Some of the most intense thunderstorms on Earth occur in the Hindu Kush Himalaya (HKH) region of South Asia (Zipser et al. 2006; Romatschke et al. 2010)—where many organizations lack the capacity needed to predict, observe, and/or effectively respond to the threats associated with high-impact convective weather. Among the hazards include tornadoes, damaging straight-line winds (known as nor’westers in the HKH region), large hail, and flash flooding, which typically peak in the pre-wet-monsoon season (approximately March–May; Das et al. 2014). The HKH region south of the Tibetan Plateau experiences favorable conditions for frequent severe weather occurrence during the premonsoon when enhanced mid- to upper-level tropospheric flow combines with increasing low-level moisture and instability (sometimes extreme, with CAPE exceeding 5,000 J kg−1). Mesoscale foci such as outflow boundaries off the Khasi Hills of northeast India and eastward-propagating drylines from the Indian subcontinent serve to initiate severe convection and also enhance low-level storm-relative helicity and storm organization, as discussed in Bikos et al. (2016). Previous studies have documented a disproportionately large number of casualties associated with intense thunderstorms in HKH (e.g., Bikos et al. 2016; Bista 1989; Dewan et al. 2017). The region is particularly vulnerable to lightning, with numerous casualties and fatalities each year (e.g., Dewan et al. 2018, 2017; Holle et al. 2019, 2018), including hundreds of fatalities during 2018–19. As a result, the Government of Bangladesh has recently declared lightning as a national disaster. Therefore, an important objective of our project’s initiative is to increase situational awareness and prediction capacity of these hazards to help reduce casualties in the HKH region.
Since meteorological observations are relatively sparse and there is limited access to operational weather radar datasets in Nepal and Bangladesh, it can be challenging to assess environmental conditions and resultant storm modes responsible for damaging thunderstorms across this region. Satellite observations, synoptic-scale reanalysis, and high-resolution regional modeling have been used to develop and refine conceptual models that explain deep convective storm patterns and behaviors in the region (e.g., Hirose and Nakamura 2005; Houze et al. 2007; Medina et al. 2010). Convection-allowing model (CAM) simulations have been used to investigate the mesoscale environment and resultant storm modes associated with high-impact tornadoes, hailstorms, and flash flooding in Nepal, Bangladesh, and northeast India (Litta et al. 2012; Bikos et al. 2016; Karmakar et al. 2017; Das et al. 2020; Meyer et al. 2021). These studies demonstrate that even CAMs at ∼4-km grid spacing can sufficiently simulate the key ingredients responsible for severe convective events, including Nepal’s first documented tornado (Meyer et al. 2021).
The purpose of this study is to evaluate a multifaceted set of satellite and numerical weather prediction (NWP) tools focused on deep convective storms that plague the HKH region. Coined the High-Impact Weather Assessment Toolkit (HIWAT), this utility was developed under the auspices of a NASA Applied Sciences Capacity Building project, SERVIR, to improve the capacity of extreme weather services in the HKH region (Gatlin et al. 2021). The short-term NWP strategies employed within HIWAT include developing real-time regional CAM ensemble guidance, similar to previous Spring Experimental Forecast Program campaigns at the NOAA Hazardous Weather Testbed (e.g., Clark et al. 2012), as well as the retired National Center for Atmospheric Research experimental ensemble prediction system (Schwartz et al. 2015, 2019). The HIWAT generates daily NWP simulations over a targeted area of the HKH region, focusing on Nepal, Bangladesh, and northeastern India. The system consists of ensemble-based forecast products similar to the High Resolution Ensemble Forecast (HREF) at the National Centers for Environmental Prediction (NCEP) Storm Prediction Center (Roberts et al. 2020; Jirak et al. 2012; Roberts et al. 2019) to offer guidance on the most likely areas of severe weather hazards for a 1–2-day forecast period. The observational components of HIWAT include satellite products from the Global Precipitation Measurement (GPM) mission (Hou et al. 2014; Skofronick-Jackson et al. 2017) and land-cover processing techniques (Bell and Molthan 2016; Bell et al. 2019; Molthan et al. 2014) that have been primarily used for case studies and validation of the HIWAT forecasts.
This article presents the implementation of a real-time ensemble forecasting experiment over the HKH region during the premonsoon months of 2018 and 2019, case studies for selected severe weather events, and quantitative validation of precipitation and lightning ensemble forecasts during 2018 and 2019. The system has been evaluated by national hydrometeorological agencies in the HKH region and found to be valuable in experimental operations. The paper is organized as follows: the NWP modeling platform and ensemble configuration are presented, the product suite and verification methods are described, results of severe weather events and precipitation/lightning verification are presented, and the paper concludes with a summary and discussion of transition activities to HKH hydrometeorological agencies.
NWP model and ensemble configuration
A virtual computing cluster in a cloud-like environment was established to support a regional CAM ensemble over south-central Asia. We applied the Unified Environmental Modeling System (UEMS; available online at http://strc.comet.ucar.edu/software/uems/) as the NWP software for an experimental forecasting experiment during spring 2018 and 2019. Maintained by the NOAA/NWS Science Operations Officers/Science and Training Resource Center, the UEMS consolidates the community Weather Research and Forecasting (WRF) NWP modeling system (Powers et al. 2017; Skamarock et al. 2008) into a streamlined open-source package for quickly and easily installing, configuring, and executing the WRF Model on a Linux-based operating environment, which is ideally suited to a cloud computing infrastructure.
The ensemble model configuration had an objective to provide 2 days of real-time forecast guidance of premonsoon thunderstorm hazards via a CAM nested-grid configuration. Given the available computational environment, the scientific/real-time objectives of the project, and the results of several studies that suggest diminishing returns with increased CAM ensemble sizes/sources (Clark et al. 2011; Schwartz et al. 2014; Jirak et al. 2016), we chose a 12-member ensemble system as a viable experimental solution. Each ensemble member ran separately on its own dedicated 32-core node, with auto postprocessing conducted on a dedicated “head” node for a 13-node computational configuration.
The model domain was designed to provide CAM guidance in a nested grid covering Nepal, Bangladesh, Bhutan, and northeastern India. The resulting 12-/4-km mesh grid is shown in Fig. 1, with configuration details common to each ensemble member summarized in Table 1. We varied both the initial/boundary conditions and physical parameterizations for achieving spread in the ensemble system. By varying select parameterization schemes, we can also glean information on the importance of certain parameterized physical processes in the complex terrain of the Himalayas. We varied the planetary boundary layer (PBL) and microphysics (MP) parameterization schemes to account for important physical processes leading up to and following convective initiation.
Grid domain d01 (12-km horizontal grid mesh) and nested domain d02 (4-km grid mesh) over south-central Asia for the WRF/UEMS ensemble system during the 2018 and 2019 forecasting experiments. Configuration details are summarized in Table 1.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
Select details of the nested grid configuration common to each ensemble member. Parameterization references are scale-aware Kain–Fritsch (Kain 2004; Zheng et al. 2016) and Rapid Radiative Transfer Model-Global (RRTM-G; Iacono et al. 2008).
A daily initialization at 1800 UTC and a 48-h integration length with hourly output frequency was applied to 1) provide several hours of model spinup prior to local sunrise (∼0000 UTC) leading into the first diurnal heating cycle, and 2) capture two full diurnal heating cycles for day-1 and day-2 thunderstorm guidance. The first ensemble member was initialized with the NCEP/Environmental Modeling Center (EMC) Global Forecast System (GFS; Han and Pan 2011; Zhou et al. 2017) while members 2–12 were initialized with arbitrarily selected members of the EMC Global Ensemble Forecast System (GEFS; Guan et al. 2015, 2019; Li et al. 2019; Zhou et al. 2017; Zhu et al. 2018), as indicated in Table 2. While suboptimal due to errors associated with 6-h GEFS forecasts, we expedite the ensemble system product generation by using the previous cycle of the GFS/GEFS (i.e., 1200 UTC) for initial and boundary conditions for the UEMS/WRF runs. Future improvements could include implementing a regional data assimilation and perturbation methodology to improve initial conditions of the ensemble system.
Matrix of initial/boundary conditions, and planetary boundary layer (PBL) and microphysics (MP) parameterization schemes comprising the 12-member ensemble system. PBL schemes: YSU (Yonsei University; Hong et al. 2006), MYJ (Mellor–Yamada–Janjić; Janjić 1994, 2001), and MYNN2 (Mellor–Yamada–Nakanishi–Niino Level 2.5 closure; Nakanishi and Niino 2009). MP schemes: Godd (Goddard 6-class cumulus ensemble with graupel; Lang et al. 2011, 2007; Tao et al. 2003, 2014), Lin 6-class [Lin et al. (1983); Thompson from 1 to 28 Mar 2018 (Thompson et al. 2008)], WSM6 [WRF Single-Moment 6-class (Hong and Lim 2006)], and Morr [Morrison double-moment; Morrison et al. (2009)]. Ensemble members are shown in bold font, and initial conditions are shown in italics. We initially ran with the Thompson microphysics scheme from 1 to 28 Mar 2018, but the Thompson ensemble members became consistently unstable with deep moist convention, so we changed to Lin microphysics on 29 Mar 2018, onward.
