Constraining Numerical Forecasts of Deep Convective Initiation through Assimilation of Surface Observations General Exam Luke Madaus 5/19/2015

Outline Background Proposed Experiments ▫Idealized Simulations ▫Observing System Simulation Experiments (OSSEs) ▫Applied Case Study Discussion

Background Forecasting deep convective initiation remains enormously challenging Despite modelling advances, forecasting storm- scale initiation still has limited skill Kain et al. 2013

Background Review of deep convective initiation (parcel theory) Limitations  Entrainment ignored, no dynamic pressure perts Surface can still play a major role! From Stensrud (2007) and Mapes (1997)

The Surface and CI Temperature  Warmer regions have increased potential for positive buoyancy and reduced CINH Moisture  Moister rising parcels reach their LCL at a lower altitude where latent heating from condensation increases buoyancy/reduces CINH Wind  Low-level convergence promotes upward vertical motion, sustaining rising parcels against CINH Pressure  Low pressure near the surface symptomatic of rising motion/convergence

The Surface and CI Observational estimates ▫1-2 m/s wind variability over 2-5 km can promote CI (Weckwerth et al. 1999; Arnott et al. 2006) ▫Moisture variations of g/kg over 5km typical in pre-CI environment (Crook 1996; Fabry 2006; Martin and Xue 2006) ▫Which is more important—temperature or moisture? Predictability limitations ▫Estimates of CI predictability on order of 1-2 hours (Weckwerth 2000; Zhang et al. 2003; Hohenegger and Schar 2007) ▫Are surface observations enough?

Proposed Experiments Storm Environment Simulations What sorts of storm-scale structures are apparent in surface fields prior to CI? OSSE Experiments How do we observe at the surface and assimilate in a way that most refines probabilistic forecasts of CI? Case Study with Novel, Existing Observations To what extent can novel, extant, dense surface observing networks improve probabilistic forecasts of CI? Idealized Real- World

What sorts of storm-scale structures are apparent in surface fields prior to CI, and are they observable given current observation limitations?

Idealized Simulations Dense surface observations already shown to improve synoptic and large mesoscale features (e.g. Madaus et al. 2014) Examine CI at its most “unpredictable,” absent larger-scale forcing ▫Isolated, “popcorn” convection ▫This has been poorly studied Clearer evaluation of lower-bound of practical forecast improvement

Idealized Simulations 30 environments chosen from Summer 2014

Idealized Simulations Use Cloud Model 1 (CM1), revision 17 (Bryan and Fritsch 2002) 104x104 km domain, doubly periodic, diurnal radiation is only forcing Homogeneous land surface properties tailored to sounding location Two types of simulations ▫200m resolution “truth” simulation ▫ member ensemble at 1km resolution  Ensemble diversity from random initial temperature perturbations only

Idealized Simulations Single Ensemble MemberComposite Storm (n=75)

Composited Surface Anomalies

Compositing summary across environments Maximum +/- anomaly magnitude along storm track Cloud Shadowing Slow increase in convergence Higher LCLs Warm anomalies “overshadowed” Weak PSFC anomalies

Future Timeline Idealized simulations ▫Complete ensembles and 200m simulations for all cases ▫Summarize the composited anomalies and examine their variability with respect to initial condition environment Optional: ▫Evaluate scale dependence of anomalies at different resolutions ▫For select cases, compare the scale and magnitude of the anomalies with respect to different microphysics schemes (Paul Markowski, personal communication)

How do we observe at the surface and assimilate in a way that most refines probabilistic forecasts of CI?

OSSE Experiments Idealized model experiments designed with OSSE application in mind Would like to use DART software for data assimilation; need to create interface with CM1 (summer project) Can vary a number of parameters to see how they affect subsequent ensemble forecasts of CI: ▫Observation density ▫Cycling frequency ▫Observation Error ▫Localization

OSSE Experiments Early estimates based on anomaly magnitude Can refine estimates knowing avg. correlation length scale

Future Timeline OSSE experiments ▫Work with the DART team to create a CM1 interface ▫Evaluate ensemble’s ability to resolve structures based on hypotheses from previous section ▫Assess how subsequent probabilistic forecasts of CI change post-assimilation Optional: ▫Establish frequency and spatial scale of observations that most improves CI forecasts ▫Examine the effect of reduced observation error on forecasts ▫Estimate localization radii that optimize assimilation impact

To what extent can novel, extant, dense surface observing networks improve probabilistic forecasts of CI?

