rainfall and streamflow were analyzed using the HEC-HMS model to understand the rainfall-runoff dynamics in the Bisalpur catchment. Continuous wavelet transform (CWT) was used to analyze the periodicity and the inter-seasonal relationship. Results showed that the determined model statistical values were in good agreement with measured values. image

: This study investigates the use of various thunderstorm indices in predicting severe thunderstorms events during the monsoon season in four different regions in India. The research evaluates the performance of the prediction model using a model skill score and utilizes the Weather Research and Forecasting (WRF) model with the double moment microphysics scheme to simulate model cases. It also compares fifteen thunderstorm indices derived from the ERA5 dataset to identify the most effective index for predicting severe thunderstorms events. The results of this study show that in-corporating thunderstorm indices with model skill scores improves severe thunderstorms fore-casting in the monsoon season in India. The result revealed that determining the optimal threshold for each index is crucial in achieving accurate predictions. The study also highlights the importance of considering multiple indices rather than relying on a single index to predict severe thunderstorms events. The advance indices such as Energy Helicity Index (EHI), Supercell Composite Parameter (SCP), mainly works well with extreme severe thunderstorms. The simplistic indices can predict the weak or severe thunderstorm easily. The use of multiple thunderstorm indices can also help mete-orologists to make more accurate predictions, which can further enhance public safety. In conclu-sion, this study demonstrates the potential of incorporating thunderstorm indices with model skill scores like HSS and TSS and combinations of different skill scores in severe thunderstorms fore-casting during the monsoon season in India. Future research can build upon the findings of this study to develop more accurate and reliable severe weather forecasting models.

Climate change’s impact on lightning and thunderstorms is uncertain. This study evaluates the sensitivity of multiple Planetary Boundary Layer (PBL) parameterisation schemes within the WRF-ELEC model for simulating a severe lightning and thunderstorm event in Bihar on 25 June 2020. The aim is to understand how these schemes affect lightning and thunderstorm intensity. The model was integrated for 54 h at 0000 UTC on 24 June 2020 using 6-hourly NCEP FNL Operational Global Analysis data at 1° × 1° resolution over Bihar with double nested domains of 9 km (D1) and 3 km (D2). It effectively captures the peak lightning and thunderstorm activity from 0000 to 0900 UTC on 25 June 2020, significantly impacting certain regions. The study utilised ERA5 and IMDAA reanalysis datasets and NASA GPM IMERG daily data to analyse the event and assess the model’s performance. Among the PBL schemes tested, ACM2, BouLac, SHsa, and MRF exhibit robust performance. Flash Origin Density (FOD) patterns broadly match observations, although occasional discrepancies occur in southern Bihar. Convective Available Potential Energy (CAPE) and precipitation (mm) analysis reveals anticipated trends. Statistical scores highlight strong performance by ACM2, UW, and GBM schemes in POFD/FAR. The MRF scheme excels in POD/Hit Rate, and the UW scheme achieves the highest score for 24-h accumulated total precipitation. HSS and GSS/ETS underscore the superior performance of the UW and GBM schemes. This study offers insights into lightning and thunderstorm simulations over Bihar with diverse PBL parameterisation schemes in the WRF-ELEC model.

This study investigates the performance of the Advanced Research Weather Research and Forecasting (WRF-ARW) model in simulating two severe hailstorm events. The simulation period covers the post-monsoon and pre-monsoon hailstorm events of 26 December 2022 in Assam and 19 May 2022 in Bihar, respectively, with a continuous model integration of 48 h. The WRF model is integrated on a double nested domain of 9 and 3 km horizontal resolutions using the Milbrandt 2-moment microphysical and Kain Fritsch cumulus schemes. Critical analyses include the mixing ratio of hydrometeors, maximum reflectivity, wind magnitude and direction, temperature contours, and relative humidity. The results were validated using ERA-5 reanalysis data. Instability indices such as Convective Available Potential Energy (CAPE), Total Totals Index (TTI), and Significant Hail Parameter (SHIP) are evaluated to assess the hailstorm environment. The hail mixing ratio peak between altitudes 600–700 hPa suggests a significant presence of hail in the mid-troposphere. A high-level jet was observed during the Assam event. The SHIP value calculated from the model output for the Assam event was 0.1, indicating a non-significant hail environment. In contrast, for Bihar, the SHIP value was 2.0, suggesting a significant hail environment. Despite a CAPE value of approximately 500 J/kg for the Assam event, typically not indicative of severe thunderstorms, hail occurred. The CAPE value of around 3000 J/kg in Bihar indicated severe thunderstorm conditions. The WRF could reasonably simulate both hailstorm events with the selected configuration.