Product suite and model validation
Environmental indices, proxy fields, and ensemble products.
We utilized the open-source Sounding and Hodograph Analysis and Research Program in Python (SHARPpy; Blumberg et al. 2017; Halbert et al. 2015) to generate multivariate composite indices from the first ensemble member to assess the environmental conditions favoring severe convection (e.g., super-cell composite parameter, significant tornado parameter, etc.). Following Kain et al. (2010), ensemble products were developed based on proxy simulated fields representing specific convective hazards. The ensemble suite included derived products similar to CAM ensembles in the continental United States (Clark et al. 2012; Schwartz et al. 2015; Roberts et al. 2019), such as postage stamp displays, ensemble mean, minimum and maximum spread, paintball plots, neighborhood probability (Schwartz and Sobash 2017), and probability matched mean (PMM; Ebert 2001; Clark 2017). For computing neighborhood probabilities, we applied a neighborhood box with a 20-km half-width to convective hazards with small spatial footprints, such as updraft helicity, derived lightning flashes, vertically integrated graupel cores, and localized regions of high 10-m winds. All of the products were generated at hourly intervals and 24-h daily temporal composites, with a summary provided in Table 3. In addition, all real-time and archived ensemble and deterministic product imagery from 2018 to 2022 were made available on the HIWAT Model Viewer at https://hmv.servirglobal.net/ (see data availability statement at end of this manuscript).
Multiparameter environmental indices, model convective proxies, and ensemble fields implemented as part of the HIWAT product suite.
Validation of events, precipitation, and lightning.
Validation of damaging thunderstorm events is highly dependent on an accurate compilation of damage reports from the field, which is a work in progress for the HKH region. We consulted available media (typically local and regional news outlets) and corresponded with collaborators in Nepal, India, and Bangladesh to document observed casualties and damage reports for select events (Gatlin et al. 2021). However, accurately documenting severe weather events and associated casualties remains an ongoing challenge in the HKH region.
Quantitative precipitation forecasts (QPF) were validated for individual ensemble members, the ensemble mean, and PMM against the final research-quality version 06B of the Integrated Multi-satellitE Retrievals for GPM (IMERG) half-hourly rain-rate product (Huffman et al. 2020; Tan et al. 2019). Half-hourly precipitation rates were combined into hourly quantitative precipitation estimates (QPE), then summed across accumulation intervals to compute skill scores using traditional contingency table verification methods. The Model Evaluation Tools (MET; Brown et al. 2020) verification package was utilized to compare the model QPF to IMERG QPE using a 20-km neighborhood window consistent with that applied to generate the ensemble probabilities. To benchmark the HIWAT QPF verification scores, we compared validation results against 12 GEFS members. Numerous metrics are available from the MET output, but we focused on Heidke skill score (HSS) for brevity’s sake, since this metric conveys model accuracy by taking into account correct forecasts due to random chance. Forecasts demonstrate skill over random chance when HSS is above 0, with perfect skill given by an HSS of unity.
Total lightning flashes from the Earth Networks Total Lightning Network (ENTLN) dataset (Bui et al. 2016; Heckman et al. 2014; Lapierre et al. 2019) served as ground truth for validating the lightning forecast algorithm (LFA; McCaul et al. 2009, 2020) simulated total lightning flash rates of each ensemble member. The ENTLN flashes were binned at ±10 min of the top of each hour, then gridded onto the 4-km model nested grid. The LFA flashes from each ensemble member were verified in a similar manner as QPF using a 20-km neighborhood window. However, given the uncertainties associated with both model-predicted lightning and the ENTLN detection efficiency in the region (Lapierre et al. 2019), we present only verification results for the occurrence of any lightning activity.
Results of 2018 and 2019 forecasting experiments
Over 300 people perished as a result of premonsoon intense thunderstorm hazards in India, Bangladesh, and Nepal during the 2018 experiment, primarily due to straight-line convective winds and lightning strikes (refer to reporting in Gatlin et al. 2021). The 2019 premonsoon season was generally less severe than 2018, but still experienced numerous casualties associated with a rare tornado event in south-central Nepal on 31 March 2019 (Shrestha et al. 2019), followed by numerous isolated lightning casualties, largely in Bangladesh. Additionally, Cyclone Fani brought numerous fatalities associated with damaging winds and flooding as it moved across the coastal areas of the northern Bay of Bengal into Bangladesh during early May 2019.
Case studies of significant events.
30 March 2018 large hail and damaging straight-line winds.
The 30 March 2018 event exhibited a classic synoptic setup for severe weather across the HKH region. The combination of a warm and moist southerly low-level flow with strong to extreme instability, an elevated mixed layer and capping inversion advecting eastward from the Indian Subcontinent, strong deep-layer wind shear, and mesoscale outflow boundaries can lead to organized intense convection in the region (see Bikos et al. 2016; Dewan et al. 2018; Mannan et al. 2015; Yamane et al. 2013, 2010a,b; Yamane and Hayashi 2006). Also, a common occurrence is that midlevel wind maxima (∼700–500-hPa layer) are diverted around the Himalayas due to the physical barrier of the high terrain, thereby enhancing the deep-layer directional and speed shear to the south of the Tibetan Plateau.
Select environmental fields in the 15-h forecast of ensemble member 1 (Fig. 2) summarize the midafternoon environment at 0900 UTC 30 March that was conducive to organized severe thunderstorm activity. A deep upper-level low resided over the Tibetan Plateau along with a substantial southwesterly subtropical jet streak at 200 hPa exceeding 80 kt (1 kt ≈ 0.51 m s−1) extending from central India across Bangladesh, Myanmar, and into southern China (Fig. 2a). A classic setup existed where the 500-hPa wind speeds were enhanced to the south of the Himalayas (Fig. 2b). At 850 hPa (Fig. 2c), a relatively weak southwesterly low-level jet ∼20+ kt occurred across Bangladesh transporting low-level moisture from the Bay of Bengal. At the surface, rich moisture with 2-m dewpoints exceeding 24°C resided near the Bangladesh coast, with a lower dewpoint maximum above 18°C extending from northern Bangladesh into southeastern Nepal (Fig. 2d). A surface low center resided over eastern India with a trough/dryline extending south from the low pressure. The resulting surface-based convective available potential energy (CAPE; Fig. 2e) exceeded 3,000 J kg−1 across portions of northern Bangladesh and northeastern India, reaching 5,000+ J kg−1 over the southeastern Indian coast. The low-level shear given by the 0–3-km storm-relative helicity (SRH) was enhanced to the east and northeast of the surface low center across northern Bangladesh into southeastern Nepal (Fig. 2f). This combination of strong instability, 150–350 m2 s−2 low-level SRH, and strong wind shear favored intense organized convection across the region. Based on the simulated midafternoon supercell composite, significant tornado, and derecho composite parameters (not shown), there was an elevated threat for supercells with large hail and damaging straight-line winds across far southeastern Nepal and much of northern and central Bangladesh.
WRF/UEMS 15-h forecast environmental variables valid at 0900 UTC 30 Mar 2018 from ensemble member 1, which serves as a deterministic output solution in HIWAT. (a) Outer 12-km domain d01 200-hPa geopotential heights (m; dashed red lines), wind barbs (kt), and color-shaded wind speeds (kt). (b) As in (a), but at 500 hPa. (c) As in (a), but at 850 hPa (blue circle denotes region of weak low-level jet). (d) Nested 4-km domain (d02) 2-m dewpoint (°C), mean sea level pressure (hPa; red dashed lines), 10-m wind barbs (kt), and position of surface low pressure and accompanying dryline (red dashed lines). (e) Domain d02 surface convective available potential energy (CAPE; J kg−1; color shaded) and convective inhibition (CIN; J kg−1; contours). (f) Domain d02 0–3-km storm relative helicity (SRH; m2 s−2; shaded according to scale).
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
The Advanced Himawari Imager observations from Japan’s Himawari-8 geostationary satellite (Bessho et al. 2016) and available storm reports in Fig. 3 show that the most intense convection tracked southeastward across northern and central Bangladesh between 0700 and 1100 UTC (Figs. 3a–c). The discrete nature of the intense thunderstorms on the Himawari true color enhanced imagery (Miller et al. 2016; Murata et al. 2018) is evident based on the distinct overshooting tops, breaks between high anvil cloud decks (especially before 1100 UTC), and sharp clearing outside of the storms, suggesting a dominant supercell convective mode. Storm reports in Fig. 3d show that predominantly large hail reports occurred across far northern and west-central Bangladesh, along with a high wind report in northeastern Bangladesh. Media reports indicated large hail (up to 7-cm diameter) penetrating roofs of homes across northern Bangladesh.
