Diagnosing Factors Leading to an Incorrect Supercell Thunderstorm Forecast

Paul D. Mykolajtchuk aDepartment of Meteorology and Atmospheric Science, The Pennsylvania State University, University Park, Pennsylvania

Search for other papers by Paul D. Mykolajtchuk in
Current site
Google Scholar
PubMed
Close
,
Keenan C. Eure aDepartment of Meteorology and Atmospheric Science, The Pennsylvania State University, University Park, Pennsylvania

Search for other papers by Keenan C. Eure in
Current site
Google Scholar
PubMed
Close
,
David J. Stensrud aDepartment of Meteorology and Atmospheric Science, The Pennsylvania State University, University Park, Pennsylvania

Search for other papers by David J. Stensrud in
Current site
Google Scholar
PubMed
Close
,
Yunji Zhang aDepartment of Meteorology and Atmospheric Science, The Pennsylvania State University, University Park, Pennsylvania

Search for other papers by Yunji Zhang in
Current site
Google Scholar
PubMed
Close
,
Steven J. Greybush aDepartment of Meteorology and Atmospheric Science, The Pennsylvania State University, University Park, Pennsylvania

Search for other papers by Steven J. Greybush in
Current site
Google Scholar
PubMed
Close
, and
Matthew R. Kumjian aDepartment of Meteorology and Atmospheric Science, The Pennsylvania State University, University Park, Pennsylvania

Search for other papers by Matthew R. Kumjian in
Current site
Google Scholar
PubMed
Close
Free access

Abstract

On 28 April 2019, hourly forecasts from the operational High-Resolution Rapid Refresh (HRRR) model consistently predicted an isolated supercell storm late in the day near Dodge City, Kansas, that subsequently was not observed. Two convection-allowing model (CAM) ensemble runs are created to explore the reasons for this forecast error and implications for severe weather forecasting. The 40-member CAM ensembles are run using the HRRR configuration of the WRF-ARW Model at 3-km horizontal grid spacing. The Gridpoint Statistical Interpolation (GSI)-based ensemble Kalman filter is used to assimilate observations every 15 min from 1500 to 1900 UTC, with resulting ensemble forecasts run out to 0000 UTC. One ensemble only assimilates conventional observations, and its forecasts strongly resemble the operational HRRR with all ensemble members predicting a supercell storm near Dodge City. In the second ensemble, conventional observations plus observations of WSR-88D radar clear-air radial velocities, WSR-88D diagnosed convective boundary layer height, and GOES-16 all-sky infrared brightness temperatures are assimilated to improve forecasts of the preconvective environment, and its forecasts have half of the members predicting supercells. Results further show that the magnitude of the low-level meridional water vapor flux in the moist tongue largely separates members with and without supercells, with water vapor flux differences of 12% leading to these different outcomes. Additional experiments that assimilate only radar or satellite observations show that both are important to predictions of the meridional water vapor flux. This analysis suggests that mesoscale environmental uncertainty remains a challenge that is difficult to overcome.

Significance Statement

Forecasts from operational numerical models are the foundation of weather forecasting. There are times when these models make forecasts that do not come true, such as 28 April 2019 when successive forecasts from the operational High-Resolution Rapid Refresh (HRRR) model predicted a supercell storm late in the day near Dodge City, Kansas, that subsequently was not observed. Reasons for this forecast error are explored using numerical experiments. Results suggest that relatively small changes to the prestorm environment led to large differences in the evolution of storms on this day. This result emphasizes the challenges to operational severe weather forecasting and the continued need for improved use of all available observations to better define the atmospheric state given to forecast models.

© 2023 American Meteorological Society. This published article is licensed under the terms of the default AMS reuse license. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: David J. Stensrud, david.stensrud@psu.edu

Abstract

On 28 April 2019, hourly forecasts from the operational High-Resolution Rapid Refresh (HRRR) model consistently predicted an isolated supercell storm late in the day near Dodge City, Kansas, that subsequently was not observed. Two convection-allowing model (CAM) ensemble runs are created to explore the reasons for this forecast error and implications for severe weather forecasting. The 40-member CAM ensembles are run using the HRRR configuration of the WRF-ARW Model at 3-km horizontal grid spacing. The Gridpoint Statistical Interpolation (GSI)-based ensemble Kalman filter is used to assimilate observations every 15 min from 1500 to 1900 UTC, with resulting ensemble forecasts run out to 0000 UTC. One ensemble only assimilates conventional observations, and its forecasts strongly resemble the operational HRRR with all ensemble members predicting a supercell storm near Dodge City. In the second ensemble, conventional observations plus observations of WSR-88D radar clear-air radial velocities, WSR-88D diagnosed convective boundary layer height, and GOES-16 all-sky infrared brightness temperatures are assimilated to improve forecasts of the preconvective environment, and its forecasts have half of the members predicting supercells. Results further show that the magnitude of the low-level meridional water vapor flux in the moist tongue largely separates members with and without supercells, with water vapor flux differences of 12% leading to these different outcomes. Additional experiments that assimilate only radar or satellite observations show that both are important to predictions of the meridional water vapor flux. This analysis suggests that mesoscale environmental uncertainty remains a challenge that is difficult to overcome.

Significance Statement

Forecasts from operational numerical models are the foundation of weather forecasting. There are times when these models make forecasts that do not come true, such as 28 April 2019 when successive forecasts from the operational High-Resolution Rapid Refresh (HRRR) model predicted a supercell storm late in the day near Dodge City, Kansas, that subsequently was not observed. Reasons for this forecast error are explored using numerical experiments. Results suggest that relatively small changes to the prestorm environment led to large differences in the evolution of storms on this day. This result emphasizes the challenges to operational severe weather forecasting and the continued need for improved use of all available observations to better define the atmospheric state given to forecast models.

© 2023 American Meteorological Society. This published article is licensed under the terms of the default AMS reuse license. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: David J. Stensrud, david.stensrud@psu.edu

1. Introduction

Supercell thunderstorms produce the majority of severe weather reports in the United States (Duda and Gallus 2010), making the prediction of supercells an important component of the National Weather Service’s watch–warning program (Johns and Doswell 1992). Discriminating between supercell and nonsupercell thunderstorms is assisted by the skillful use of sounding-derived parameters, such as convective available potential energy (CAPE), vertical wind shear, storm-relative helicity (SRH), and various combinations thereof (e.g., Moller et al. 1994; Rasmussen and Blanchard 1998; Thompson et al. 2007). These parameters can be calculated from rawinsonde observations or numerical weather prediction model output and are used to define regions where supercells are likely to develop given convection initiation (CI). In recent years, convection-allowing models (CAMs) have provided valuable information on convective mode, which has further assisted in supercell forecasting (Done et al. 2004; Weisman et al. 2008; Kain et al. 2008). CAMs often predict CI time reasonably well, while being less accurate in location (Kain et al. 2013; Johnson et al. 2016; Stelten and Gallus 2017).

Admittedly, CI remains challenging in model forecasts because it requires an adequate representation of processes that assist parcels in reaching their level of free convection (LFC), including synoptic-scale factors involving air masses associated with baroclinic storm systems, and mesoscale factors involving regions of localized low-level convergence ranging from sea breezes, drylines, orography, and fronts (e.g., Brooks et al. 1994). For accurate prediction of convective mode, accurately representing the overall airmass instability also is important, as synoptic-scale systems often provide the broad conditions required for severe weather outbreaks (Johns and Doswell 1992; Tochimoto and Niino 2016). Any misplacement or misrepresentation of these environmental variables in model guidance has the potential to lead to missed or inaccurate convective forecasts. Therefore, it is key to correctly depict these environmental variables (Zhang et al. 2015). This is particularly important within severe weather environments owing to the destructive nature of severe weather outbreaks in which errors in the predictions of convection can have large detrimental effects on preparedness and response (NCEI 2022).

One of the CAMs currently run operationally by NOAA is the High-Resolution Rapid Refresh (HRRR; Benjamin et al. 2016; Dowell et al. 2022), which uses 3-km horizontal grid spacing and is run every hour in a rapid cycling mode. HRRR forecasts extend out to at least 18 h, with some runs providing forecasts out to 48 h. The HRRR has become an important component of operational severe weather prediction and has demonstrated ability to predict supercells, with results showing a HRRR ensemble correctly predicts convective mode for 65% of the events (Grim et al. 2022).

On 28 April 2019, the operational HRRR forecasts consistently developed a supercell thunderstorm in southwestern Kansas by late in the day (Fig. 1). It is likely that these forecasts influenced the operational severe weather guidance of the day, as the NOAA Storm Prediction Center (SPC) convective outlook shifted southward to include southwestern Kansas with the 1630 UTC update. However, no supercell thunderstorm developed in southwestern Kansas. The consistent HRRR forecasts of this unobserved supercell merit attention to determine potential reasons for these inaccuracies.

Fig. 1.
Fig. 1.

Composite reflectivity forecasts from the HRRR valid at 0000 UTC 29 Apr 2019 from runs started at (a) 1200 UTC, (b) 1500 UTC, and (c) 1900 UTC 28 Apr 2019. The unobserved supercell is indicated by the black circle in each panel.

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

The goal of this study is to diagnose what factors could have led to an hourly updated CAM consistently producing a supercell where none was observed. Three hypotheses for the inaccurate supercell forecast are that the HRRR predicted 1) CI that did not occur and these storms developed into the unobserved supercell, 2) CI accurately but had environmental conditions overly favorable to supercells leading to more intense predicted storms than seen in observations, and 3) some combination thereof. To explore these hypotheses, we produce and compare two CAM ensembles. One ensemble is designed to resemble the operational HRRR forecasts; this ensemble assimilates only conventional observations from this day. The second ensemble assimilates the same conventional observations plus three additional remotely sensed observations chosen to help better define the preconvective environment and the physical processes leading to CI. As seen below, many forecasts from this ensemble correctly predict only weak convection near Dodge City. These two ensemble forecasts are compared to determine what factors may have led to the inaccurate supercell forecasts on 28 April 2019. Section 2 provides an overview of the event, followed by a description of the model, assimilation system, observational data, and experiment design in section 3. Results follow in section 4, with a discussion in section 5.

2. The 28 April 2019 case

The first SPC convective outlook of the day at 0545 UTC 28 April 2019 had a slight risk in north-central Kansas, with southwestern Kansas having relatively low (2%) tornado probabilities. By the 1630 UTC outlook, the slight risk area had expanded to the south and covered most of Kansas. Tornado probability also had increased to 5% near Dodge City, Kansas; this region is where the HRRR forecasts consistently produced a supercell storm. A severe thunderstorm watch was issued at 2150 UTC and covered most of western Kansas. Although severe weather was reported in northern Kansas, the reports were mostly for wind and hail. There were no severe reports within the southern third of this watch, near Dodge City, Kansas.