Applied case study High-density observations Smartphones Weather Underground

Applied case study Case Study ▫July 27-29, 2014 ▫NE CONUS ▫Repeated rounds of convective initiation Ensemble Design ▫Based on HRRR model ▫HRRR boundary conditions ▫Use SKEBS for ensemble diversity

Applied case study How good are these new observations?

Future Timeline Applied Case Study ▫Work with collaborators (Cliff Mass, Conor McNicholas) to quality control smartphone and Weather Underground Observations ▫Complete assimilation cycling and produce ensemble forecasts for CI periods ▫Use storm-scale verification to assess how much CI forecasts may have been improved with additional observations Optional: ▫Expand with OSSE-type experiments to evaluate even greater observation density ▫Attempt to partition the observation impact into larger mesoscale and storm-scale adjustments (challenge; may not be possible)

Summary Surface weather quantities are anticipated to have strong connections to convective initiation Coherent surface anomalies are observed preceding convective initiation in idealized simulations A hypothetical observing network can be constructed to attempt to capture this pre-CI variability Real test---does resolving the fine-scale surface structure indeed improve ensemble forecasts of CI? How well can existing, underutilized surface observing networks replicate the idealized results?

Thank You!

References Kain, J. S., and Coauthors, 2013: A feasibility study for probabilistic convection initiation forecasts based on explicit numerical guidance. Bull. Amer. Meteor. Soc., 94, 1213–1225, doi: /BAMS-D Stensrud, D. J., 2007: Parameterization Schemes: Keys to Understanding Numerical Weather Prediction Models. 1st ed., Cambridge University Press, New York. Mapes, B. E., 1997: Equilibrium vs. activation control of large-scale variations of tropical deep convection. The Physics and Parameterization of Moist Atmospheric Convection, R. K. Smith, Ed., Kluwer Academic Publishers, Norwell, Massachusetts. Weckwerth, T.M., T.W. Horst, and J.W.Wilson, 1999: An observational study of the evolution of horizontal convective rolls. Mon. Wea. Rev., 127, 2160–2179, doi: / (1999)127h2160:AOSOTEi2.0. CO;2. Arnott, N. R., Y. P. Richardson, J. M. Wurman, and E. M. Rasmussen, 2006: Relationship between a weakening cold front, misocyclones, and cloud development on 10 june 2002 during ihop. Mon. Wea. Rev., 134, 311–335, doi: /MWR Crook, N. A., 1996: Sensitivity of moist convection forced by boundary layer processes to low-level thermodynamic fields. Mon. Wea. Rev., 124, 1767–1785, doi: / (1996)124h1767:SOMCFBi2.0.CO;2. Fabry, F., 2006: The spatial variability of moisture in the boundary layer and its effect on convection initiation: Project-long characterization. Mon. Wea. Rev., 134, 79–91, doi: /MWR Martin, W. J., and M. Xue, 2006: Sensitivity analysis of convection of the 24 may 2002 ihop case using very large ensembles. Mon. Wea. Rev., 134, 192–207, doi: /MWR Weckwerth, T. M., 2000: The effect of small-scale moisture variability on thunderstorm initiation. Mon. Wea. Rev., 128, 4017–4030, doi: / (2000)129h4017:TEOSSMi2.0.CO;2. Zhang, F., C. Snyder, and R. Rotunno, 2003: Effects of moist convection on mesoscale predictability. J. Atmos. Sci., 60, 1173–1185, doi: / (2003)060h1173:EOMCOMi2.0.CO;2. Hohenegger, C., and C. Sch¨ar, 2007: Atmospheric predictability at synoptic versus cloud-resolving scales. Bull. Amer. Meteor. Soc., 88, 1783–1793, doi: /BAMS Bryan, G. H., and J. M. Fritsch, 2002: A benchmark simulation for moist nonhydrostatic numerical models. Mon. Wea. Rev., 130, 2917–2928, doi: / (2002)130h2917:ABSFMNi2.0.CO;2. Madaus, L. E., G. J. Hakim, and C. F. Mass, 2014: Utility of dense pressure observations for improving mesoscale analyses and forecasts. Mon. Wea. Rev., 142, 2398–2413, doi: /MWR-D

Idealized Simulations Environment characteristics

Idealized Simulations Domain- averaged ensemble variance with time

Idealized Simulations Surface temperature (from PBL scheme) First model level temperature (40m)

Idealized Simulations Before clouds Clouds developingPrecipitation present Evolution of correlation pattern/length scale

Applied case study Weather Underground Temperature