Himawari-8 true color enhanced imagery centered over Bangladesh from 30 Mar 2018 at valid times (a) 0700, (b) 0900, and (c) 1100 UTC. (d) Storm reports of large, damaging hail (blue markers) wind damage (red markers), and lightning fatalities (orange markers) collected from available media reports. Himawari-8 images courtesy of the Japan Meteorological Agency. Storm reports not considered comprehensive.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
The ensemble forecast initialized on 1800 UTC 29 March captured fairly well the areas of lightning, large hail, and damaging convective winds observed in the 30 March event. The postage stamp view of the 12 ensemble members of composite LFA during the first 24 h of the forecast (Fig. 4) show a consistent pattern of widespread simulated total lightning across southeastern Nepal, much of central and northern Bangladesh, and portions of northeastern India. A more conditional lightning threat was indicated by the ensemble members across the southeast India coast, where forecast lightning activity was more scattered and not present in all ensemble members. The results of Fig. 4 suggest that the microphysics schemes play more of a role in determining the areal coverage of total lightning compared to the choice of PBL scheme, as seen by the greater overall coverage with the Goddard and Morrison schemes compared to the others (see McCaul et al. 2020). The resulting day-1 20-km neighborhood probability of lightning activity at any time during the first 24 h (Fig. 5a) indicate probability exceeded 90% across most of the northern half of Bangladesh, northeast India, and southeast Nepal. Observed total lightning activity composited during the same 24-h period from ENTLN (Fig. 5b) show a northwest–southeast pattern of lightning activity from Nepal into Bangladesh and northeastern India, quite similar to the pattern and orientation of the highest HIWAT day-1 lightning probabilities. The more conditional lightning threat along the southeast India coast near Kolkata did not verify in the observed ENTLN, which at the time had a detection efficiency ∼50%–60% in this region of India and western Bangladesh (Lapierre et al. 2019).
Postage stamp composite summary of the day-1 lightning forecast algorithm (LFA) temporal maximum total lightning flash rates at every model grid point during forecast hours 1–24, for the 12 WRF Model ensemble members initialized at 1800 UTC 29 Mar 2018. Variations in microphysics scheme are given along the columns, while planetary boundary layer schemes are displayed along rows. Initial and boundary condition data sources are listed below each panel along with the maximum LFA value simulated during forecast hours 1–24.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
(a) Day-1 summary of probability (%) of any lightning occurring within 20 km of a location during forecast hours 1–24, for the WRF Model ensemble forecast initialized at 1800 UTC 29 Mar 2018. (b) Total lightning flashes detected by the Earth Networks Total Lightning Network (ENTLN) during the same time frame of the ensemble probability map, spanning 1800 UTC 29 Mar–1800 UTC 30 Mar 2018. Reference cities/landmarks and sounding sites are denoted by stars and circles, respectively.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
The composite day-1 probability maps for the other convective hazards on 30 March are presented in Fig. 6. The threat for general deep convection is represented by probability of simulated reflectivity exceeding 50 dBZ (Fig. 6a), which indicated a high threat for deep convection across much of the same region as the lightning threat in Fig. 5a. The probability of 10-m wind speeds exceeding 50 kt (∼25.7 m s−1; Fig. 6b) had generally lower probabilities, with the highest probability across eastern Bangladesh not far from the observed wind damage report in Fig. 3d. The hail proxy threat given by the total column graupel exceeding 40 kg m−2 showed the highest probabilities in two areas across central and northern Bangladesh (Fig. 6c), closely corresponding to the two areas of clustered large hail reports in Fig. 3d. Similarly, the highest probabilities of updraft helicity exceeding 100 m2 s−2 occurred across central and northern Bangladesh, and along an axis extending from southeast Nepal into northeast India. Oftentimes the updraft helicity threat closely corresponds to the total column graupel (hail) threats because long-lived, rotating convective updrafts (i.e., mesocyclones or supercell thunderstorms associated with updraft helicity) often favor the growth of large hail aloft in thunderstorms (Knight and Knight 2001). These temporal daily composited probabilities of thunderstorm hazards can subsequently be examined in hourly forecasts to determine the most likely timing of each hazard, albeit limited by the skill of the overall ensemble system.
Day-1 probability summaries (%) of convective hazards occurring within 20 km of a location during forecast hours 1–24, for the WRF Model ensemble forecast initialized at 1800 UTC 29 Mar 2018: (a) composite reflectivity exceeding 50 dBZ, (b) 10-m wind speed exceeding 50 kt, (c) total column graupel exceeding 40 kg m−2, and (d) 2–5-km updraft helicity exceeding 100 m2 s−2. See text for details regarding specific convective hazards. Reference cities/landmarks and sounding sites are denoted by stars and circles, respectively.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
17 April 2018 Kolkata severe windstorm.
On 17 April 2018, a severe windstorm associated with intense convection affected Kolkata, India, during the evening hours. Numerous fatalities and injuries occurred associated with downed trees from the severe thunderstorms in the Bengal state of India, as observed wind gusts peaked at 98 km h−1 (∼27 m s−1; HT Correspondent 2018). The damage caused by this storm was observed first-hand by several coauthors of this manuscript, who traveled to Kolkata about a week after the storm and noted numerous billboards destroyed, as well as extensive large tree limbs downed across the city. The Himawari-8 combined visible–IR imagery (Fig. 7) depicts the development of scattered convection to the northwest of Kolkata at 1000 UTC (Fig. 7a), which tracked southeastward over the next few hours (Figs. 7b–d). The most intense of the storms is evident over Kolkata at 1200 and 1300 UTC given by the bright white in Figs. 7c and 7d associated with cold cloud-top temperatures in the band-13 IR imagery.
Himawari-8 visible+IR combined images from the 17 Apr 2018 Kolkata, India, severe windstorm, valid at (a) 1000, (b) 1100, (c) 1200, and (d) 1300 UTC. Images courtesy of the Japan Meteorological Agency.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
The daily HIWAT probability forecasts of 10-m wind speed exceeding 40 kt (∼20.6 m s−1; Figs. 8a,b) show that the day-2 forecast initialized at 1800 UTC 15 April was more aggressive in depicting higher probabilities of strong convective winds across southeastern India compared to the day-1 forecast initialized at 1800 UTC 16 April. However, despite some decrease in probability from day 2 to day 1, the maximum 10-m wind speed from any ensemble member at 1100 UTC (Figs. 8c,d) depicts the potential for localized 10-m winds of 60+ kt (31 m s−1) near Kolkata in both ensemble runs, comparable to the observed intensity. The probability patterns for strong 10-m wind speed and maximum forecast values suggest a conditional threat of damaging winds dependent on the development of scattered convection.
Comparison of day-2 vs day-1 probability summaries (%) of 10-m wind speed exceeding 40 kt within 20 km of a location during (a) day-2 forecast hours 25–48 for the ensemble initialized at 1800 UTC 15 Apr 2018, and (b) day-1 forecast hours 1–24 for the ensemble initialized at 1800 UTC 16 Apr 2018. Comparison of day-2 vs day-1 ensemble maximum from any member 10-m wind speed valid at 1100 UTC 17 Apr for the (c) 41-h forecast from the 15 Apr initialization, and (d) 17-h forecast from the 16 Apr initialization. Reference cities/landmarks and sounding sites are denoted by stars and circles, respectively.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
The hourly progression from 0900 to 1200 UTC of PMM composite reflectivity and probability of 10-m winds exceeding 40 kt from the 15 April ensemble forecast is presented in Fig. 9. The composite reflectivity PMM represents the intensity resampling of the ensemble mean composite reflectivity from the combined probability density function of the individual ensemble members, thereby building back amplitude that tends to be washed out during the averaging process. Scattered convection organized and intensified across southeast India into southern Bangladesh in the reflectivity PMM between 0900 and 1200 UTC (Figs. 9a,c,e,g) as the simulated storms moved southeastward toward Kolkata and the Bay of Bengal coast. The relatively high probabilities of 10-m wind speeds exceeding 40 kt accompanied the reflectivity PMM evolution (Figs. 9b,d,f,h), with the highest probability indicative of the most intense simulated convection occurring west of Kolkata. Overall, the nature of the convective mode and character were well represented by the ensemble up to two days in advance for this high-end, relatively isolated convective event, albeit with some decreased skill in the day-1 forecast.