Multiple factors contributed to the development of convective storms on 28 April 2019. First, a warm front was gradually progressing northeastward and passed Dodge City, Kansas by 2100 UTC, according to the Weather Prediction Center (WPC) surface analysis. Second, a midlevel shortwave trough was moving across Colorado, providing midlevel ascent across much of eastern Colorado. In response, a broad region of low pressure developed in southeastern Colorado by late in the day. The northeastward movement of the warm front provided the airmass change necessary for convective development, with southerly winds advecting higher dewpoint temperatures after passage of the warm front. There also was a sharp zonal dewpoint gradient associated with a dryline in eastern Colorado that appeared to initiate some of the convection. With westerly flow aloft, storms progressed eastward into an air mass conducive for severe weather, with operational analyses depicting a strong low-level jet, SRH values > 200 m2 s−2, and mixed-layer CAPE values approaching 2000 J kg−1 in a moist tongue that extends northward along the Colorado–Kansas border.

Widespread convection first initiated in northeastern Colorado just prior to 1900 UTC, with several isolated cells developing in southeastern Colorado 30 min later (Fig. 2a, groups A and B). The large areal extent of convective cell development suggests that initiation was closely tied to ascent associated with the passage of the midlevel shortwave across eastern Colorado. Most of these cells persisted and moved eastward toward Kansas over the next few hours. The cells that entered northwestern Kansas intensified, with radar reflectivity factor (hereafter reflectivity) values > 60 dBZ observed by 2130 UTC, whereas the groups of cells that moved into southwestern Kansas remained smaller in spatial coverage and had modest reflectivity (20–30 dBZ) with isolated pockets of ∼40 dBZ (Fig. 2b, groups A and B). Over the next 2 h, additional convective cells developed over eastern Colorado, including a small group of cells that initiated at 2200 UTC in southeastern Colorado (Fig. 2b, group C). The storms in northwestern Kansas developed into an east–west-aligned convective line with a growing area of stratiform rain to the east by 0000 UTC 29 April, while the cells in southwestern Kansas remained weak (Fig. 2c). Indeed, the A group cells dissipated by 2300 UTC as they reached Dodge City, Kansas, leaving only the B and C cell groups in southwestern Kansas at 0000 UTC 29 April 2019. Thus, weak convection persisted along a roughly west–east line in southwestern Kansas, with a parallel line of intense storms to the north that created an expansive region of stratiform rain to the east of this convective line (Fig. 2c). By 0300 UTC 29 April 2019, the weak convection in southwestern Kansas had dissipated completely, while to the north a bowing convective line was located along the warm frontal boundary and moving southeastward (not shown). This convective line crossed the entire state of Kansas over the next 7 h. In contrast, the HRRR forecasts started prior to CI continued to maintain a significant supercell near Dodge City at 0300 UTC, which may have negatively influenced the prediction of the bowing convective line to the north. It is unclear from the available operational data why the convection across southwestern Kansas remained weak, while the convection that moved across northern Kansas was much more intense.

Fig. 2.
Fig. 2.

Observed Multi-Radar/Multi-Sensor (MRMS) base reflectivity (dBZ) at (a) 2000 UTC 28 Apr 2019, near the time of CI in southeastern Colorado; (b) 2200 UTC 28 Apr 2019; and (c) 0000 UTC 29 Apr 2019, from the Iowa State University Mesonet archive. Letters A, B, and C indicate distinct convective cell groups that developed in Colorado and moved eastward into Kansas and are tracked via visual inspection of the 15-min radar observations. Cells in groups A and B initiated at 1930 UTC, while cells in group C initiated at 2200 UTC.

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

3. Methods

a. Numerical model

The High-Resolution Rapid Refresh (HRRR) configuration of the Advanced Research and Forecasting WRF (WRF-ARW) Model version 3.8.1 (Skamarock et al. 2008; Benjamin et al. 2016; Dowell et al. 2022) is used in this study. The model grid is a 600 km × 600 km region, with 51 vertical levels (domain shown in Fig. 1). The horizontal grid spacing is 3 km, and the vertical levels extend to 50 hPa at model top. This model domain is selected to include the incorrect supercell forecast across eastern Colorado and western Kansas, while also including regions of the Texas Panhandle where the impact of the dryline could be important for the convective forecast. The domain is large enough to capture the passage of the warm front that took place during the afternoon hours from southwest to northeast.

The physical parameterization schemes chosen are the same as used in the HRRR, and no convective parameterization is used. For the boundary layer, the Mellor–Yamada–Nakanishi–Niino (MYNN) 2.5 level turbulent kinetic energy (TKE) parameterization is selected (Nakanishi and Niino 2009; Olson et al. 2019), with the Rapid Update Cycle (RUC) land surface model from Benjamin et al. (2004). The Thompson et al. (2008) microphysics scheme is used to predict hydrometeor mixing ratios and the Rapid Radiative Transfer Model is used for both shortwave and longwave radiation (Iacono et al. 2008).

Initial conditions for a 40-member ensemble are created using this HRRR configuration of the WRF-ARW Model. The ensemble uses 20 members from the 0000 UTC Global Ensemble Forecast System (GEFS) runs, and another 20 members from the 0600 UTC GEFS runs on 28 April, in which all ensemble member fields are then centered around the 1200 UTC HRRR analysis as in Eure et al. (2023). Both ensembles use these initial and boundary conditions, with forecasts starting at 1200 UTC 28 April and ending 12 h later at 0000 UTC 29 April.

b. Data assimilation approach and observations for assimilation

The community Gridpoint Statistical Interpolation (GSI; version 3.7) based Ensemble Kalman Filter (EnKF, version 1.3) (Wu et al. 2002; Kleist et al. 2009; Wang and Lei 2014; Wang et al. 2013; Johnson et al. 2015; Liu et al. 2018) with extensions to ingest satellite radiances (Jones et al. 2020; Johnson et al. 2022) is used to assimilate observations in this study. This version of the EnKF system uses a square root EnKF filter (EnSRF; Whitaker and Hamill 2002). Relaxation to prior perturbation is used to generate sufficient variance among the ensemble members (Zhang et al. 2004), where 80% of the perturbation inflation occurred with the prior ensemble and 20% of the additional perturbations occurred with the posterior ensemble. Although Whitaker and Hamill (2012) propose a relaxation to prior spread inflation technique, this study follows the technique outlined in Zhang et al. (2004). Consistency ratios from the assimilated observations are between 0.4 and 1.2, indicating that the ensemble spread is sufficient (Dowell et al. 2004).

Three different observation groups are assimilated in the two ensemble experiments that follow. The first group consists of conventional observations, which are routinely collected and assimilated by the National Weather Service. The second group consists of WSR-88D radar observations of clear-air radial winds and convective boundary layer (CBL) height, which is diagnosed from polarimetric radar observations (Banghoff et al. 2018). These observations provide information on the strength and orientation of the low-level flow near the radars and help constrain CBL depth during the data assimilation period. Assimilation of these radar observations should influence predictions of water vapor flux into the moist tongue that extends northward into western Kansas. The third group consists of GOES-16 observations of all-sky infrared brightness temperature (BT). Assimilation of these satellite observations should influence environmental thermodynamic structures in the middle troposphere while also helping the locations and tops of existing clouds and influence CI. Combined, it is expected that these radar and satellite observations should improve the forecasts of CAPE, LFC, and wind shear (primarily via changes in the low-level winds) across the model domain in comparison to experiments that only assimilate conventional observations, thereby providing insight into why the unobserved supercell developed in the operational model runs. A more extensive study on the assimilation of these observations is provided by Eure et al. (2023) and shows positive impacts on CI forecasts from both the radar and the satellite observations.

1) Conventional observations

The conventional observations used in the experiments are archived at the National Center for Atmospheric Research (NCAR) and are assimilated in both experiments. This dataset contains observations from rawinsondes, meteorological aerodrome reports (METARs), synoptic observations (SYNOPs), airplane reports, and radar-derived wind data, which, together, provide surface pressure, specific humidity, dry-bulb temperature, and the zonal and meridional wind components. Due to the different time scales at which the observations are available, the number of conventional observations that are assimilated with each cycle varies. With the assimilation window from 1500 to 1900 UTC the observations assimilated are primarily surface weather station observations. A horizontal covariance localization length of 300 km is used, with a vertical covariance localization scale factor of 0.40 scale height, which translates to a surface observation no longer having a discernable effect above 670 hPa. The observation error comes directly from GSI and varies based on the kind of conventional observation that is being assimilated. More information on the error descriptions for each conventional observation type is found at the Environmental Modeling Center website (https://www.emc.ncep.noaa.gov/mmb/data_processing/prepbufr.doc/.htm).

2) WSR-88D radar observations

Six WSR-88D radar sites within the model domain are used to obtain clear-air radial velocity observations: Amarillo, Texas (KAMA); Dodge City, Kansas (KDDC); Goodland, Kansas (KGLD); Pueblo, Colorado (KPUX); Hastings, Nebraska (KUEX); and Vance Air Force Base, Oklahoma (KVNX). Clear-air radial velocity observations are from the Level-II WSR-88D radar data. Prior to precipitation formation, WSR-88D radars generally are operated in clear-air mode and provide observations within the CBL roughly every 10 min for five elevation angles from 0.5° to 4.5°. Owing to the focus on improvements to the preconvective environment, radar reflectivity observations are not assimilated. To reduce the quantity of the observations to be assimilated, superobservations (SOs) are created in which only data from the lowest and highest elevation angle are used. The observations are quality controlled, including dealiasing the radial velocities values following Eilts and Smith (1990), and then the SOs are created following the procedures in Zhang et al. (2009). The resulting clear-air SOs are available every 15 min from 1500 UTC 28 April to 0000 UTC 29 April 2019.

The clear-air radial velocity SOs have a horizontal covariance localization length of 30 km with a vertical covariance localization scale factor of 0.36 scale height, which translates to a pressure level of roughly 700 hPa as the cutoff threshold for when the observation no longer has an effect vertically for an observation at 1000 hPa. The observation error used is 3 m s−1. The observation operator for radial winds is taken from Johnson et al. (2015).