Severe thunderstorms pose significant challenges to communities and infrastructure, with timely and accurate prediction increasingly difficult. This study compares a control run (CNTL) with a 3DVAR run by assimilating upper-air observational data, focusing on severe thunderstorm events in Bihar on 25th June, 2020, and West Bengal on 07th June, 2021. It assesses the impact of data assimilation on severe thunderstorm initiation and intensification, considering factors such as Convective Available Potential Energy (CAPE), Convective Inhibition (CIN), rainfall, wind, and relative humidity. Using 0.25° * 0.25° NCEP GDAS FNL data, the model was integrated for 48 h with eight 6-hourly assimilation cycles for 3DVAR for both cases. Bihar exhibited a CAPE of 1800 J/kg, while West Bengal recorded 5000 J/kg, indicating severe thunderstorm conditions. Model performance metrics, including Accuracy (ACC), Equitable Threat Score (ETS), Heidke Skill Score (HSS), Probability of Detection (POD), True Skill Statistic (TSS), Critical Success Index (CSI), and False Alarm Ratio (FAR) were utilised. Assimilating upper air observational data notably improved model-simulated results, rectifying issues such as overestimated rainfall by 20–40 mm in Bihar, correcting overestimations of 100 mm, and underestimating 80–100 mm in northern and southern parts of West Bengal. Additionally, 3DVAR reduced spatial and temporal lags in both cases. ACC approached 1 for Bihar rainfall, while FAR neared 0 for West Bengal CIN during event time in the 3DVAR run. Each model runs showed similar patterns during peak intervals, but the 3DVAR approach generally exhibited closer alignment with the observation, emphasising its effectiveness in enhancing severe thunderstorm prediction accuracy.

A reliability-based assessment of heavy rainfall events can provide insights for developing appropriate measures and predictions to control flood risk. One of the contemporary challenges is to model heavy rainfall events, which depend on different synoptic conditions and model configurations. In this study, the regional mesoscale Weather Research and Forecasting (WRF) model was used to evaluate the applicability of different parameterization schemes for simulating rainfall events over the Mahi River basin between 22 and 26 August 2020 with the maximum rainfall occurring on 23 August 2020. A high-resolution triple-nested (27, 9, and 3 km) WRF model was configured, and data for initial boundary conditions were obtained from NCEP/NCAR FNL with a resolution of 1 degrees x 1 degrees (six-hourly). The model results were evaluated using gridded rainfall of ECMWF Reanalysis v5 (ERA5) and India Meteorological Department (IMD) at a spatial resolution of 0.25 degrees x 0.25 degrees. Further to verify the model output, GPM-IMERG was used to compute the skill score (bias, threat score, probability of detection, and accuracy) metrics at 24 hourly and 3 hourly scales. Eight sensitivity tests were conducted to identify the optimal sets of microphysics (WSM6 and Goddard), cumulus (Kain and BMJ), and planetary boundary layer (YSU and ACM2) parameterization schemes for rainfall simulation of the selected events. The Kain-Goddard-YSU and KainGoddard-ACM2 parameterization schemes demonstrated the optimal performance for simulating the rainfall event. Relative humidity and reflectivity were used for analyzing moisture and cloud brightness temperature, respectively. The model score revealed that the Kain-Goddard scheme outperformed other schemes in case of heavy precipitation, with a threshold of 60 mm/day over the Mahi basin. The findings of this study showed that the simulation of heavy rainfall events with a set of selected combinations of parameterization schemes can reproduce the structure of convective organization as well as prominent synoptic features associated with heavy rainfall events over the Mahi basin. Moreover, this information may help to better understand the future forecast trend of precipitation and explore solutions for sustainable water management to support decision-makers and operational water managers.