Hourly composite reflectivity (a),(c),(e),(g) probability matched mean (PMM; dBZ) and (b),(d),(f),(h) probability of 10-m wind speed exceeding 40 kt within 20 km of a location (%) from the ensemble initialized at 1800 UTC 15 Apr 2018. The 39–42-h forecasts are shown, valid 0900–1200 UTC 17 Apr. Reference cities/landmarks and sounding sites are denoted by stars and circles, respectively.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
31 March 2019 south-central Nepal tornado.
A rare documented tornado, rated as EF3 (wind speed between 180 and 265 km h−1) on the Enhanced Fujita scale, impacted the Bara and Parsa districts in south-central Nepal during the evening of 31 March 2019 (Meyer et al. 2021). Thirty people were killed and over 1,150 injured (Shrestha et al. 2019) when a tornado tracked on a northwest–southeast path for about 35 km, leaving a damage scar across the region that is clearly visible from low-Earth orbit in the Sentinel-2 imagery shown in Fig. 10. In addition, six people were killed along with extensive property damage over western and central Bangladesh associated with a documented nor’wester (NIRAPAD 2019). The environmental conditions favoring severe convective winds and even tornado potential were depicted by the HIWAT deterministic (ensemble member 1) 36–42-h forecasts, as shown in Fig. 11. A strong midlevel wind maximum from a northwesterly direction at 700 hPa of ∼40–50 kt is seen in the 36-h forecast valid 0600 UTC 31 March, several hours before the tornado occurred, and positioned unusually close to the Himalaya mountains across extreme southern Nepal and northeastern India (Fig. 11a). At the surface, a broad southeasterly flow advected a relatively moist air mass from the Bay of Bengal into southern and eastern Nepal, with 2-m dewpoint temperatures of 15°–18°C+ (Fig. 11b). The southeasterly flow underlying the northwest flow at midlevels led to large directional shear and a moderately strong low-level environmental helicity that contributed to the potential for rotating thunderstorms. The combination of moderate instability [most unstable CAPE (MUCAPE) ∼1,500–2,500 J kg−1; Fig. 11c] and strong low- to midlevel shear led to a risk of rotating (supercell) thunderstorms across south-central Nepal, as seen in the SCP forecast from ensemble member one 42-h forecast valid at 1200 UTC 31 March (Fig. 11d), near the time of the tornado event. However, the forecast STP from the first ensemble member remained less than one leading up to the tornado event (not shown).
European Space Agency’s Copernicus Sentinel-2 30-m resolution images at 0525 UTC 1 Apr 2019 centered over south-central Nepal, depicting the ∼35-km-long tornado damage path evident in (a) a true color image, and (b) a false color composite image utilizing the shortwave infrared channel that helps highlight the moisture content in vegetation. Arrows denote the approximate tornado path while the solid line indicates the Nepal–India border. Images contains Modified Copernicus Sentinel data 2021/Sentinel Hub.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
Deterministic (from ensemble member 1) and probabilistic fields from the ensemble initialized at 1800 UTC 29 Mar 2019: (a) 12-km outer grid 700-hPa 36-h forecast winds (kt) valid 0600 UTC 31 Mar, (b) 4-km nest 2-m dewpoint (°C) and 10-m wind (kt) 36-h forecast valid 0600 UTC 31 Mar, (c) regional zoom of 12-km grid 36-h forecast most unstable CAPE (MUCAPE; J kg−1) valid 0600 UTC 31 Mar, (d) regional zoom of 12-km grid 42-h forecast supercell composite parameter (SCP) valid 1200 UTC 31 Mar, (e) day-2 ensemble probability summary (%; 25–48-h forecasts valid 1800 UTC 30 March–1800 UTC 31 March) of 2–5-km updraft helicity exceeding 50 m2 s−2 within 20 km of a location, and (f) day-2 ensemble probability summary (%; 25–48-h forecasts) of 10-m wind speed exceeding 40 kt within 20 km of a location. Red circles in (e) and (f) highlight the area of interest in central Nepal. Reference cities/landmarks and sounding sites are denoted by stars and circles, respectively, in (e) and (f).
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
The day-2 composite probability products also indicated the potential for rotating thunderstorms and damaging straight-line winds in central Nepal (red circled areas in Figs. 11e,f). The probability of updraft helicity exceeding 50 m2 s−2 was relatively low across central Nepal compared to the high probabilities across northern Bangladesh and northeast India; however, a secondary maximum of probability > 50% extended northwestward into central Nepal, indicating a nontrivial threat for rotating updrafts in convective storms on 31 March (Fig. 11e). Also worth noting is the secondary axis of high probability for damaging winds across southeastern India (Fig. 11f) coinciding with near zero probability in updraft helicity (Fig. 11e), suggesting mainly a damaging straight-line wind threat there. However, no storm reports from India were identified for this day.
Similarly, the potential for damaging straight-line winds was present across central Nepal according to the day-2 composite probability of 10-m wind speeds exceeding 40 kt (Fig. 11f). Again, the highest probabilities for damaging convective winds occurred over India and Bangladesh, but a secondary maximum of probabilities greater than 50% is seen over central Nepal, in the vicinity of where the tornado occurred. In fact, the simulated timing and location of convective initiation for the responsible storms was nearly coincident in many members to the observed initiation as documented by Shrestha et al. (2019); however, the model runs did not sustain this deep convection long enough. Similar to the Kolkata windstorm event of 17 April 2018, this event saw HIWAT probabilities diminish somewhat from the day-2 to day-1 forecasts.
Cyclone Fani: Early May 2019.
The primary focus of this ensemble forecasting system is on premonsoon severe convection; however, the ensemble and probability products also demonstrated utility for tropical cyclone forecast scenarios, such as with Cyclone Fani of early May 2019. Cyclone Fani developed in the southern Bay of Bengal and intensified to an equivalent category 4 hurricane on the Saffir–Simpson scale [winds > 209 km h−1 (58 m s−1)] across the west-central Bay of Bengal. The cyclone made landfall across southeastern India causing extensive damage and impacts (see Indian Red Cross Society 2019; Wikipedia 2019), before weakening as it tracked inland into western Bangladesh (Fig. 12). The primary impacts on Bangladesh included storm surge along the coast to the east of the cyclone center, flooding, and high winds. Although not known to have occurred with Fani, tropical cyclones that recurve northeastward can sometimes produce mini tornado outbreaks (e.g., McCaul 1991, 1987) that cause further severe damage, albeit more localized than the broader cyclone impact, and several ensemble members depicted a corridor of updraft helicity swaths across southern and central Bangladesh associated with Fani’s landfall.
Six-hourly position and intensity track of Cyclone Fani during its lifespan from 26 Apr to 5 May 2019, reaching a maximum intensity equivalent to a category 4 storm on the Saffir–Simpson intensity scale prior to landfall in the India state of Odisha.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
In this event, we focus on two specific ensemble products: 1) the maximum 10-m wind speed from all ensemble members and 2) the probability of 10-m wind speed exceeding 40 kt. As illustrated in Fig. 13, these two products provide information about the potential variability and uncertainty in Cyclone Fani’s track and accompanying potential wind impact, as well as the reduction in high wind probabilities as the cyclone weakened during its forecast track across Bangladesh. The maximum 10-m wind speed at forecast hours 30, 36, and 42 from the ensemble cycle initialized at 1800 UTC 2 May 2019 (Figs. 13a,c,e) depict the forecast variability of Cyclone Fani by the ensemble members, as can be gleaned through multiple circular wind maxima. The contributing ensemble members that simulated the center of Fani closer to the coast retained the highest wind speeds (especially over water), whereas the other circular wind maximum further inland were more ill defined and weaker in magnitude. The overall trend of all ensemble members during these 12 h was for rapid weakening in maximum 10-m wind speeds from ∼60–70+ kt at forecast hour 30 (Fig. 13a) to ∼45–55 kt by forecast hour 42 (Fig. 13e). A similar trend occurred in the probability of 10-m wind speed exceeding 40 kt (Figs. 13b,d,f). At forecast hour 30 (Fig. 13b), the probability of exceeding 40 kt was greater than 80% across the southeast coastal area of India and adjacent waters. These probabilities decreased to ∼50%–60% across portions of southern Bangladesh by forecast hour 36 (Fig. 13d) and then less than 40% with greatly diminished spatial coverage by forecast hour 42 (Fig. 13f). These select ensemble products highlighted in Fig. 13 indicate the potential utility of applying some of the convective ensemble diagnostics to tropical cyclone forecast scenarios in this region.
Six-hourly sequence of (a),(c),(e) maximum 10-m wind speed from all ensemble members and (b),(d),(f) probability (%) of 10-m wind speed exceeding 40 kt within 20 km of a location from the WRF Model ensemble forecast initialized at 1800 UTC 2 May 2019. Forecast hours shown are 30, 36 and 42, valid at 0000, 0600, and 1200 UTC 4 May, respectively. Reference cities/landmarks and sounding sites are denoted by stars and circles, respectively.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
Precipitation and total lightning verification.