Level-II data from the same WSR-88D sites are used to estimate CBL heights from differential reflectivity (ZDR) measurements. Small-scale turbulent mixing of drier free-tropospheric air with moister boundary layer air occurs at the top of the CBL and creates strong gradients in the air’s refractive index (or “refractivity”). Radar signals scatter from these turbulent refractivity fluctuations. For turbulent structures with scales equal to half the radar wavelength, the backscattered signals constructively interfere to produce an observable signal in what is called Bragg scattering (e.g., Doviak and Zrnić 1993). Turbulent structures at these scales (∼5 cm for S-band wavelengths) are isotropic; thus, the resulting backscattered signal strength is equal at horizontal and vertical polarizations, leading to intrinsic ZDR values near 0 dB (Melnikov et al. 2011, 2013). In contrast, the CBL often is filled with insects and birds that produce very large intrinsic ZDR values. The combination of biota and Bragg scattering often lead to a local ZDR minimum at the top of the CBL (Banghoff et al. 2018). The height of this ZDR minimum is determined from ZDR quasi-vertical profiles (QVPs; Ryzhkov et al. 2016) constructed using the 4.5° elevation angle scan at each radar site, following Banghoff et al. (2018). These single point observations are provided every 15 min from 1500 UTC 28 April to 0000 UTC 29 April 2019. A horizontal localization of 600 km is used, as suggested by analyses of the horizontal correlation structures, and is similar to the value used for rawinsonde observations. Correlations between CBL height and model variables like potential temperature and water vapor mixing ratio exist well into the free troposphere (Eure et al. 2023), leading to a vertical localization scale factor of 2.3, such that the observations at 1000 hPa have no impact above 100 hPa. The observation error used is 250 m (Banghoff et al. 2018). The observation operator for CBL height is calculated from the mean value of model diagnosed PBL height over a 10 × 10 gridpoint region centered on the grid point closest to the WSR-88D radar, which is roughly the area over which clear-air returns are available.

3) Brightness temperatures from GOES-16 ABI

All-sky (i.e., clear and cloud-affected) infrared BT observations from channel 10 of the GOES-16 Advanced Brightness Imager (ABI; Schmit et al. 2017) and archived on NOAA’s Comprehensive Large Array-data Stewardship System (CLASS) also are assimilated. These observations represent atmospheric thermodynamic conditions over a deep layer when skies are clear but represent cloud-top conditions when skies are cloudy. Because the satellite often views Earth at an angle other than nadir, cloud-top BT observations are displaced (on the order of tens of kilometers) in space relative to their actual geographic location. Given the high resolution of the BT data, it is important to correct for even slight discrepancies in observation location, as such errors could lead to inaccurate placement of convection, and so this parallax error is corrected (Soler and Eisemann 1994; Jones et al. 2020). Cloud height is determined using a cloud-top height product (ACHA) available on CLASS. Because the cloud-top height data are at coarser horizontal resolution (12 km) relative to the BT data (∼2.5 km), the cloud-top height data are interpolated to the scale of BT observations. The Community Radiative Transfer Model (CRTM) is used to generate simulated BTs from the WRF-ARW Model output (Han et al. 2006).

For the assimilation of satellite all-sky infrared BT observations, we apply an adaptive observational error inflation (Minamide and Zhang 2017) using the cloud/clear sky observation errors from Jones et al. (2020) of 4 K for cloudy sky and 2 K for clear sky. When clouds are absent, the observation height value assigned is 5000 m, which is the height of the ABI weighting function peak for channel 10. If there are clouds present, the height assigned is the value derived from the cloud top height product for the interpolated BT coordinates. Following Zhang et al. (2022) a horizontal covariance localization length of 30 km is used, with a vertical covariance localization scale height of 4.0, which translates to the observation no longer having an impact above 18 hPa. The BT observations are available at least every 15 min from 1500 UTC 28 April to 0000 UTC 29 April 2019.

c. Experiment design

All 40 ensemble members are run for 3 h, starting at 1200 UTC 28 April 2019, to allow errors to grow and covariances between model variables to develop. The selected observations are then assimilated over a 4-h window from 1500 to 1900 UTC, with observations for each experiment being assimilated serially every 15 min. A 5-h forecast follows without any data assimilation, providing ensemble predictions for 40 members out to 0000 UTC 29 April 2019.

The first experiment, CONV, only assimilates conventional observations. This experiment assimilates many (but not all) of the observations available operationally to the HRRR. As such, we expect the simulations to strongly resemble the HRRR forecasts. In the second experiment, ALL, conventional observations are assimilated first, followed by WSR-88D radar clear-air radial velocity measurements, then CBL heights derived from WSR-88D ZDR QVPs, and, finally, GOES-16 BTs from channel 10. At any given 15-min assimilation period, there are several hundred conventional observations in the model domain, over 10 000 clear-air SOs, 6 CBL height observations (one for each radar), and over 50 000 BT observations. Calculations of root-mean-square innovation (RMSI) indicate that the observations are being assimilated effectively in all the experiments (not shown).

Two additional experiments are conducted to better understand the contributions from radar and satellite observations to the ensemble forecasts for this case. The RAD experiment assimilates conventional and radar observations (both clear-air radial velocity and CBL heights), whereas the SAT experiment assimilates conventional and GOES-16 BTs from channel 10. All other aspects of these experiments are identical to ALL. Eure et al. (2023) assimilate these same radar and satellite observations within a similar set of experiments for a different case and show that both radar and satellite observations lead to improved forecasts of convection initiation. The results from RAD and SAT are briefly discussed at the end of section 4, as the main message of the study is conveyed by comparing the CONV and ALL ensembles.

4. Results

a. CONV ensemble

The CONV ensemble strongly resembles the operational HRRR forecasts at 0000 UTC 29 April 2019, as expected. As seen in Fig. 1, the operational HRRR has a storm in southwestern Kansas near 38°N and 100.5°W at this time in successive hourly runs. The CONV ensemble produces deep convection in the same location, as seen in both the ensemble mean reflectivity and probability of reflectivity > 35 dBZ fields (Fig. 3). Several grid points have more than 80% of the ensemble members with reflectivity > 35 dBZ (Fig. 3b), while many grid points have ensemble probabilities greater than 50%. Analysis of updraft helicity (UH; Kain et al. 2008), a parameter used to identify midlevel rotation in modeled storms, indicates that all ensemble members produce storms with UH > 50 m2 s−2 in southwestern Kansas, albeit with slight shifts in storm location. Farther north, the CONV ensemble also captures the southwest–northeast-oriented convective line stretching from the Colorado border northeastward (Fig. 3a) that also is seen in the operational HRRR forecasts (cf. Fig. 1).

Fig. 3.
Fig. 3.

Plots of (a) CONV ensemble mean reflectivity (dBZ), (b) pointwise CONV ensemble probability of reflectivity > 35 dBZ, and (c) simulated reflectivity from CONV ensemble member 17 valid at 0000 UTC 29 Apr 2019. Letters A and C indicate regions of convection in the ensemble that are tracked starting at CI. Cells associated with letter B have dissipated. Ensemble member 17 is chosen subjectively as the member that most closely resembles the ensemble mean in terms of location of the convective structures.

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

Our first hypothesis for the consistent prediction of the unobserved supercell is an incorrect prediction of CI by the model. The 2000 UTC ensemble mean reflectivity and the probability of CONV ensemble members with reflectivity > 35 dBZ show that the ensemble produces CI along a west–east line in southeastern Colorado in the same region as observed CI (Figs. 4b,c, groups A and B), with probabilities exceeding 50%. These storms in CONV move eastward into southwestern Kansas and become supercells as soon as they cross the Kansas border, indicated by values of UH exceeding 50 m2 s−2 (not shown). The observed convective cells in northeastern Colorado also are captured by CONV as seen in the ensemble mean reflectivity (Fig. 4b). The locations of these storms differ in the various ensemble members, leading to point-wise ensemble probabilities below 10%. Curiously, CONV also produces a high probability of convection in far northeastern New Mexico. However, this convection decays subsequently and does not appear to have a direct impact on the prediction of the southwestern Kansas supercell. Because convective cells in CONV initiate in the same location and move eastward into southwestern Kansas as observed (albeit more intense than observed), erroneous CI is not responsible for producing the unobserved supercell.

Fig. 4.
Fig. 4.

Plots of (a) observed reflectivity (dBZ), (b) CONV ensemble mean reflectivity, (c) pointwise CONV ensemble probability of reflectivity > 35 dBZ, and (d) simulated reflectivity from CONV ensemble member 17 valid at 2000 UTC 28 Apr 2019, slightly after observed CI in southwestern Kansas. Letters A and B indicate convective cell groups in the ensemble that are tracked starting at CI. Ensemble member 17 is chosen subjectively as the member that most closely resembles the ensemble mean in terms of location of the convective structures.

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

Comparing the simulated and observed storms’ evolution could provide further insight. The observed group A cells initiate at 2000 UTC, move eastward into southwestern Kansas, and decay by 0000 UTC. The observed group B cells that also initiate at 2000 UTC and group C cells that initiate at 2200 UTC are the two cell groups still ongoing by 0000 UTC. In contrast, the group A cells in CONV persist and are supercells at 0000 UTC (Fig. 3), the group B cells have largely dissipated by 0000 UTC, and the group C cells remain weak as they move into Kansas (as observed). Further, the predicted group A cells are supercellular and have motions deviant from the cloud-layer mean wind, whereas the observed group A, B and C storms are ordinary single- or multicellular convection and move largely with the cloud-layer mean wind. Thus, although the cells in CONV all initiated in approximately the same locations as observed, the evolution of cell groups A and B is distinctly different from observations.

This analysis suggests that the predicted environment was favorable for supercells, possibly in contradistinction from the observed environment. Rasmussen and Blanchard (1998) show that supercells tend to occur in environments with CAPE > 1000 J kg−1, 0–6-km bulk wind difference > 19 m s−1, and SRH > 100 m2 s−2. At 2100 UTC 28 April 2019, as the cells in CONV were moving into southwestern Kansas, the ensemble mean values of most unstable CAPE (hereafter MUCAPE1) exceed 2000 J kg−1 in southwestern Kansas (Fig. 5a), indicating an environment thermodynamically favorable for deep convection. In this same region the 0–6-km bulk wind difference values generally are >20 m s−1, except in a local minimum just east of the Colorado–Kansas border (Fig. 5c), and SRH values are >300 m2 s−2 across Kansas (Fig. 5d). These fields all resemble the operational analyses of these variables (not shown) and are within the ranges associated with supercells.

Fig. 5.
Fig. 5.

Plots of (a) MUCAPE (J kg−1), (b) LFC (m), (c) 0–6-km wind shear (m s−1), and (d) SRH (m2 s−2) from the CONV ensemble mean at 2100 UTC 28 Apr 2019.

