Convective-scale data assimilation of high-resolution wind and thermodynamic observations during PECAN and VORTEX2 James Marquis, Glen Romine, Josh Wurman, Tammy Weckwerth, Jim Wilson, Kristen Rasmussen Introduction: Two research projects are underway to explore improvements in numerical representation of environments supportive of nocturnal and daytime convection by assimilating a variety of unique observations to mesoscale and storm-scale ensembles using the WRF model and Data Assimilation Research Testbed (DART) software with the Ensemble Kalman Filter (EnKF) technique.

Goals: Our first project (left) assimilates stationary and mobile radar, aircraft, sonde, and profiler data collected during PECAN to Goals: Our second project (right) assimilates different combinations of routine radar, sonde, and commercial aircraft nested domains with 3-km and sub-1-km grid spacing to accurately reproduce detailed environments of nocturnal convection data, and atmospheric motion vectors (AMVs) derived from super-rapid-scan GOES operations to determine the relative initiation (NCI). Resulting fine-scale EnKF analyses will be used to evaluate environments and triggers dictating specifc locations impact of high-resolution satellite observations on the mesocale shear field surrounding tornadic supercells. Principle of NCI, including understanding the roles of nocturnal low-level jets (LLJ), surface boundaries, and influences from previous data for tests was collected during the second VORTEX project, and will advise upon operational utility of AMVs from the convection. Analyses also will be used as intitial conditions for forecast sensitivity studies to understand impact of novel upcoming GOES-R. Fine-scale simulations (<1-km grid spacing) of supercells that employ initial conditions produced high-resolution observations on the predictability of NCI. with and without AMVs will indicate storm-scale sensitivities of potentially more detailed environments.

Method: Initial EnKF work has begun on the 23-24 June 2015 Early Results: Preliminary EnKF analyses assimilate MADIS surface, radiosonde, and commercial Method: Production of a preliminary set of AMVs, derived PECAN case. This was a combined LLJ-NCI mission in SE from GOES-11 imagery on 5 June 2009, is calculated by AMVs from GOES11 (15-min) 5 June 2009 Commercial aircraft winds (15-min window) aircraft in situ observations (small black dots) hourly to a 3-km mesoscale ensemble. Select 49 Nebraska, sampling the LLJ along its forecasted axis. NCI occured tracking visible, infrared, and water vapor objects at 15-min 1700 UTC posterior analyses are shown below, next to corresponding WSR-88D and other observations. Altitude 30 m/s above 350 mb just NE and W of Omaha. intervals (3-min for visible). Super-rapid-scan operations 47 750 - 350 mb Upcoming analyses will assimilate 88D (black circles) reflectivity, clear-air velocity data (typical below 750 mb coverage extends ~50 km from each radar, large black circles), and PECAN fixed PISA sondes were conducted on 5 June 2009 starting at approximately 24 June 2015 Lines of NCI 45 NEXRAD 1900 UTC (scans at variable time increments, as small as deployment map (~ 0440 UTC) (black squares) leading up to the IOP. Sioux City 1.5-min); although, sufficient data were available to produce 43 2-m T (K) KLNX 23 June 2015, 1400 UTC (NCI -15 hr) With only MADIS observations assimilated, the mesoscale analyses, start- AMVs prior to this time in routine operation mode. A sample Isolated NCI Mobile ing the morning prior to NCI, develops mesocale/synoptic-scale features (~ 0510 UTC) Radar observed on 23 June, including a northward-moving warm front in central of the AMVs is compared to contemporaneous upper air 41 KS. It is believed this warm front plays a pivotal role in forcing the NCI. Day- CSU KOAX time CI occurs along KS-NE border. 1400 UTC, 23 June commercial aircraft (ACARS) data, WSR-88D locations, and 39 Omaha Fixed 88D re . 1400 UTC, 23 June routine sondes (only at 0000 and 1200 UTC) in the figure to DOW8 Profiling SR2 System DOW7 37 50 km the right. Grand Lincoln Island NSSL FP4 35 SR1 Mobile KUEX Profiling Early Results: There are significantly higher density MIPSNOXP System DOW6 Morning CI Latitude (degrees) mesoscale wind observations resulting from SRSO 33 SPARC MGAUS GOES-derived AMV calculations than from routine upper air UWKA St. Joseph 31 UWKA 49 TWOLF flight observations (e.g., ACARS, 12-hourly sondes, and 1800 UTC CLAMPS path spatially-limited 88D-derived VAD profiles). Owing to Latitude (degrees) 47 23 June 2015, 2000 UTC (NCI - 9 hr) 2-m T (K) The morning convection develops into a disorganized MCS that moves east- upper-level cloud cover, many AMVs over the high plains Availability of routine and PECAN observations for assimilation to ward into Missouri. This convection produces a cold pool that moves north- ward into the target area for NCI that night. It is not obvious that the surface are limited to high altitudes (above 350 mb); thus, mid- and 45 the mesoscale and storm-scale grids, starting 12 UTC on 23 June. gust front triggers NCI, but the role that the cold pool plays in NCI is unknown low-level observations are generally distant from the and a focus of this investigation. 43 VORTEX2-observed storm (located in purple box on the 88D re . 2000 UTC, 23 June KOAX re . 2323 UTC, 23 June Future locations of 41 NCI (5.5 hrs later) right). However, assimilation of these AMV observations 0.05 g/kg KOAX rain mixing over much of the plains may radically increase details of the ratio boundary layer flow in the TX-OK-KS region, and the 39 upper-level flow in the CO-NE-WY-SD region, which could 37