Select precipitation and lightning verification results are given in Figs. 14 and 15. The 2018 premonsoon 6-h accumulated precipitation Heidke skill score (HSS) at 25+-mm intensity indicates that the HIWAT ensemble members had the highest overall skill in forecasting heavy precipitation in the day-1 time frame over Bangladesh (blue shaded boxplots in Fig. 14a, bottom right), where the distribution of HSS scores approached 0.4 for many ensemble members and for the PMM. The HSS results were generally between 0.2 and 0.3 in the more mountainous countries of Nepal and Bhutan (top row of Fig. 14a), and around 0.3+ over northeastern India. Interestingly, Nepal and Bhutan experienced the greatest intermember variability in skill among the four countries, suggesting that the choice of model physical parameterization schemes may be more important in areas of complex terrain during the premonsoon. In fact, an examination of the individual scores comprising the boxplots reveal that the microphysics schemes largely contributed to the variability in verification during the 2018 premonsoon months in Nepal. In general for most countries, the day-2 precipitation skill is comparable to or slightly lower than the day-1 skill except for Bangladesh, which experienced the largest drop in HSS between the day-1 and day-2 forecasts. The PMM (orange dots) generally had competitive or higher skill relative to the individual ensemble members, as should be the case. The HSS results for the 2019 premonsoon season (Fig. 14b) have slightly lower scores overall, but support the same conclusions. When compared to GEFS member verification scores (red boxplots in Fig. 14), the HIWAT members had superior skill at this higher precipitation rate threshold (as expected), since higher model resolution is crucial in areas of complex terrain. When verifying lower-intensity thresholds (e.g., 1 and 5 mm per 6 h), the GEFS and HIWAT members had more comparable skill.
Boxplot distributions of Heidke skill score for 25+ mm (6 h)−1 accumulated precipitation within 20 km of a location for (a) 2018 and (b) 2019 premonsoon months of March–May (MAM), validated against GPM/IMERG-Final QPE and masked by country. Boxplot distributions comprise the 4-km nested grid of the 12 HIWAT members in blue shading, along with 12 GEFS members in red shading (GEFS member 1 plus the 11 used to initialize HIWAT members 2–12). Skill scores of the ensemble mean (EM) and probability matched mean (PMM) are also indicated by green and orange dots, respectively.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
The same analysis of HSS scores for the 2018 and 2019 premonsoon season total lightning forecasts are given in Fig. 15. As previously mentioned, we are only examining the occurrence of any lightning during the 24-h day-1 and day-2 intervals. The results indicate that the highest overall skill in predicting lightning occurred over Bangladesh (lower right) whereas the lowest skill was found in Bhutan (upper right), with very marked differences in these countries’ scores. The most likely explanation for this behavior is that the frequency of lightning is much higher over Bangladesh compared to Bhutan, which is supported by satellite-based lightning studies (e.g., Cecil et al. 2014). Similar to the precipitation HSS, we see the largest decline in lightning forecast skill over Bangladesh from day-1 to day-2 in both premonsoon seasons, likely due to the more highly variable nature of propagating convective systems dependent on synoptic and mesoscale forcing features that can be more uncertain as forecast time increases. The skill is more similar from day-1 to day-2 in Nepal and Bhutan because of the dominant topographic forcing that drives precipitation and lightning production in these areas (e.g., Barros et al. 2004; Houze et al. 2007; Medina et al. 2010; Romatschke and Houze 2011). The HSS scores for northeastern India (Fig. 15; lower-left panel) lie between that of Bangladesh and Nepal. It is interesting to note that the ensemble intermember skill variability for lightning is not nearly as large as the variability in forecasting heavy precipitation in the countries with the most complex terrain (Nepal and Bhutan), suggesting that the LFA algorithm is not as sensitive to (micro)physics parameterization choice as is the overarching precipitation production. So despite the inferred LFA coverage variability as a function of microphysics scheme in Fig. 4 for the 30 March 2018 event, the bulk verification scores suggest that the ensemble members have similar skill in diagnostic lightning activity during the premonsoon months. In general, the skill in forecasting any lightning activity during the premonsoon seasons was comparable to, or slightly less than the skill in predicting heavy rainfall within a 24-h period, except in Bhutan where the lightning skill was lowest (not shown).
Boxplot distributions of Heidke skill score for the lightning forecast algorithm (LFA) occurrence of any lightning in 24-h forecast intervals for the 2018 and 2019 premonsoon months of March–May (MAM), validated against gridded ENTLN total lightning observations and masked by country. Boxplot distributions comprise the 4-km nested grid of the 12 HIWAT members.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
Summary and transition activities
This paper presented an experimental–operational CAM ensemble forecasting system configured for the HKH region of south-central Asia. The focus and objectives have been to increase situational awareness and provide a computationally affordable prototype that can enable probability forecasts of specific hazardous convective phenomena during the synoptically active premonsoon season. The HKH ensemble was tailored to emulate similar CAM ensemble systems in the U.S. in order to provide guidance in forecasting threats of damaging straight-line winds, hail, supercells/tornadoes, lightning, and flash flooding. A 12-member solution was developed using the WRF NWP modeling system and a nested domain configuration with 12- and 4-km horizontal grid spacing. In addition to the ensemble suite of products, the first ensemble member also provided deterministic forecast guidance, including geographical subsets of severe weather indices that combine shear, instability, and other relevant environmental factors into normalized parameters that effectively convey the threat of supercells, significant tornadoes, damaging wind, and hail. Several high-impact case studies were presented in this paper to describe the potential range of applications of the ensemble system. While the exact placement and timing of convection is not typically expected to match observations in CAMs, the overall ensemble depiction of the nature and character of simulated severe thunderstorm elements was reasonably well captured in the cases highlighted herein. These select cases and other events not shown suggest that this relatively simple configuration of a CAM ensemble system can provide added value to the forecast process and enhanced situational awareness in the HKH region associated with premonsoon organized convection, and even tropical cyclone forecast impacts.
Composite verification results of precipitation and lightning from the 2018 and 2019 premonsoon seasons revealed that day-1 forecasts had the highest skill over Bangladesh and that the greatest intermember variability in skill was seen in precipitation forecasts over Nepal, suggesting sensitivity to and importance of physical parameterization selection in areas of complex terrain. Both precipitation and lightning forecasts proved most skillful during the most convectively active portions of the diurnal cycle. Despite bulk verification scores generally showing more skill on day 1, we noted that at times HIWAT depicted larger day-2 hazard probabilities for some high-impact events, which is worth investigating in greater detail for future analysis.
This manuscript focused on the premonsoon seasons, but HIWAT was extended into the wet monsoon months from 2018 to 2022 at the request of hydrometeorological organizations to help assess the flooding threat during the wet monsoon. The hourly PMM fields have been used each day as input into hydrological models for providing guidance on streamflow and inundation potential (Nelson et al. 2019). The 2018–19 demonstration of HIWAT provided forecasters at the Bangladesh Meteorological Department (BMD) and Nepal’s Department of Hydrology and Meteorology (DHM) with new information to enhance their early warning services in the region, in particular as they pertain to forecasting hazards associated with severe convective weather. In early 2022, a customized version of the ensemble forecasting system was installed onto a high-performance computing platform procured by BMD specifically for running HIWAT in-house and incorporating its guidance into their forecast process. The adoption of HIWAT by BMD represents a research-to-operations and capacity-building success for the SERVIR program in the HKH region, helping to meet the need of thunderstorm hazard forecasting capabilities for the protection of life and property.
Although we presented promising examples of several high-impact weather events, the system would benefit from additional testing and evaluation with the aid of a centralized storm reporting and archiving system to perform such studies. The thresholds applied in the probabilistic products may need further refinement so they are more appropriate to the region(s) of interest. For example, convection in the higher terrain of Nepal likely produces damaging hail at lower reflectivity and/or vertically integrated graupel thresholds than storms in Bangladesh, which tend to be deeper and have a higher melting level. However, these adjustments are not limited to the product configuration, but also extend to the model configurations. Further investigation to assess the impact of physical parameterizations and perform quantitative comparisons with satellite, radar and in situ observations will benefit forecaster understanding of individual modeling member accuracy and the ensemble output representativeness.
Acknowledgments.
This research was funded by NASA’s Applied Sciences Capacity Building Program Manager, Dr. Nancy Searby, and the SERVIR Project Scientist, Dr. Ashutosh Limaye, with support for ICIMOD from the United States Agency for International Development (USAID). We are grateful to Dr. Robert Rozumalski of NOAA for his help with updates and fixes to the UEMS for our application. We thank Dr. Lee Ellenburg and Timothy Mayer for their leadership in coordinating SERVIR-HKH collaborations. We thank Jeff Knickerbocker and the SERVIR Geospatial Informatics Team for assistance with configuring and hosting the virtual computer cluster for conducting simulations. We are also grateful to Dr. Christopher Hain, Paul Meyer, and Roger Allen of the NASA SPoRT Center for hosting real-time web graphics and providing in-kind computational resources for the first year of the project. We want to thank Dr. Jim Nelson and Sarva Pulla for development of the hydrological and web-mapping interfaces for HIWAT. Additionally, we sincerely thank Mr. Birendra Bajracharya, Dr. Mir Matin, and the North Eastern Space Applications Centre of India for hosting and facilitating trainings and presentations in Nepal, Bangladesh, and India. Mention of a copyrighted, trademarked, or proprietary product, service, or document does not constitute endorsement thereof by the authors, ENSCO Inc., the National Aeronautics and Space Administration, the SERVIR program, the United States Government, the International Centre for Integrated Mountain Development, or the Bangladesh Meteorological Department. Any such mention is solely for the purpose of fully informing the reader of the resources used to conduct the work reported herein.
Data availability statement.