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

Time series of maximum ensemble mean MUCAPE, minimum ensemble mean CIN, and minimum ensemble mean LFC predicted at any grid point within a 1° × 1° box in far southwestern Kansas show that maximum MUCAPE doubles from 1500 to 3000 J kg−1 between 1900 and 2200 UTC (Fig. 6). Simultaneously the values of minimum CIN decrease from 75 to 25 J kg−1 and remain at this low value for the next 3 h. Along with the decrease in CIN there is a decrease in the minimum LFC, which reaches values ≤ 1500 m by 2200 UTC. Thus, not only are values of MUCAPE increasing but the environment is evolving favorably for convection initiation (see Lock and Houston 2014). These changes suggest sustained low-level lift has deepened boundary layer moisture and weakened the overlying capping inversion, which is verified by examination of soundings in the region (not shown). Thus, it is unsurprising that the group A cells in CONV persist as they move into southwestern Kansas. To evaluate whether the CONV-predicted environment was overly favorable for supercells, we assess the ALL experiment, which is expected to better represent the preconvective environment owing to the assimilation of additional remotely sensed observations.

Fig. 6.
Fig. 6.

Time series of ensemble mean (a) maximum MUCAPE (J kg−1), (b) minimum LFC (m), and (c) minimum CIN (J kg−1) at any grid point within a 1° × 1° box in far southwestern Kansas.

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

b. ALL ensemble

The ALL ensemble mean reflectivity and probability of reflectivity > 35 dBZ in southwestern Kansas are smaller than CONV at 0000 UTC 29 April 2019 (cf. Figs. 3 and 7). The ALL probabilities are <40%, half those in CONV, and the ALL ensemble mean reflectivity is <25 dBZ (Fig. 7a). In part, this is because fewer ALL members produce convection in southwestern Kansas compared to CONV. Yet, farther north, the convective line near the Colorado border in ALL has larger ensemble mean reflectivity than CONV. The convective line in ALL also is shifted slightly westward from its location in CONV, which qualitatively agrees better with observations in which the line’s westward extent reaches the Colorado–Kansas border.

Fig. 7.
Fig. 7.

Plots of (a) ALL ensemble mean reflectivity (dBZ), (b) pointwise ALL ensemble probability of reflectivity > 35 dBZ, and (c) simulated reflectivity from ALL ensemble member 13 valid at 0000 UTC 29 Apr 2019. Letters B and C indicate regions of convection in the ensemble that are tracked starting at CI. Ensemble member 13 is chosen subjectively as the member that most closely resembles the ensemble mean in terms of location of the convective structures.

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

The ALL ensemble mean reflectivity and the probability of ensemble members with reflectivity > 35 dBZ shortly after the observed CI in southwestern Colorado at 2000 UTC shows that ALL produces CI in several convective cell groups in the same region as observed (Fig. 8, groups A and B), with probabilities exceeding 40%. These probabilities are up to 10% less than seen with CONV, but in similar locations. The region of CI in northeastern New Mexico also occurs, but as seen with CONV, these cells decay over the next few hours and do not appear to influence the cells to their north.

Fig. 8.
Fig. 8.

Plots of (a) observed reflectivity (dBZ), (b) ALL ensemble mean reflectivity, (c) pointwise ALL ensemble probability of reflectivity > 35 dBZ, and (d) simulated reflectivity from ALL ensemble member 13 valid at 2000 UTC 28 Apr 2019, slightly after observed CI in southwestern Kansas. Points A and B indicate regions of convection in the ensemble that are tracked starting at CI. Ensemble member 13 is chosen subjectively as the member that most closely resembles the ensemble mean in terms of location of the convective structures.

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

Further, the evolution of the convective cell groups in ALL differs from that seen in CONV. In ALL, many ensemble members produce cells in group A that move eastward into Kansas and begin to dissipate by 2200 UTC, consistent with observations. However, other ensemble members produce cells in group A that are supercells, as in CONV. The group B cells in ALL also move eastward into Kansas and persist until 0000 UTC 29 April, including more supercells. Overall, half of the ALL members feature supercells, with the majority tracing their roots to their initiation with the group B cells. Thus, not only does the location and timing of CI in ALL agree reasonably well with observations, but the demise of cell group A convection in many ensemble members also parallels the observations. This suggests that predicted environmental differences between CONV and ALL are the dominant reason for the distinct behaviors of the simulated storms.

Indeed, the ALL ensemble mean environmental MUCAPE and LFC at 2100 UTC (Fig. 9) differ markedly from those in CONV (cf. Fig. 5), with MUCAPE < 1500 J kg−1 in much of western Kansas (Fig. 9a), and higher values of LFC (Fig. 9b). These fields indicate an environment that, while still supportive of deep convection, is also less favorable for deep convection than predicted by CONV. These differences persist as seen in the time series of ensemble mean values in southwestern Kansas, which show that CONV has larger maximum MUCAPE, lower minimum LFC, and lower minimum CIN than ALL (Fig. 6). Differences in wind shear and SRH between CONV and ALL are smaller, although the wind shear in ALL is slightly larger than in CONV in parts of western Kansas. After an examination of the ensemble members, we hypothesize that the reduced MUCAPE and higher LFC and CIN in ALL are related to a reduction in low-level northward moisture flux, which modifies the environment to hamper renewed cell development for some of the group A cells.

Fig. 9.
Fig. 9.

Plots of (a) MUCAPE (J kg−1), (b) LFC (m), (c) 0–6-km wind shear (m s−1), and (d) SRH (m2 s−2) from the ALL ensemble mean at 2100 UTC 28 Apr 2019. Color bars for each panel indicate the values of the diagnosed variables.

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

To evaluate this hypothesis, we investigate the ensemble mean lower-tropospheric moisture flux into the convective region. Results show that the northward moisture flux into the region is reduced in ALL compared to CONV (Fig. 10), helping to explain the reduced MUCAPE and increased LFC and CIN. This reduced moisture flux is attributed to both weaker low-level winds in ALL, as well as reduced water vapor mixing ratios above and near the top of the boundary layer owing to the assimilation of clear-air radial velocity, CBL depth, and BT observations, as shown in the 2200 UTC predicted soundings at Dodge City, Kansas (Fig. 11). It appears that the changing thermodynamic environment plays a role in the cell group evolution. The CONV ensemble appears to be a few hours faster in producing the more favorable environment for deep convection compared to ALL, which leads to the group A cells persisting in CONV (not shown). In contrast, the group A cells in ALL dissipate while group B and C cells persist. The operational forecasts from the HRRR evolve the environment in southwestern Kansas similar to that in CONV (not shown), providing evidence that the incorrect prediction of environments overly favorable for supercells may have contributed to the erroneous storm forecasts.

Fig. 10.
Fig. 10.

Ensemble mean meridional water vapor flux (kg m s−1) vs time (UTC) across 37°N between the surface and 700 hPa and extending from the western edge of the model domain to 100°W to capture the extent of the moist tongue. Values are calculated once an hour from 1500 UTC 28 Apr to 0000 UTC 29 Apr 2019 from CONV (blue line), RAD (short dash orange line), SAT (long dash gray line), and ALL (gold line) ensembles.

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

Fig. 11.
Fig. 11.

Ensemble mean sounding from the grid point closest to Dodge City, KS, at 2200 UTC 28 Apr 2019 from (a) CONV and (b) ALL ensembles.

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

To further test the importance of the meridional low-level water vapor flux to convective evolution, 5 members from ALL that produce weak convection near Dodge City are selected (best members) along with 5 members from ALL that produce supercells near Dodge City (worst members). The mean meridional water vapor flux is calculated from the two 5-member groups and compared, with results showing that the best members have smaller meridional water vapor flux than the worst members (Fig. 12); the total flux over the 9-h period is 12% less for the best members compared to the worst members. The best members also have smaller values of MUCAPE, higher LFC, larger CIN, smaller 0–6-km shear, and smaller SRH than the worst members (not shown).

Fig. 12.
Fig. 12.

Ensemble mean meridional water vapor flux (kg m s−1) vs time (UTC) across 37°N between the surface and 700 hPa and extending from the western edge of the model domain to 100°W to capture the extent of the moist tongue. Values are calculated once an hour from 1500 UTC 28 Apr to 0000 UTC 29 Apr 2019 from the five best performing members in ALL (gold line) and the five worst performing members in ALL (blue line).

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

If meridional low-level water vapor flux is a key determining factor in determining convective mode for this case, then one would expect different convective modes for runs with the largest and smallest meridional low-level water vapor fluxes. The CONV and ALL ensemble members are combined into an 80-member group and the 5 members with the largest and smallest meridional low-level water vapor fluxes are compared. Results show that the 5 members with the largest meridional water vapor flux all have strong supercells in southwestern Kansas, whereas the 5 members with the smallest water vapor flux all have only weak convection in southwestern Kansas (not shown). These analyses confirm that the strength of the meridional low-level water vapor flux played an important role in the development of supercell storms in the ensembles for this case.

c. RAD and SAT ensembles

The relative contributions from assimilating radar and satellite observations on the resulting ensemble forecasts can be investigated by comparing forecasts from the RAD and SAT ensembles with CONV and ALL. The predictions of CI from all four ensembles are very similar to each other (not shown). Thus, the focus again turns to differences in the preconvective environment and the mean meridional water vapor flux. The MUCAPE and LFC fields at 2100 UTC from RAD strongly resemble these fields in CONV (cf. Figs. 5a,b and 13a,b), although RAD shows a reduction in MUCAPE and an increase in LFC in southeastern Colorado as the early cells move into Kansas. In contrast, the MUCAPE and LFC fields at 2100 UTC from SAT are a blend of CONV and ALL (cf. Figs. 5a,b, 9a,b, and 13c,d). Values of MUCAPE > 1500 J kg−1 extend farther northward into southwestern Kansas in SAT compared to ALL, but not as far northward compared to CONV. SAT has the lowest values of LFC within the moist tongue in comparison with CONV, RAD, and ALL. When evaluating 0–6-km wind differences and SRH, RAD again resembles CONV and SAT is a blend of CONV and ALL (not shown).

Fig. 13.
Fig. 13.

Plots of (a) MUCAPE (J kg−1) and (b) LFC (m) from the RAD ensemble mean, and (c) MUCAPE and (d) LFC from the SAT ensemble mean at 2100 UTC 28 Apr 2019. Color bars for each panel indicate the values of the diagnosed variables. Compare with Figs. 5 and 9 for CONV and ALL.