10-m Latitude (degrees) N’ward moving horizontal KOAX gust front wind improve details of the mesocale and synoptic environment moreso than by only using traditionally-assimilated routine 35

2-km observations. 33 horizontal wind 2-m T (K) No neighboring convection exists by the time of NCI. The warm front contin- 24 June 2015, 0200 UTC (NCI -3 hr) 31 . . ues northward. Southerly flow at z = 2 km is enhanced north of the front, 49 .... though less so south of it. Wind veers in the LLJ south of the front accurately, 1900 UTC Availability of upper air observations during the 5-6 June 2009, Goshen, WY supercell event Assimilated but analyzed winds aloft are slightly weaker than observations (e.g., at FP2). ACARS and Assimilation of PECAN and radar are expected to improve the analyses. GOES-11 scans 47 METAR obs routine operations super rapid scan operations 0130 UTC, 24 June; Winds from FP2 88D re . 0200 UTC, 24 June 45 GOES-11 AMVs WSR- 50-km 43 88D range ring ACARS winds x Latitude (degrees) 41

PECAN (km)Height mobile FP2 88D volumes (VADs) profiler 1600 1700 1800 1900 2000 2100 2200 2300 0000 0100 39 WRF Specs: FP2

Time (UTC) 3-km grid spacing, 733x433x50 grid points, dt = 10s, x 37 50 ensemble members, 2-m T (K) Convection initiates in broken bands north of the warm front within the broad o PECAN 24 June 2015, 0600 UTC (NCI + 1hr) Initial conditions from GFS 12 UTC 23 June 2015 (0.25 -res) mobile convergence zone leading the LLJ and over-riding flow at approximately 0500 Experiments: Using a similar cycled mesoscale (3-km grid 35 radar 50-km range UTC on 24 June. Proximity soundings suggest unstable air originates from spacing) data assimilation system as the PECAN-focused Boundary conditions from GFS analyses, ring elevated sources (above ~700 mb). It is not yet clear why the convection 33 Parameterizations: No convective param, MYJ PBL, forms in three parallel bands in Iowa and single bands in Nebraska, with gaps project (left), we will evaluate changes in mesoscale shear in between. This organization of NCI is a primary focus of this research. Future < 1-km and its effect on storm morphology and evolution when 31 Thompson MP, RRTMG SW & LW radiation, NOAH land-sfc storm-scale 88D re . 0600 UTC, 24 June CSU sonde, 0600 UTC 49 ensemble domain AMVs are assimilated at 15-min intervals, when they are not 2000 UTC NWS sonde location assimilated, and when they are assimilated at coarser WSR-88D location DART specs: Latitude (degrees) 47 PECAN-targeted NCI temporal/spatial resolution commensurate with routine and Gaspari & Cohn (1999) localization, sonde Latitude (degrees) past satellite observations. Results will inform upon the 45 Observations smoothed/thinned prior to assimilation, utility of routinely assimilating new GOES-R AMV data sets 43 Adaptive Inflation for the use of convective-scale operational NWP. Longitude (degrees) 41

Potentially valuable fine-scale mobile radar observations to assimilate. Future Work: Perform similar mesoscale experiments for Future Work: Nested sub-1km-grid centered on IOP and NCI (orange box DOW 8 refl & vel 39 DOW 6 vel, 0500 UTC more GOES and upcoming GOES-R SRSO-derived AMV in panels to the right). Assimilate routine in situ observations, nearby 0420 UTC DOW 7 vel, 0650 UTC Small scale waves 37 z = 1 km z = 1 kmdata sets to understand common impact of different WSR-88D observations, mobile PISA sondes and profilers, and mobile observed cloud layers on storm development. Also, to 0.4km 35 radars at 10-min intervals from 0200 UTC throughout the NCI event. Radar fine-lines quantify expectations for coverage and utility of SRSO Evaluate origins of convection with parcel trajectory analysis and 4-D AMVs over the U.S. Plains for convective events. 33 0520 UTC mapping of thermodynamic and flow fields. Explore the value of Perform storm-scale (sub-1-km) resolution simulations 0.9 km 31 assimilating SPOL refractivity, DIAL, and AERI data. Begin similar work on with AMV and non-AMV analyses as initial conditions to test -111 -109 -107 -105 -103 -101 -99 -97 -95 -93 -91 -89 -109 -107 -105 -103 -101 -99 -97 -95 -93 -91 -89 Longitude (degrees) 3-4 July 2015 case. Discrete shear layers impact of improved mesoscale flow (shear) fields.

Acknowledgements: This PECAN analysis project was funded by the National Science Foundation; AGS-1541624. Acknowledgements: This project is funded by NASA grant NNX15AR58G. Simulations are performed on the Simulations are performed on the Yellowstone supercomputer, with CISL support. Thank you to all dedicated participants of NASA advanced supercomputing (NAS) facility, Pleiades. Special thank you to Bob Rabin (NOAA/NSSL and the PECAN project for collecting valuable scientific data at all times of the night. UW-Madison/CIMSS) for guidance with and building the SRSO AMV data set for this case.