The datasets used to generate the regional ensemble forecasts include personal tiles of the 0.25° resolution GFS model forecasts at 3-hourly intervals, and 0.50° resolution GEFS modeling members at 3–6-hourly intervals for initial and boundary conditions. The GFS and GEFS model archives are available from the National Centers for Environmental Information and are also archived on the project computing infrastructure. Output of UEMS/WRF Model simulations and derived ensemble products are archived on the cloud-like infrastructure from 2018 to present (March–August/September each year), with the intent of enabling open science by providing the data to the general public, university researchers, meteorological organizations, and decision-makers through open-access via a NASA DAAC and/or cloud platform. The GPM/IMERG-Final dataset for rainfall estimates and total lightning flashes are available from the NASA GPM project archive and Earth Networks, respectively. Both real-time and archived forecast imagery are available on the HIWAT Model Viewer page (https://hmv.servirglobal.net/), which includes deterministic output from ensemble member 1 and ensemble fields and probabilities for each daily forecast. An example of the HIWAT Model Viewer is shown in Fig. A1 of the appendix, displaying the total accumulated precipitation PMM from an excessive rainfall event in northeastern India during mid-May 2022.
Appendix: HIWAT Model Viewer Visualization Repository
All daily ensemble model imagery are archived to the HIWAT model visualization respository available at https://hmv.servirglobal.net. A screen capture of the user interface given in Fig. A1.
HIWAT Model Viewer showing the 54-h total accumulated precipitation PMM from the ensemble initialized at 1200 UTC 15 May 2022. The HIWAT Model Viewer is available online at https://hmv.servirglobal.net/.
Citation: Bulletin of the American Meteorological Society 104, 5; 10.1175/BAMS-D-21-0260.1
References
Barros, A. P., G. Kim, E. Williams, and S. W. Nesbitt, 2004: Probing orographic controls in the Himalayas during the monsoon using satellite imagery. Nat. Hazards Earth Syst. Sci., 4, 29–51, https://doi.org/10.5194/nhess-4-29-2004.
Bell, J., and A. Molthan, 2016: Evaluation of approaches to identifying hail damage to crop vegetation using satellite imagery. J. Oper. Meteor., 4, 142–159, https://doi.org/10.15191/nwajom.2016.0411.
Bell, J., E. Gebremichael, A. Molthan, L. Schultz, F. Meyer, and S. Shrestha, 2019: Synthetic aperture radar and optical remote sensing of crop damage attributed to severe weather in the Central United States. 2019 Int. Geoscience and Remote Sensing Symp., Yokohama, Japan, IEEE, 9938–9941, https://doi.org/10.1109/IGARSS.2019.8899775.
Bessho, K., and Coauthors, 2016: An introduction to Himawari-8/9 — Japan’s new-generation geostationary meteorological satellites. J. Meteor. Soc. Japan, 94, 151–183, https://doi.org/10.2151/jmsj.2016-009.
Bikos, D., J. Finch, and J. L. Case, 2016: The environment associated with significant tornadoes in Bangladesh. Atmos. Res., 167, 183–195, https://doi.org/10.1016/j.atmosres.2015.08.002.
Bista, D. B., 1989: Nepal in 1988: Many losses, some gains. Asian Surv., 29, 223–228, https://doi.org/10.2307/2644583.
Blumberg, W. G., K. T. Halbert, T. A. Supinie, P. T. Marsh, R. L. Thompson, and J. A. Hart, 2017: SHARPpy: An open-source sounding analysis toolkit for the atmospheric sciences. Bull. Amer. Meteor. Soc., 98, 1625–1636, https://doi.org/10.1175/BAMS-D-15-00309.1.
Brown, B., and Coauthors, 2020: The Model Evaluation Tools (MET): More than a decade of community-supported forecast verification. Bull. Amer. Meteor. Soc., 102, E782–E807, https://doi.org/10.1175/BAMS-D-19-0093.1.
Bui, V., L. C. Chang, and S. Heckman, 2016: A performance study of Earth Networks Total Lighting Network (ENTLN) and Worldwide Lightning Location Network (WWLLN). Proc. 2015 Int. Conf. on Computational Science and Computational Intelligence, Las Vegas, NV, IEEE, 386–391, https://doi.org/10.1109/CSCI.2015.120.
Cecil, D. J., D. E. Buechler, and R. J. Blakeslee, 2014: Gridded lightning climatology from TRMM-LIS and OTD: Dataset description. Atmos. Res., 135–136, 404–414, https://doi.org/10.1016/j.atmosres.2012.06.028.
Clark, A. J., 2017: Generation of ensemble mean precipitation forecasts from convection-allowing ensembles. Wea. Forecasting, 32, 1569–1583, https://doi.org/10.1175/WAF-D-16-0199.1.
Clark, A. J., and Coauthors, 2011: Probabilistic precipitation forecast skill as a function of ensemble size and spatial scale in a convection-allowing ensemble. Mon. Wea. Rev., 139, 1410–1418, https://doi.org/10.1175/2010MWR3624.1.
Clark, A. J., and Coauthors, 2012: An overview of the 2010 Hazardous Weather Testbed experimental forecast program spring experiment. Bull. Amer. Meteor. Soc., 93, 55–74, https://doi.org/10.1175/BAMS-D-11-00040.1.
Das, M. K., A. K. M. S. Islam, S. Karmakar, J. U. Khan, K. Mohammed, G. M. T. Islam, and S. K. Bala, 2020: Investigation of flash flood producing rainstorm in Northeast Bangladesh using WRF model. Water, Flood Management and Water Security under a Changing Climate, A. Haque and A. I. A. Chowdhury, Eds., Springer, 163–175, https://doi.org/10.1007/978-3-030-47786-8_11.
Das, S., and Coauthors, 2014: The SAARC STORM: A coordinated field experiment on severe thunderstorm observations and regional modeling over the South Asian region. Bull. Amer. Meteor. Soc., 95, 603–617, https://doi.org/10.1175/BAMS-D-12-00237.1.
Dewan, A., M. F. Hossain, M. M. Rahman, Y. Yamane, and R. L. Holle, 2017: Recent lightning-related fatalities and injuries in Bangladesh. Wea. Climate Soc., 9, 575–589, https://doi.org/10.1175/WCAS-D-16-0128.1.
Dewan, A., E. T. Ongee, M. M. Rahman, R. Mahmood, and Y. Yamane, 2018: Spatial and temporal analysis of a 17-year lightning climatology over Bangladesh with LIS data. Theor. Appl. Climatol., 134, 347–362, https://doi.org/10.1007/s00704-017-2278-3.
Ebert, E. E., 2001: Ability of a poor man’s ensemble to predict the probability and distribution of precipitation. Mon. Wea. Rev., 129, 2461–2480, https://doi.org/10.1175/1520-0493(2001)129<2461:AOAPMS>2.0.CO;2.
Evans, J. S., and C. A. Doswell, 2001: Examination of derecho environments using proximity soundings. Wea. Forecasting, 16, 329–342, https://doi.org/10.1175/1520-0434(2001)016<0329:EODEUP>2.0.CO;2.
Gatlin, P. N., J. L. Case, and J. Srikishen, 2021: The High-Impact Weather Assessment Toolkit. Earth Observation Science and Applications for Risk Reduction and Enhanced Resilience in Hindu Kush Himalaya Region, B. Bajracharya, R. B. Thapa, and M. A. Matin, Eds., Springer, 231–250.
Guan, H., B. Cui, and Y. Zhu, 2015: Improvement of statistical postprocessing using GEFS reforecast information. Wea. Forecasting, 30, 841–854, https://doi.org/10.1175/WAF-D-14-00126.1.
Guan, H., Y. Zhu, E. Sinsky, W. Li, X. Zhou, D. Hou, C. Melhauser, and R. Wobus, 2019: Systematic error analysis and calibration of 2-m temperature for the NCEP GEFS reforecast of the Subseasonal Experiment (SubX) project. Wea. Forecasting, 34, 361–376, https://doi.org/10.1175/WAF-D-18-0100.1.
Halbert, K. T., W. G. Blumberg, and P. T. Marsh, 2015: SHARPpy: Fueling the Python cult. Fifth Symp. on Advances in Modeling and Analysis Using Python, Phoenix, AZ, Amer. Meteor. Soc., 402, https://ams.confex.com/ams/95Annual/webprogram/Paper270233.html.
Han, J., and H.-L. Pan, 2011: Revision of convection and vertical diffusion schemes in the NCEP Global Forecast System. Wea. Forecasting, 26, 520–533, https://doi.org/10.1175/WAF-D-10-05038.1.
Heckman, S., C. Liu, and C. Sloop, 2014: Earth Networks lightning overview. 2014 Int. Conf. on Lightning Protection (ICLP), Shanghai, China, IEEE, 1866–1869, https://doi.org/10.1109/ICLP.2014.6973433.
Hirose, M., and K. Nakamura, 2005: Spatial and diurnal variation of precipitation systems over Asia observed by the TRMM precipitation radar. J. Geophys. Res., 110, D05106, https://doi.org/10.1029/2004JD004815.