Citation: Weather and Forecasting 38, 10; 10.1175/WAF-D-23-0010.1

The similarity of the environments from RAD with CONV suggests that the impact of radar observations decays within a few hours after the end of the assimilation window. In contrast, the differences between the environments from SAT and CONV suggests that the impact of satellite observations is retained for a longer time after the assimilation window. However, 38 ensemble members in RAD and 33 ensemble members in SAT produce supercells, albeit there is greater variation in supercell location in SAT compared to RAD and CONV (not shown).

Additional information is provided by examining the time series of northward low-level water vapor flux within the moist tongue (Fig. 10). Ensemble mean water vapor flux from RAD and SAT are between the values from CONV and ALL from 1730 to 2100 UTC. Early in this time window RAD has smaller meridional fluxes while after 2000 UTC SAT has smaller meridional fluxes. The meridional fluxes in SAT closely parallel those in ALL from 2000 to 2400 UTC and are much smaller than seen in RAD and CONV. These results suggest that radar observations have the largest impact on water vapor fluxes during the assimilation window and extending for 2 h into the ensemble forecasts after which time the satellite observations have the largest impact. The combined changes in meridional water vapor flux from assimilating data from both observational systems are not simply the sums of the two parts, but rather these observation platforms provide synergistic information that leads to even smaller meridional water vapor fluxes in ALL.

5. Discussion

The hourly HRRR forecasts on 28 April 2019 repeatedly produced a supercell storm in southwestern Kansas late in the day that was not observed, an unusual situation that merits exploration. Results from the CONV ensemble that only assimilated conventional observations indicate that all members produce long-lived supercells in southwestern Kansas, in agreement with the operational HRRR forecasts. In contrast, the ALL ensemble that assimilates WSR-88D and GOES-16 observations in addition to the conventional observations has 50% fewer supercells in this region, with many members predicting a convective evolution closer to observations. The assimilated WSR-88D observations are clear-air radial velocities and CBL depth, while the GOES-16 observations are all-sky BTs. Results show that the predictions of CI from the two ensembles are very similar in both timing and location, suggesting that differences in initiation in the two ensembles are not responsible for the improvements in convective evolution seen in ALL. Further exploration of the two ensembles indicates that their preconvective environments differ in the values of MUCAPE, LFC, and CIN within the moist tongue, with the ALL run predicting smaller MUCAPE, larger CIN, and higher LFC compared to CONV. Comparisons of various ensemble members suggest that these differences are due to a reduced northward low-level water vapor flux within the moist tongue in ALL compared to CONV.

The relative contributions from assimilating radar and satellite observations on the resulting ensemble forecasts are investigated by exploring the forecasts from the RAD and SAT ensembles. Results suggest that radar observations have the largest impact on water vapor fluxes during the assimilation window and for a few hours afterward, likely owing to their direct influence on the moisture and wind fields. The impacts of satellite observations are retained longer into the forecast through more accurate feedbacks between convection and the environment. The results clearly show that only by assimilating both radar and satellite observations can the ensemble forecasts reduce the number of members with the unobserved supercell storm by half. Radar and satellite observations provide synergistic information that change the environment beyond a simple additive effect, as also suggested by Eure et al. (2023).

It is unexpected that the HRRR forecasts on this day repeatedly produced a supercell storm in southwestern Kansas that was not observed, especially when new observations are assimilated every hour. The 40-member CONV ensemble, which assimilates similar observations, replicates this outcome. This situation is reminiscent of behaviors seen with large-scale models in which environmental errors prove to be difficult to remove from one assimilation cycle to the next (e.g., Fritsch et al. 2000). It often takes multiple assimilation cycles to correct model errors even when presented with observations that correctly depict the environment. The persistent unobserved supercell prediction from the HRRR suggests that the first guess from the earlier HRRR runs dominate during data assimilation without copious observations to refute this first guess.

Results from the ALL ensemble show that the assimilation of WSR-88D clear-air radial velocities, WSR-88D diagnosed CBL heights, and GOES-16 all-sky infrared BTs alter the prediction of MUCAPE, CIN, and LFC to be less favorable for deep convection, owing in large part to a notable reduction in the low-level meridional water vapor flux. Given that both ensembles assimilate surface observations, these results suggest that the WSR-88D and GOES-16 observations are modifying the atmosphere within the upper portion of the boundary layer and into the middle troposphere. Thus, it appears that the conventional observation gap between 10 m above the surface and a few kilometers AGL (see NRC 2009) impacted this event negatively, which was then ameliorated to some extent by assimilating the WSR-88D and GOES-16 observations. The assimilation of these observations has lesser, but still notable, impacts on 0–6-km wind shear and SRH. While the relative value of these observations compared to other operational data is unknown, their assimilation altered the ALL ensemble in ways that moved the convective evolution closer to observations. Further studies assimilating these observations are needed to determine their impact over many cases. As these observations are available from current operational data streams, they could more easily be included in data assimilation systems to help reduce the low-level observation gap. Boundary layer sensors, such as profilers, lidars, and radiometers, also can help to reduce the conventional observation gap and studies have shown that these observations lead to improvements in the resulting forecasts when assimilated (e.g., Otkin et al. 2011; Hartung et al. 2011; Coniglio et al. 2019; Degelia et al. 2019, 2020).

This case also raises questions about mesoscale forecast uncertainty and how forecasters should interpret forecast-to-forecast consistency from operational modeling systems. One of the hopes of ensemble forecasting systems is that they are more likely to capture the observations within the ensemble forecast. Results from CONV indicate that nearly all members produced a supercell in southwestern Kansas; it would be hard for a forecaster to dismiss the consistency seen across members in these forecasts even a few hours in advance of the predicted events. Dealing with these types of incorrect forecast events will likely continue to challenge forecasters until we can provide observations to data assimilation systems that capture mesoscale structures routinely within the lower tropospheric conventional observational gap.

1

The parcel used to calculate MUCAPE, convective inhibition (CIN), and LFC has the largest equivalent potential temperature when averaged over a 500-m-deep layer in the lowest 3000 m of the column.

Acknowledgments.

We are very thankful for and acknowledge the assistance that both Sam Degelia and Dr. Xuguang Wang of the University of Oklahoma provided in modifying code to successfully assimilate the satellite and radar observations. John Banghoff is thanked for all his help regarding the QVP measurements. The helpful and constructive reviews of three anonymous reviewers improved the manuscript and are appreciated. Furthermore, we gratefully acknowledge the countless contributions from colleagues at NOAA’s Global System Division (GSD) at Earth System Research Laboratory (ESRL) in Boulder, Colorado; the Environmental Modeling Center (EMC) within the National Center for Environmental Prediction (NCEP); as well as the National Severe Storms Laboratory (NSSL) in Norman, Oklahoma. This project began as a collaborative project with Dr. Fuqing Zhang of The Pennsylvania State University, and we honor his many contributions to this project and others. He is missed. Funding for this work is provided by NOAA Award NA18OAR4590369 as well as the Future Investigators in NASA Earth and Space Science Technology (FINESST) Grant 80NSSC20K1619. This work solely represents the view of the authors and does not necessarily represent the view of NOAA or NASA. Initial experiments were conducted on the University of Texas’s Stampede2 and Hera supercomputers.

Data availability statement.

WSR-88D observations used in this study are available from the National Center for Environmental Information (NCEI). GOES-16 satellite brightness temperatures are available from NOAA’s CLASS. Conventional observations are archived at NCAR. The GSI-EnKF is available from the Developmental Testbed Center (DTC) of UCAR. Version 3.8 of WRF was downloaded from the University Corporation for Atmospheric Research (UCAR). Experiment results can be shared upon request to the authors.

REFERENCES

  • Banghoff, J. R., D. J. Stensrud, and M. R. Kumjian, 2018: Convective boundary layer depth estimation from S-band dual-polarization radar. J. Atmos. Oceanic Technol., 35, 17231733, https://doi.org/10.1175/JTECH-D-17-0210.1.

    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., G. A. Grell, J. M. Brown, and T. G. Smirnova, 2004: Mesoscale weather prediction with the RUC hybrid isentropic-terrain-following coordinate model. Mon. Wea. Rev., 132, 473494, https://doi.org/10.1175/1520-0493(2004)132<0473:MWPWTR>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., and Coauthors, 2016: A North American hourly assimilation and model forecast cycle: The Rapid Refresh. Mon. Wea. Rev., 144, 16691694, https://doi.org/10.1175/MWR-D-15-0242.1.

    • Search Google Scholar
    • Export Citation
  • Brooks, H. E., C. A. Doswell III, and J. Cooper, 1994: On the environments of tornadic and nontornadic mesocyclones. Wea. Forecasting, 9, 606618, https://doi.org/10.1175/1520-0434(1994)009<0606:OTEOTA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Coniglio, M. C., G. S. Romine, D. D. Turner, and R. D. Torn, 2019: Impacts of targeted AERI and Doppler lidar wind retrievals on short-term forecasts of the initiation and early evolution of thunderstorms. Mon. Wea. Rev., 147, 11491170, https://doi.org/10.1175/MWR-D-18-0351.1.

    • Search Google Scholar
    • Export Citation
  • Degelia, S. K., X. Wang, and D. J. Stensrud, 2019: An evaluation of the impact of assimilating AERI retrievals, kinematic profilers, rawinsondes, and surface observations on a forecast of a nocturnal convection initiation event during the PECAN field campaign. Mon. Wea. Rev., 147, 27392764, https://doi.org/10.1175/MWR-D-18-0423.1.

    • Search Google Scholar
    • Export Citation
  • Degelia, S. K., X. Wang, D. J. Stensrud, and D. D. Turner, 2020: Systematic evaluation of the impact of assimilating a network of ground-based remote sensing profilers for forecasts of nocturnal convection initiation during PECAN. Mon. Wea. Rev., 148, 47034728, https://doi.org/10.1175/MWR-D-20-0118.1.

    • Search Google Scholar
    • Export Citation
  • Done, J., C. Davis, and M. Weisman, 2004: The next generation of NWP: Explicit forecasts of convection using the Weather Research and Forecasting (WRF) Model. Atmos. Sci. Lett., 5, 110117, https://doi.org/10.1002/asl.72.

    • Search Google Scholar
    • Export Citation
  • Doviak, R. J., and D. S. Zrnić, 1993: Doppler Radar and Weather Observations. Academic Press, 562 pp.

  • Dowell, D. C., F. Zhang, L. J. Wicker, C. Snyder, and N. A. Crook, 2004: Wind and temperature retrievals in the 17 May 1981 Arcadia, Oklahoma, supercell: Ensemble Kalman filter experiments. Mon. Wea. Rev., 132, 19812005, https://doi.org/10.1175/1520-0493(2004)132<1982:WATRIT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Dowell, D. C., and Coauthors, 2022: The High-Resolution Rapid Refresh (HRRR): An hourly updating convection-allowing forecast model. Part I: Motivation and system description. Wea. Forecasting, 37, 13711395, https://doi.org/10.1175/WAF-D-21-0151.1.

    • Search Google Scholar
    • Export Citation
  • Duda, J. D., and W. A. Gallus Jr., 2010: Spring and summer Midwestern severe weather reports in supercells compared to other morphologies. Wea. Forecasting, 25, 190206, https://doi.org/10.1175/2009WAF2222338.1.

    • Search Google Scholar
    • Export Citation
  • Eilts, M. D., and S. D. Smith, 1990: Efficient dealiasing of Doppler velocities using local environment constraints. J. Atmos. Oceanic Technol., 7, 118128, https://doi.org/10.1175/1520-0426(1990)007<0118:EDODVU>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Eure, K. C., P. D. Mykolajtchuk, Y. Zhang, D. J. Stensrud, F. Zhang, S. J. Greybush, and M. R. Kumjian, 2023: Simultaneous assimilation of planetary boundary layer observations from radar and all-sky satellite observations to improve forecasts of convection initiation. Mon. Wea. Rev, 151, 795813, https://doi.org/10.1175/MWR-D-22-0188.1.

    • Search Google Scholar
    • Export Citation
  • Fritsch, J. M., J. Hilliker, J. Ross, and R. L. Vislocky, 2000: Model consensus. Wea. Forecasting, 15, 571582, https://doi.org/10.1175/1520-0434(2000)015<0571:MC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Grim, J. A., J. O. Pinto, T. Blitz, K. Stone, and D. C. Dowell, 2022: Biases in the prediction of convective storm characteristics with a convection allowing ensemble. Wea. Forecasting, 37, 6583, https://doi.org/10.1175/WAF-D-21-0106.1.

    • Search Google Scholar
    • Export Citation
  • Han, Y., P. van Delst, Q. Liu, F. Weng, B. Yan, R. Treadon, and J. Derber, 2006: JCSDA Community Radiative Transfer Model (CRTM)—Version 1. NOAA Tech. Rep. NESDIS 122, 33 pp., https://repository.library.noaa.gov/view/noaa/1157.

    • Search Google Scholar
    • Export Citation
  • Hartung, D. C., J. A. Otkin, R. A. Petersen, D. D. Turner, and W. F. Feltz, 2011: Assimilation of surface-based boundary layer profiler observations during a cool-season weather event using an observing system simulation experiment. Part II: Forecast assessment. Mon. Wea. Rev., 139, 23272346, https://doi.org/10.1175/2011MWR3623.1.

    • Search Google Scholar
    • Export Citation
  • 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.

    • Search Google Scholar
    • Export Citation
  • Johns, R. H., and C. A. Doswell III, 1992: Severe local storm forecasting. Wea. Forecasting, 7, 588612, https://doi.org/10.1175/1520-0434(1992)007<0588:SLSF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Johnson, A., X. Wang, J. R. Carley, L. J. Wicker, and C. Karstens, 2015: A comparison of multiscale GSI-based EnKF and 3DVar data assimilation using radar and conventional observations for midlatitude convective-scale precipitation forecasts. Mon. Wea. Rev., 143, 30873108, https://doi.org/10.1175/MWR-D-14-00345.1.

    • Search Google Scholar
    • Export Citation
  • Johnson, A., X. Wang, and S. Degelia, 2016: Design and implementation of a GSI-based convection-allowing ensemble based data assimilation and forecast system for the PECAN field experiment. Part II: Overview and evaluation of real-time system. Wea. Forecasting, 32, 12271251, https://doi.org/10.1175/WAF-D-16-0201.1.

    • Search Google Scholar
    • Export Citation
  • Johnson, A., X. Wang, and T. Jones, 2022: Impacts of assimilating GOES-16 ABI channels 9 and 10 clear air and cloudy radiance observations with additive inflation and adaptive observation error in GSI-EnKF for a case of rapidly evolving severe supercells. J. Geophys. Res. Atmos., 127, e2021JD036157, https://doi.org/10.1029/2021JD036157.

    • Search Google Scholar
    • Export Citation
  • Jones, T. A., and Coauthors, 2020: Assimilation of GOES-16 radiances and retrievals into the Warn-on-Forecast System. Mon. Wea. Rev., 148, 18291859, https://doi.org/10.1175/MWR-D-19-0379.1.

    • Search Google Scholar
    • Export Citation
  • Kain, J. S., and Coauthors, 2008: Some practical considerations regarding horizontal resolution in the first generation of operational convection-allowing NWP. Wea. Forecasting, 23, 931952, https://doi.org/10.1175/WAF2007106.1.

    • Search Google Scholar
    • Export Citation
  • Kain, J. S., and Coauthors, 2013: A feasibility study for probabilistic convection initiation forecasts based on explicit numerical guidance. Bull. Amer. Meteor. Soc., 94, 12131225, https://doi.org/10.1175/BAMS-D-11-00264.1.

    • Search Google Scholar
    • Export Citation
  • Kleist, D. T., D. F. Parrish, J. C. Derber, R. Treadon, W.-S. Wu, and S. Lord, 2009: Introduction of the GSI into the NCEP global data assimilation system. Wea. Forecasting, 24, 16911705, https://doi.org/10.1175/2009WAF2222201.1.

    • Search Google Scholar
    • Export Citation
  • Liu, H., M. Hu, G. Ge, D. Stark, H. Shao, K. Newman, and J. Whitaker, 2018: Ensemble Kalman filter (EnKF) user’s guide version 1.3. Developmental Testbed Center Doc., 80 pp., https://dtcenter.org/community-code/ensemble-kalman-filter-system-enkf/documentation.

  • Lock, N. A., and A. L. Houston, 2014: Empirical examination of the factors regulating thunderstorm initiation. Mon. Wea. Rev., 142, 240258, https://doi.org/10.1175/MWR-D-13-00082.1.

    • Search Google Scholar
    • Export Citation
  • Melnikov, V. M., R. J. Doviak, D. S. Zrnić, and D. J. Stensrud, 2011: Mapping Bragg scatter with a polarimetric WSR-88D. J. Atmos. Oceanic Technol., 28, 12731285, https://doi.org/10.1175/JTECH-D-10-05048.1.

    • Search Google Scholar
    • Export Citation
  • Melnikov, V. M., R. J. Doviak, D. S. Zrnić, and D. J. Stensrud, 2013: Structures of Bragg scatter observed with the polarimetric WSR-88D. J. Atmos. Oceanic Technol., 30, 12531258, https://doi.org/10.1175/JTECH-D-12-00210.1.

    • Search Google Scholar
    • Export Citation
  • Minamide, M., and F. Zhang, 2017: Adaptive observation error inflation for assimilating all-sky satellite radiance. Mon. Wea. Rev., 145, 10631081, https://doi.org/10.1175/MWR-D-16-0257.1.

    • Search Google Scholar
    • Export Citation
  • Moller, A. R., C. A. Doswell III, M. P. Foster, and G. R. Woodall, 1994: The operational recognition of supercell thunderstorm environments and storm structures. Wea. Forecasting, 9, 327347, https://doi.org/10.1175/1520-0434(1994)009<0327:TOROST>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Nakanishi, M., and H. Niino, 2009: Development of an improved turbulence closure model for the atmospheric boundary layer. J. Meteor. Soc. Japan, 87, 895912, https://doi.org/10.2151/jmsj.87.895.

    • Search Google Scholar
    • Export Citation
  • National Research Council, 2009: Observing Weather and Climate from the Ground Up: A Nationwide Network of Networks. National Academies Press, 234 pp., https://doi.org/10.17226/12540.

  • NCEI, 2022: U.S. billion-dollar weather and climate disasters, 1980-present. NOAA/NCEI, accessed 15 August 2022, https://doi.org/10.25921/stkw-7w73.

  • Olson, J. B., J. S. Kenyon, W. A. Angevine, J. M. Brown, M. Pagowski, and K. Sušelj, 2019: A description of the MYNN-EDMF scheme and the coupling to other components in WRF-ARW. NOAA Tech. Memo. OAR GSD 61, 37 pp., https://doi.org/10.25923/n9wm-be49.

  • Otkin, J. A., D. C. Hartung, D. D. Turner, R. A. Petersen, W. F. Feltz, and E. Janzon, 2011: Assimilation of surface-based boundary layer profiler observations during a cool-season weather event using an observing system simulation experiment. Part I: Analysis impact. Mon. Wea. Rev., 139, 23092326, https://doi.org/10.1175/2011MWR3622.1.

    • Search Google Scholar
    • Export Citation
  • Rasmussen, E. N., and D. O. Blanchard, 1998: A baseline climatology of sounding-derived supercell and tornado forecast parameters. Wea. Forecasting, 13, 11481164, https://doi.org/10.1175/1520-0434(1998)013<1148:ABCOSD>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Ryzhkov, A., P. Zhang, H. Reeves, M. Kumjian, T. Tschallener, S. Trömel, and C. Simmer, 2016: Quasi-vertical profiles—A new way to look at polarimetric radar data. J. Atmos. Oceanic Technol., 33, 551562, https://doi.org/10.1175/JTECH-D-15-0020.1.

    • Search Google Scholar
    • Export Citation
  • Schmit, T. J., P. Griffith, M. M. Gunshor, J. M. Daniels, S. J. Goodman, and W. J. Lebair, 2017: A closer look at the ABI on the GOES-R series. Bull. Amer. Meteor. Soc., 98, 681698, https://doi.org/10.1175/BAMS-D-15-00230.1.

    • Search Google Scholar
    • Export Citation
  • Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp., https://doi.org/10.5065/D68S4MVH.

  • Soler, T., and D. W. Eisemann, 1994: Determination of look angles to geostationary communication satellites. J. Surv. Eng., 120, 115127, https://doi.org/10.1061/(ASCE)0733-9453(1994)120:3(115).

    • Search Google Scholar
    • Export Citation
  • Stelten, S., and W. A. Gallus Jr., 2017: Pristine nocturnal convective initiation: A climatology and preliminary examination of predictability. Wea. Forecasting, 32, 16131635, https://doi.org/10.1175/WAF-D-16-0222.1.

    • Search Google Scholar
    • Export Citation
  • 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, 50955115, https://doi.org/10.1175/2008MWR2387.1.