Holle, R. L., D. Ashraf, S. Mohammad, R. Karim, and F. Hosain, 2018: Lightning fatalities and injuries in Bangladesh from 1990 through 2017. 25th Int. Lightning Detection Conf./Seventh Int. Lightning Meteorology Conf., Ft. Lauderdale, FL, Vaisala, 6 pp., www.vaisala.com/sites/default/files/documents/Lightning%20Fatalities%20and%20Injuries%20in%20Bangladesh%20from%201990%20through%202017_R.L.%20Holle%20et%20al.pdf.
Holle, R. L., A. Dewan, R. Said, W. A. Brooks, M. F. Hossain, and M. Rafiuddin, 2019: Fatalities related to lightning occurrence and agriculture in Bangladesh. Int. J. Disaster Risk Reduct., 41, 101264, https://doi.org/10.1016/j.ijdrr.2019.101264.
Hong, S.-Y., and J.-O. J. Lim, 2006: The WRF Single-moment 6-Class Microphysics scheme (WSM6). J. Korean Meteor. Soc., 42, 129–151.
Hong, S. Y., Y. Noh, and J. Dudhia, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 2318–2341, https://doi.org/10.1175/MWR3199.1.
Hou, A. Y., and Coauthors, 2014: The Global Precipitation Measurement mission. Bull. Amer. Meteor. Soc., 95, 701–722, https://doi.org/10.1175/BAMS-D-13-00164.1.
Houze, R. A., D. C. Wilton, and B. F. Smull, 2007: Monsoon convection in the Himalayan region as seen by the TRMM Precipitation Radar. Quart. J. Roy. Meteor. Soc., 133, 1389–1411, https://doi.org/10.1002/qj.106.
HT Correspondent, 2018: Twin storms kill 15 in West Bengal as winds top 98 kmph, disrupt life in Kolkata. Hindustan Times, 9 March, www.hindustantimes.com/india-news/ten-killed-as-storm-hits-kolkata-and-suburbs-rail-road-air-traffic-disrupted/story-DQpFAfgTzV5uz9QowYcAMJ.html.
Huffman, G. J., and Coauthors, 2020: Integrated multi-satellite retrievals for the Global Precipitation Measurement (GPM) Mission (IMERG). Satellite Precipitation Measurement, V. Levizzani et al., Eds., Advances in Global Change Research, Vol. 67, Springer, 343–353, https://doi.org/10.1007/978-3-030-24568-9_19.
Iacono, M. J., J. S. Delamere, E. J. Mlawer, M. W. Shephard, S. A. Clough, and W. D. Collins, 2008: Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, https://doi.org/10.1029/2008JD009944.
Indian Red Cross Society, 2019: Odisha FANI cyclone assessment report. Indian Red Cross Society, 14 pp., https://reliefweb.int/sites/reliefweb.int/files/resources/OdishaFaniAsessmentReport.pdf.
Janjić, Z. I., 1994: The step-mountain eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev., 122, 927–945, https://doi.org/10.1175/1520-0493(1994)122<0927:TSMECM>2.0.CO;2.
Janjić, Z. I., 2001: Nonsingular implementation of the Mellor-Yamada level 2.5 scheme in the NCEP Meso model. NCEP Office Note 437, 61 pp., https://repository.library.noaa.gov/view/noaa/11409.
Jirak, I. L., S. J. Weiss, and C. J. Melick, 2012: The SPC storm-scale ensemble of opportunity: Overview and results from the 2012 hazardous weather testbed spring forecasting experiment. 26th Conf. on Severe Local Storms, Nashville, TN, Amer. Meteor. Soc., P137, https://ams.confex.com/ams/26SLS/webprogram/Paper211729.html.
Jirak, I. L., A. J. Clark, C. J. Melick, and S. J. Weiss, 2016: Investigation of the impact of convection-allowing ensemble size for severe weather forecasting. 28th Conf. on Severe Local Storms, Portland, OR, Amer. Meteor. Soc., 15B.5, https://ams.confex.com/ams/28SLS/webprogram/Paper300907.html.
Johnson, A., and K. Sugden, 2014: Evaluation of sounding-derived thermodynamic and wind-related parameters associated with large hail events. E-J. Severe Storms Meteor., 9 (5), 1–42, https://ejssm.org/archives/2014/vol-9-5-2014/.
Kain, J. S., 2004: The Kain - Fritsch convective parameterization: An update. J. Appl. Meteor., 43, 170–181, https://doi.org/10.1175/1520-0450(2004)043<0170:TKCPAU>2.0.CO;2.
Kain, J. S., S. R. Dembek, S. J. Weiss, J. L. Case, J. J. Levit, and R. Sobash, 2010: Extracting unique information from high-resolution forecast models: Monitoring selected fields and phenomena every time step. Wea. Forecasting, 25, 1536–1542, https://doi.org/10.1175/2010WAF2222430.1.
Karmakar, S., D. A. Quadir, and M. K. Das, 2017: Numerical simulation of physical and dynamical characteristics associated with the severe thunderstorm on April 5, 2015 at Kushtia and Jhenaidah. Nat. Hazards, 86, 1127–1146, https://doi.org/10.1007/s11069-016-2733-y.
Knight, C. A., and N. C. Knight, 2001: Hailstorms. Severe Convective Storms, Meteor. Monogr., No. 50, Amer. Meteor. Soc., 223–254.
Lang, S., W. K. Tao, R. Cifelli, W. Olson, J. Halverson, S. Rutledge, and J. Simpson, 2007: Improving simulations of convective systems from TRMM LBA: Easterly and westerly regimes. J. Atmos. Sci., 64, 1141–1164, https://doi.org/10.1175/JAS3879.1.
Lang, S. E., W. K. Tao, X. Zeng, and Y. Li, 2011: Reducing the biases in simulated radar reflectivities from a bulk microphysics scheme: Tropical convective systems. J. Atmos. Sci., 68, 2306–2320, https://doi.org/10.1175/JAS-D-10-05000.1.
Lapierre, J., M. Hoekzema, M. Stock, C. Merrill, and S. C. Thangaraj, 2019: Earth networks lightning network and dangerous thunderstorm alerts. 11th Asia-Pacific Int. Conf. on Lightning, Hong Kong, China, IEEE, 1–5, https://doi.org/10.1109/APL.2019.8816032.
Li, W., and Coauthors, 2019: Evaluating the MJO prediction skill from different configurations of NCEP GEFS extended forecast. Climate Dyn., 52, 4923–4936, https://doi.org/10.1007/s00382-018-4423-9.
Lin, Y. L., R. D. Farley, and H. D. Orville, 1983: Bulk parameterization of the snow field in a cloud model. J. Climate Appl. Meteor., 22, 1065–1092, https://doi.org/10.1175/1520-0450(1983)022<1065:BPOTSF>2.0.CO;2.
Litta, A. J., U. C. Mohanty, S. Das, and S. Mary Idicula, 2012: Numerical simulation of severe local storms over east India using WRF-NMM mesoscale model. Atmos. Res., 116, 161–184, https://doi.org/10.1016/j.atmosres.2012.04.015.
Mannan, M. A., M. N. Ahasan, and M. S. Alam, 2015: Study of severe thunderstorms over Bangladesh and its surrounding areas during pre-monsoon season of 2013 using WRF-ARW model. High-Impact Weather Events over the SAARC Region, K. Ray et al., Eds., Springer, 3–22.
McCaul, E. W., 1987: Observations of the Hurricane “Danny” tornado outbreak of 16 August 1985. Mon. Wea. Rev., 115, 1206–1223, https://doi.org/10.1175/1520-0493(1987)115<1206:OOTHTO>2.0.CO;2.
McCaul, E. W., 1991: Buoyancy and shear characteristics of hurricane-tornado environments. Mon. Wea. Rev., 119, 1954–1978, https://doi.org/10.1175/1520-0493(1991)119<1954:BASCOH>2.0.CO;2.
McCaul, E. W., S. J. Goodman, K. M. LaCasse, and D. J. Cecil, 2009: Forecasting lightning threat using cloud-resolving model simulations. Wea. Forecasting, 24, 709–729, https://doi.org/10.1175/2008WAF2222152.1.
McCaul, E. W., G. Priftis, J. L. Case, T. Chronis, P. N. Gatlin, S. J. Goodman, and F. Kong, 2020: Sensitivities of the WRF lightning forecasting algorithm to parameterized microphysics and boundary layer schemes. Wea. Forecasting, 35, 1545–1560, https://doi.org/10.1175/WAF-D-19-0101.1.
Medina, S., R. A. Houze, A. Kumar, and D. Niyogi, 2010: Summer monsoon convection in the Himalayan region: Terrain and land cover effects. Quart. J. Roy. Meteor. Soc., 136, 593–616, https://doi.org/10.1002/qj.601.
Meyer, J. D. D., B. Pokharel, and R. R. Gillies, 2021: Simulating the storm environment responsible for Nepal’s first observed tornado. Wea. Climate Extremes, 34, 100368, https://doi.org/10.1016/j.wace.2021.100368.
Miller, S. D., T. L. Schmit, C. J. Seaman, D. T. Lindsey, M. M. Gunshor, R. A. Kohrs, Y. Sumida, and D. Hillger, 2016: A sight for sore eyes: The return of true color to geostationary satellites. Bull. Amer. Meteor. Soc., 97, 1803–1816, https://doi.org/10.1175/BAMS-D-15-00154.1.
Molthan, A., J. Bell, T. Cole, and J. Burks, 2014: Satellite-based identification of tornado damage tracks from the 27 April 2011 severe weather outbreak. J. Oper. Meteor., 2, 191–208, https://doi.org/10.15191/nwajom.2014.0216.
Morrison, H., G. Thompson, and V. Tatarskii, 2009: Impact of cloud microphysics on the development of trailing stratiform precipitation in a simulated squall line: Comparison of one- and two-moment schemes. Mon. Wea. Rev., 137, 991–1007, https://doi.org/10.1175/2008MWR2556.1.
Murata, H., K. Saitoh, and Y. Sumida, 2018: True color imagery rendering for Himawari-8 with a color reproduction approach based on the CIE XYZ color system. J. Meteor. Soc. Japan, 96B, 211–238, https://doi.org/10.2151/jmsj.2018-049.
Nakanishi, M., and H. Niino, 2009: Development of an improved turbulence closure model for the atmospheric boundary layer. J. Meteor. Soc. Japan, 87, 895–912, https://doi.org/10.2151/jmsj.87.895.
Nelson, E. J., and Coauthors, 2019: Enabling stakeholder decision-making with Earth observation and modeling data using Tethys platform. Front. Environ. Sci., 7, 148, https://doi.org/10.3389/fenvs.2019.00148.
NIRAPAD, 2019: Hazard incidents in Bangladesh. Reliefweb, 31 March, https://reliefweb.int/report/bangladesh/hazard-incidents-bangladesh-march-2019.
Powers, J. G., and Coauthors, 2017: The Weather Research and Forecasting Model: Overview, system efforts, and future directions. Bull. Amer. Meteor. Soc., 98, 1717–1737, https://doi.org/10.1175/BAMS-D-15-00308.1.
Roberts, B., I. L. Jirak, A. J. Clark, S. J. Weiss, and J. S. Kain, 2019: Postprocessing and visualization techniques for convection-allowing ensembles. Bull. Amer. Meteor. Soc., 100, 1245–1258. https://doi.org/10.1175/BAMS-D-18-0041.1
Roberts, B., B. T. Gallo, I. L. Jirak, A. J. Clark, D. C. Dowell, X. Wang, and Y. Wang, 2020: What does a convection-allowing ensemble of opportunity buy us in forecasting thunderstorms? Wea. Forecasting, 35, 2293–2316, https://doi.org/10.1175/WAF-D-20-0069.1.
Romatschke, U., and R. A. Houze, 2011: Characteristics of precipitating convective systems in the premonsoon season of South Asia. J. Hydrometeor., 12, 157–180, https://doi.org/10.1175/2010JHM1311.1.
Romatschke, U., S. Medina, and R. A. Houze, 2010: Regional, seasonal, and diurnal variations of extreme convection in the South Asian region. J. Climate, 23, 419–439, https://doi.org/10.1175/2009JCLI3140.1.
Schwartz, C. S., and R. A. Sobash, 2017: Generating probabilistic forecasts from convection-allowing ensembles using neighborhood approaches: A review and recommendations. Mon. Wea. Rev., 145, 3397–3418, https://doi.org/10.1175/MWR-D-16-0400.1.
Schwartz, C. S., G. S. Romine, K. R. Smith, and M. L. Weisman, 2014: Characterizing and optimizing precipitation forecasts from a convection-permitting ensemble initialized by a mesoscale ensemble Kalman filter. Wea. Forecasting, 29, 1295–1318, https://doi.org/10.1175/WAF-D-13-00145.1.
Schwartz, C. S., G. S. Romine, R. A. Sobash, K. R. Fossell, and M. L. Weisman, 2015: NCAR’s experimental real-time convection-allowing ensemble prediction system. Wea. Forecasting, 30, 1645–1654, https://doi.org/10.1175/WAF-D-15-0103.1.
Schwartz, C. S., G. S. Romine, R. A. Sobash, K. R. Fossell, and M. L. Weisman, 2019: NCAR’s real-time convection-allowing ensemble project. Bull. Amer. Meteor. Soc., 100, 321–343, https://doi.org/10.1175/BAMS-D-17-0297.1.
Shrestha, A., D. Pradhananga, and J. Karmacharya, 2019: Report on Bara-Parsa tornado. Small Earth Nepal, 82 pp., www.smallearth.org.np/wp-content/uploads/2019/04/Report-on-Bara-Parsa-Tornado.pdf.
Skamarock, W. C., and Coauthors, 2008: A description of the advanced research WRF version 3. NCAR Tech. Note NCAR/TN-468+STR, 113 pp., https://doi.org/10.5065/D68S4MVH.
Skofronick-Jackson, G., and Coauthors, 2017: The Global Precipitation Measurement (GPM) mission for science and society. Bull. Amer. Meteor. Soc., 98, 1679–1695, https://doi.org/10.1175/BAMS-D-15-00306.1.
Tan, J., G. J. Huffman, D. T. Bolvin, and E. J. Nelkin, 2019: IMERG V06: Changes to the morphing algorithm. J. Atmos. Oceanic Technol., 36, 2471–2482, https://doi.org/10.1175/JTECH-D-19-0114.1.
Tao, W. K., and Coauthors, 2003: Microphysics, radiation and surface processes in the Goddard Cumulus Ensemble (GCE) model. Meteor. Atmos. Phys., 82, 97–137, https://doi.org/10.1007/s00703-001-0594-7.
Tao, W. K., and Coauthors, 2014: The Goddard Cumulus Ensemble model (GCE): Improvements and applications for studying precipitation processes. Atmos. Res., 143, 392–424, https://doi.org/10.1016/j.atmosres.2014.03.005.
Thompson, G., P. R. Field, R. M. Rasmussen, and W. D. Hall, 2008: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part II: Implementation of a new snow parameterization. Mon. Wea. Rev., 136, 5095–5115, https://doi.org/10.1175/2008MWR2387.1.
Thompson, R. L., R. Edwards, J. A. Hart, K. L. Elmore, and P. Markowski, 2003: Close proximity soundings within supercell environments obtained from the Rapid Update Cycle. Wea. Forecasting, 18, 1243–1261, https://doi.org/10.1175/1520-0434(2003)018<1243:CPSWSE>2.0.CO;2.
Thompson, R. L., E. Edwards, and C. M. Mead, 2004: An update to the supercell composite and significant tornado parameters. 22nd Conf. on Severe Local Storms, Hyannis, MA, Amer. Meteor. Soc., P8.1, https://ams.confex.com/ams/11aram22sls/techprogram/paper_82100.htm.
Wikipedia, 2019: Cyclone Fani impact and aftermath. Wikipedia, accessed 22 June 2022, https://en.wikipedia.org/wiki/Cyclone_Fani#Impact_and_aftermath.
Yamane, Y., and T. Hayashi, 2006: Evaluation of environmental conditions for the formation of severe local storms across the Indian subcontinent. Geophys. Res. Lett., 33, L17806, https://doi.org/10.1029/2006GL026823.
Yamane, Y., T. Hayashi, A. M. Dewan, and F. Akter, 2010a: Severe local convective storms in Bangladesh: Part I. Climatology. Atmos. Res., 95, 400–406, https://doi.org/10.1016/j.atmosres.2009.11.004.
Yamane, Y., T. Hayashi, A. M. Dewan, and F. Akter, 2010b: Severe local convective storms in Bangladesh: Part II. Environmental conditions. Atmos. Res., 95, 407–418, https://doi.org/10.1016/j.atmosres.2009.11.003.
Yamane, Y., T. Hayashi, M. Kiguchi, F. Akter, and A. M. Dewan, 2013: Synoptic situations of severe local convective storms during the pre-monsoon season in Bangladesh. Int. J. Climatol., 33, 725–734, https://doi.org/10.1002/joc.3460.
Zheng, Y., K. Alapaty, J. A. Herwehe, A. D. Del Genio, and D. Niyogi, 2016: Improving high-resolution weather forecasts using the Weather Research and Forecasting (WRF) Model with an updated Kain–Fritsch scheme. Mon. Wea. Rev., 144, 833–860, https://doi.org/10.1175/MWR-D-15-0005.1.
Zhou, X., Y. Zhu, D. Hou, Y. Luo, J. Peng, and R. Wobus, 2017: Performance of the new NCEP global ensemble forecast system in a parallel experiment. Wea. Forecasting, 32, 1989–2004, https://doi.org/10.1175/WAF-D-17-0023.1.