    • Search Google Scholar
    • Export Citation
  • Thompson, R. L., C. M. Mead, and R. Edwards, 2007: Effective storm-relative helicity and bulk shear in supercell thunderstorm environments. Wea. Forecasting, 22, 102115, https://doi.org/10.1175/WAF969.1.

    • Search Google Scholar
    • Export Citation
  • Tochimoto, E., and H. Niino, 2016: Structural and environmental characteristics of extratropical cyclones that cause tornado outbreaks in the warm sector: A composite study. Mon. Wea. Rev., 144, 945969, https://doi.org/10.1175/MWR-D-15-0015.1.

    • Search Google Scholar
    • Export Citation
  • Wang, X., and T. Lei, 2014: GSI-based four-dimensional ensemble–variational (4DEnsVar) data assimilation: Formulation and single-resolution experiments with real data for NCEP global forecast system. Mon. Wea. Rev., 142, 33033325, https://doi.org/10.1175/MWR-D-13-00303.1.

    • Search Google Scholar
    • Export Citation
  • Wang, X., D. Parrish, D. Kleist, and J. Whitaker, 2013: GSI 3DVar-based ensemble-variational hybrid data assimilation for NCEP global forecast system: Single resolution experiments. Mon. Wea. Rev., 141, 40984117, https://doi.org/10.1175/MWR-D-12-00141.1.

    • Search Google Scholar
    • Export Citation
  • Weisman, M. L., C. Davis, W. Wang, K. W. Manning, and J. B. Klemp, 2008: Experiences with 0–36-h explicit convective forecasts with the WRF-ARW Model. Wea. Forecasting, 23, 407437, https://doi.org/10.1175/2007WAF2007005.1.

    • Search Google Scholar
    • Export Citation
  • Whitaker, J. S., and T. M. Hamill, 2002: Ensemble data assimilation without perturbed observations. Mon. Wea. Rev., 130, 19131924, https://doi.org/10.1175/1520-0493(2002)130<1913:EDAWPO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Whitaker, J. S., and T. M. Hamill, 2012: Evaluating methods to account for system errors in ensemble data assimilation. Mon. Wea. Rev., 140, 30783089, https://doi.org/10.1175/MWR-D-11-00276.1.

    • Search Google Scholar
    • Export Citation
  • Wu, W.-S., R. J. Pursuer, and D. F. Parrish, 2002: Three-dimensional variational analysis with spatially inhomogeneous covariances. Mon. Wea. Rev., 130, 29052916, https://doi.org/10.1175/1520-0493(2002)130<2905:TDVAWS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Zhang, F., C. Snyder, and J. Sun, 2004: Impacts of initial estimate and observation availability on convective-scale data assimilation with an ensemble Kalman filter. Mon. Wea. Rev., 132, 12381253, https://doi.org/10.1175/1520-0493(2004)132<1238:IOIEAO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Zhang, F., Y. Weng, J. A. Sippel, Z. Meng, and C. H. Bishop, 2009: Cloud-resolving hurricane initialization and prediction through assimilation of Doppler radar observations with an ensemble Kalman filter. Mon. Wea. Rev., 137, 21052125, https://doi.org/10.1175/2009MWR2645.1.

    • Search Google Scholar
    • Export Citation
  • Zhang, Y., D. J. Stensrud, and F. Zhang, 2015: Practical predictability of the 20 May 2013 tornadic thunderstorm event in Oklahoma: Sensitivity to synoptic timing and topographical influence. Mon. Wea. Rev., 143, 29732997, https://doi.org/10.1175/MWR-D-14-00394.1.

    • Search Google Scholar
    • Export Citation
  • Zhang, Y., E. E. Clothiaux, and D. J. Stensrud, 2022: Correlation structures between satellite all-sky infrared brightness temperatures and the atmospheric state at storm scales. Adv. Atmos. Sci., 39, 714732, https://doi.org/10.1007/s00376-021-0352-3.

    • Search Google Scholar
    • Export Citation
Save
  • Banghoff, J. R., D. J. Stensrud, and M. R. Kumjian, 2018: Convective boundary layer depth estimation from S-band dual-polarization radar. J. Atmos. Oceanic Technol., 35, 17231733, https://doi.org/10.1175/JTECH-D-17-0210.1.

    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., G. A. Grell, J. M. Brown, and T. G. Smirnova, 2004: Mesoscale weather prediction with the RUC hybrid isentropic-terrain-following coordinate model. Mon. Wea. Rev., 132, 473494, https://doi.org/10.1175/1520-0493(2004)132<0473:MWPWTR>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., and Coauthors, 2016: A North American hourly assimilation and model forecast cycle: The Rapid Refresh. Mon. Wea. Rev., 144, 16691694, https://doi.org/10.1175/MWR-D-15-0242.1.

    • Search Google Scholar
    • Export Citation
  • Brooks, H. E., C. A. Doswell III, and J. Cooper, 1994: On the environments of tornadic and nontornadic mesocyclones. Wea. Forecasting, 9, 606618, https://doi.org/10.1175/1520-0434(1994)009<0606:OTEOTA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Coniglio, M. C., G. S. Romine, D. D. Turner, and R. D. Torn, 2019: Impacts of targeted AERI and Doppler lidar wind retrievals on short-term forecasts of the initiation and early evolution of thunderstorms. Mon. Wea. Rev., 147, 11491170, https://doi.org/10.1175/MWR-D-18-0351.1.

    • Search Google Scholar
    • Export Citation
  • Degelia, S. K., X. Wang, and D. J. Stensrud, 2019: An evaluation of the impact of assimilating AERI retrievals, kinematic profilers, rawinsondes, and surface observations on a forecast of a nocturnal convection initiation event during the PECAN field campaign. Mon. Wea. Rev., 147, 27392764, https://doi.org/10.1175/MWR-D-18-0423.1.

    • Search Google Scholar
    • Export Citation
  • Degelia, S. K., X. Wang, D. J. Stensrud, and D. D. Turner, 2020: Systematic evaluation of the impact of assimilating a network of ground-based remote sensing profilers for forecasts of nocturnal convection initiation during PECAN. Mon. Wea. Rev., 148, 47034728, https://doi.org/10.1175/MWR-D-20-0118.1.

    • Search Google Scholar
    • Export Citation
  • Done, J., C. Davis, and M. Weisman, 2004: The next generation of NWP: Explicit forecasts of convection using the Weather Research and Forecasting (WRF) Model. Atmos. Sci. Lett., 5, 110117, https://doi.org/10.1002/asl.72.

    • Search Google Scholar
    • Export Citation
  • Doviak, R. J., and D. S. Zrnić, 1993: Doppler Radar and Weather Observations. Academic Press, 562 pp.

  • Dowell, D. C., F. Zhang, L. J. Wicker, C. Snyder, and N. A. Crook, 2004: Wind and temperature retrievals in the 17 May 1981 Arcadia, Oklahoma, supercell: Ensemble Kalman filter experiments. Mon. Wea. Rev., 132, 19812005, https://doi.org/10.1175/1520-0493(2004)132<1982:WATRIT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Dowell, D. C., and Coauthors, 2022: The High-Resolution Rapid Refresh (HRRR): An hourly updating convection-allowing forecast model. Part I: Motivation and system description. Wea. Forecasting, 37, 13711395, https://doi.org/10.1175/WAF-D-21-0151.1.

    • Search Google Scholar
    • Export Citation
  • Duda, J. D., and W. A. Gallus Jr., 2010: Spring and summer Midwestern severe weather reports in supercells compared to other morphologies. Wea. Forecasting, 25, 190206, https://doi.org/10.1175/2009WAF2222338.1.

    • Search Google Scholar
    • Export Citation
  • Eilts, M. D., and S. D. Smith, 1990: Efficient dealiasing of Doppler velocities using local environment constraints. J. Atmos. Oceanic Technol., 7, 118128, https://doi.org/10.1175/1520-0426(1990)007<0118:EDODVU>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Eure, K. C., P. D. Mykolajtchuk, Y. Zhang, D. J. Stensrud, F. Zhang, S. J. Greybush, and M. R. Kumjian, 2023: Simultaneous assimilation of planetary boundary layer observations from radar and all-sky satellite observations to improve forecasts of convection initiation. Mon. Wea. Rev, 151, 795813, https://doi.org/10.1175/MWR-D-22-0188.1.

    • Search Google Scholar
    • Export Citation
  • Fritsch, J. M., J. Hilliker, J. Ross, and R. L. Vislocky, 2000: Model consensus. Wea. Forecasting, 15, 571582, https://doi.org/10.1175/1520-0434(2000)015<0571:MC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Grim, J. A., J. O. Pinto, T. Blitz, K. Stone, and D. C. Dowell, 2022: Biases in the prediction of convective storm characteristics with a convection allowing ensemble. Wea. Forecasting, 37, 6583, https://doi.org/10.1175/WAF-D-21-0106.1.

    • Search Google Scholar
    • Export Citation
  • Han, Y., P. van Delst, Q. Liu, F. Weng, B. Yan, R. Treadon, and J. Derber, 2006: JCSDA Community Radiative Transfer Model (CRTM)—Version 1. NOAA Tech. Rep. NESDIS 122, 33 pp., https://repository.library.noaa.gov/view/noaa/1157.

    • Search Google Scholar
    • Export Citation
  • Hartung, D. C., J. A. Otkin, R. A. Petersen, D. D. Turner, and W. F. Feltz, 2011: Assimilation of surface-based boundary layer profiler observations during a cool-season weather event using an observing system simulation experiment. Part II: Forecast assessment. Mon. Wea. Rev., 139, 23272346, https://doi.org/10.1175/2011MWR3623.1.

    • Search Google Scholar
    • Export Citation
  • 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.

    • Search Google Scholar
    • Export Citation
  • Johns, R. H., and C. A. Doswell III, 1992: Severe local storm forecasting. Wea. Forecasting, 7, 588612, https://doi.org/10.1175/1520-0434(1992)007<0588:SLSF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Johnson, A., X. Wang, J. R. Carley, L. J. Wicker, and C. Karstens, 2015: A comparison of multiscale GSI-based EnKF and 3DVar data assimilation using radar and conventional observations for midlatitude convective-scale precipitation forecasts. Mon. Wea. Rev., 143, 30873108, https://doi.org/10.1175/MWR-D-14-00345.1.

    • Search Google Scholar
    • Export Citation
  • Johnson, A., X. Wang, and S. Degelia, 2016: Design and implementation of a GSI-based convection-allowing ensemble based data assimilation and forecast system for the PECAN field experiment. Part II: Overview and evaluation of real-time system. Wea. Forecasting, 32, 12271251, https://doi.org/10.1175/WAF-D-16-0201.1.

    • Search Google Scholar
    • Export Citation
  • Johnson, A., X. Wang, and T. Jones, 2022: Impacts of assimilating GOES-16 ABI channels 9 and 10 clear air and cloudy radiance observations with additive inflation and adaptive observation error in GSI-EnKF for a case of rapidly evolving severe supercells. J. Geophys. Res. Atmos., 127, e2021JD036157, https://doi.org/10.1029/2021JD036157.

    • Search Google Scholar
    • Export Citation
  • Jones, T. A., and Coauthors, 2020: Assimilation of GOES-16 radiances and retrievals into the Warn-on-Forecast System. Mon. Wea. Rev., 148, 18291859, https://doi.org/10.1175/MWR-D-19-0379.1.

    • Search Google Scholar
    • Export Citation
  • Kain, J. S., and Coauthors, 2008: Some practical considerations regarding horizontal resolution in the first generation of operational convection-allowing NWP. Wea. Forecasting, 23, 931952, https://doi.org/10.1175/WAF2007106.1.

    • Search Google Scholar
    • Export Citation
  • Kain, J. S., and Coauthors, 2013: A feasibility study for probabilistic convection initiation forecasts based on explicit numerical guidance. Bull. Amer. Meteor. Soc., 94, 12131225, https://doi.org/10.1175/BAMS-D-11-00264.1.

    • Search Google Scholar
    • Export Citation
  • Kleist, D. T., D. F. Parrish, J. C. Derber, R. Treadon, W.-S. Wu, and S. Lord, 2009: Introduction of the GSI into the NCEP global data assimilation system. Wea. Forecasting, 24, 16911705, https://doi.org/10.1175/2009WAF2222201.1.

    • Search Google Scholar
    • Export Citation
  • Liu, H., M. Hu, G. Ge, D. Stark, H. Shao, K. Newman, and J. Whitaker, 2018: Ensemble Kalman filter (EnKF) user’s guide version 1.3. Developmental Testbed Center Doc., 80 pp., https://dtcenter.org/community-code/ensemble-kalman-filter-system-enkf/documentation.

  • Lock, N. A., and A. L. Houston, 2014: Empirical examination of the factors regulating thunderstorm initiation. Mon. Wea. Rev., 142, 240258, https://doi.org/10.1175/MWR-D-13-00082.1.

    • Search Google Scholar
    • Export Citation
  • Melnikov, V. M., R. J. Doviak, D. S. Zrnić, and D. J. Stensrud, 2011: Mapping Bragg scatter with a polarimetric WSR-88D. J. Atmos. Oceanic Technol., 28, 12731285, https://doi.org/10.1175/JTECH-D-10-05048.1.

    • Search Google Scholar
    • Export Citation
  • Melnikov, V. M., R. J. Doviak, D. S. Zrnić, and D. J. Stensrud, 2013: Structures of Bragg scatter observed with the polarimetric WSR-88D. J. Atmos. Oceanic Technol., 30, 12531258, https://doi.org/10.1175/JTECH-D-12-00210.1.

    • Search Google Scholar
    • Export Citation
  • Minamide, M., and F. Zhang, 2017: Adaptive observation error inflation for assimilating all-sky satellite radiance. Mon. Wea. Rev., 145, 10631081, https://doi.org/10.1175/MWR-D-16-0257.1.

    • Search Google Scholar
    • Export Citation
  • Moller, A. R., C. A. Doswell III, M. P. Foster, and G. R. Woodall, 1994: The operational recognition of supercell thunderstorm environments and storm structures. Wea. Forecasting, 9, 327347, https://doi.org/10.1175/1520-0434(1994)009<0327:TOROST>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Nakanishi, M., and H. Niino, 2009: Development of an improved turbulence closure model for the atmospheric boundary layer. J. Meteor. Soc. Japan, 87, 895912, https://doi.org/10.2151/jmsj.87.895.

    • Search Google Scholar
    • Export Citation
  • National Research Council, 2009: Observing Weather and Climate from the Ground Up: A Nationwide Network of Networks. National Academies Press, 234 pp., https://doi.org/10.17226/12540.

  • NCEI, 2022: U.S. billion-dollar weather and climate disasters, 1980-present. NOAA/NCEI, accessed 15 August 2022, https://doi.org/10.25921/stkw-7w73.

  • Olson, J. B., J. S. Kenyon, W. A. Angevine, J. M. Brown, M. Pagowski, and K. Sušelj, 2019: A description of the MYNN-EDMF scheme and the coupling to other components in WRF-ARW. NOAA Tech. Memo. OAR GSD 61, 37 pp., https://doi.org/10.25923/n9wm-be49.

  • Otkin, J. A., D. C. Hartung, D. D. Turner, R. A. Petersen, W. F. Feltz, and E. Janzon, 2011: Assimilation of surface-based boundary layer profiler observations during a cool-season weather event using an observing system simulation experiment. Part I: Analysis impact. Mon. Wea. Rev., 139, 23092326, https://doi.org/10.1175/2011MWR3622.1.

    • Search Google Scholar
    • Export Citation
  • Rasmussen, E. N., and D. O. Blanchard, 1998: A baseline climatology of sounding-derived supercell and tornado forecast parameters. Wea. Forecasting, 13, 11481164, https://doi.org/10.1175/1520-0434(1998)013<1148:ABCOSD>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Ryzhkov, A., P. Zhang, H. Reeves, M. Kumjian, T. Tschallener, S. Trömel, and C. Simmer, 2016: Quasi-vertical profiles—A new way to look at polarimetric radar data. J. Atmos. Oceanic Technol., 33, 551562, https://doi.org/10.1175/JTECH-D-15-0020.1.

    • Search Google Scholar
    • Export Citation
  • Schmit, T. J., P. Griffith, M. M. Gunshor, J. M. Daniels, S. J. Goodman, and W. J. Lebair, 2017: A closer look at the ABI on the GOES-R series. Bull. Amer. Meteor. Soc., 98, 681698, https://doi.org/10.1175/BAMS-D-15-00230.1.

    • Search Google Scholar
    • Export Citation
  • Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp., https://doi.org/10.5065/D68S4MVH.

  • Soler, T., and D. W. Eisemann, 1994: Determination of look angles to geostationary communication satellites. J. Surv. Eng., 120, 115127, https://doi.org/10.1061/(ASCE)0733-9453(1994)120:3(115).

    • Search Google Scholar
    • Export Citation
  • Stelten, S., and W. A. Gallus Jr., 2017: Pristine nocturnal convective initiation: A climatology and preliminary examination of predictability. Wea. Forecasting, 32, 16131635, https://doi.org/10.1175/WAF-D-16-0222.1.

    • Search Google Scholar
    • Export Citation
  • 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, 50955115, https://doi.org/10.1175/2008MWR2387.1.

    • Search Google Scholar
    • Export Citation
  • Thompson, R. L., C. M. Mead, and R. Edwards, 2007: Effective storm-relative helicity and bulk shear in supercell thunderstorm environments. Wea. Forecasting, 22, 102115, https://doi.org/10.1175/WAF969.1.

    • Search Google Scholar
    • Export Citation
  • Tochimoto, E., and H. Niino, 2016: Structural and environmental characteristics of extratropical cyclones that cause tornado outbreaks in the warm sector: A composite study. Mon. Wea. Rev., 144, 945969, https://doi.org/10.1175/MWR-D-15-0015.1.

    • Search Google Scholar
    • Export Citation
  • Wang, X., and T. Lei, 2014: GSI-based four-dimensional ensemble–variational (4DEnsVar) data assimilation: Formulation and single-resolution experiments with real data for NCEP global forecast system. Mon. Wea. Rev., 142, 33033325, https://doi.org/10.1175/MWR-D-13-00303.1.

    • Search Google Scholar
    • Export Citation
  • Wang, X., D. Parrish, D. Kleist, and J. Whitaker, 2013: GSI 3DVar-based ensemble-variational hybrid data assimilation for NCEP global forecast system: Single resolution experiments. Mon. Wea. Rev., 141, 40984117, https://doi.org/10.1175/MWR-D-12-00141.1.

    • Search Google Scholar
    • Export Citation
  • Weisman, M. L., C. Davis, W. Wang, K. W. Manning, and J. B. Klemp, 2008: Experiences with 0–36-h explicit convective forecasts with the WRF-ARW Model. Wea. Forecasting, 23, 407437, https://doi.org/10.1175/2007WAF2007005.1.

    • Search Google Scholar
    • Export Citation
  • Whitaker, J. S., and T. M. Hamill, 2002: Ensemble data assimilation without perturbed observations. Mon. Wea. Rev., 130, 19131924, https://doi.org/10.1175/1520-0493(2002)130<1913:EDAWPO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Whitaker, J. S., and T. M. Hamill, 2012: Evaluating methods to account for system errors in ensemble data assimilation. Mon. Wea. Rev., 140, 30783089, https://doi.org/10.1175/MWR-D-11-00276.1.

    • Search Google Scholar
    • Export Citation
  • Wu, W.-S., R. J. Pursuer, and D. F. Parrish, 2002: Three-dimensional variational analysis with spatially inhomogeneous covariances. Mon. Wea. Rev., 130, 29052916, https://doi.org/10.1175/1520-0493(2002)130<2905:TDVAWS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Zhang, F., C. Snyder, and J. Sun, 2004: Impacts of initial estimate and observation availability on convective-scale data assimilation with an ensemble Kalman filter. Mon. Wea. Rev., 132, 12381253, https://doi.org/10.1175/1520-0493(2004)132<1238:IOIEAO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Zhang, F., Y. Weng, J. A. Sippel, Z. Meng, and C. H. Bishop, 2009: Cloud-resolving hurricane initialization and prediction through assimilation of Doppler radar observations with an ensemble Kalman filter. Mon. Wea. Rev., 137, 21052125, https://doi.org/10.1175/2009MWR2645.1.

    • Search Google Scholar
    • Export Citation
  • Zhang, Y., D. J. Stensrud, and F. Zhang, 2015: Practical predictability of the 20 May 2013 tornadic thunderstorm event in Oklahoma: Sensitivity to synoptic timing and topographical influence. Mon. Wea. Rev., 143, 29732997, https://doi.org/10.1175/MWR-D-14-00394.1.

    • Search Google Scholar
    • Export Citation
  • Zhang, Y., E. E. Clothiaux, and D. J. Stensrud, 2022: Correlation structures between satellite all-sky infrared brightness temperatures and the atmospheric state at storm scales. Adv. Atmos. Sci., 39, 714732, https://doi.org/10.1007/s00376-021-0352-3.

    • Search Google Scholar
    • Export Citation