7B.4 CONVECTION AND APPALACHIAN COLD-AIR DAMMING 1. INTRODUCTION Adm K. Bker * nd Gry M. Lckmnn North Crolin Stte University, Rleigh, North Crolin Cold-ir dmming (CAD) is common occurrence long the estern slopes of the Applchin Mountins in the southestern US. Although there re multiple clssifictions of CAD (e.g., Biley et l. 2003), this typiclly occurs when reltively cold ir is driven southwrd initilly in the form of ckdoor cold front (e.g., Crr 1951) nd undergoes geostrophic djustment with the incresing effect of the Coriolis force. Consequently, the flow of cold ir is redirected towrd the mountins nd ecomes locked y the mountin rrier to form shllow cold dome (e.g., Bell nd Bosrt 1988). As the cross-rrier force lnce djusts towrds geostrophy, lowlevel mountin-prllel jet forms nd serves to help mintin the cold dome, driving the cold ir frther southwrd (e.g., Fores et l. 1987). This cold dome cn extend s fr southwrd s centrl Georgi (e.g., Fig. 1) nd persist for severl dys efore erosion occurs. While cold seson CAD rings mny chllenges to winter wether forecsting (e.g., Keeter et l. 1995), there is dditionl concern mong opertionl forecsters with the role of CAD (including the wrm seson) in convective development. Although the cold dome stilizes the lower-troposphere, the periphery of the CAD dome, or wedge front, hs een oserved to trigger convective development (e.g., Fig. 1). When mient moisture nd instility re sufficient for convection to develop ner the wedge front, the sher environment of the cold dome hs een hypothesized to enhnce the possiility of severe convection. The ojectives of this reserch re to clrify nd quntify the CAD cold dome influence on convection nd the convective environment. Specificlly, our gol is to isolte conditions in which the presence of CAD sufficiently lters convective storm structure nd intensity. We hypothesize tht the CAD cold dome mkes n pprecile difference with convective intensity nd ehvior. * Corresponding uthor ddress: Adm K. Bker, North Crolin Stte University, Dept. of Mrine, Erth nd Atmospheric Sciences, Rleigh, NC 27695. E-mil: kker@ncsu.edu Figure 1. () Mnul surfce temperture ( F) nd frontl nlysis t 2300Z 20 Mrch 2003 nd () rdr composite reflectivity t 0000Z 21 Mrch 2003 (ottom). Dshed white lines denote pproximte loction of temperture grdient long CAD dome periphery nd white ovl outlines developing convection. 2. DATA AND METHODS In order to successfully isolte the effects of CAD nd ppropritely test the hypothesis, the following steps were tken: (1) identify cses of ctive convection long the CAD dome periphery from comprehensive dtse of CAD event dys, (2) isolte the most representtive cses, nd (3) use rel cse dt to initilize model simultions for testing.
2.1 Cse identifiction The process of identifying cses of ctive convection long the wedge front involved detiled criteri in multiple fields commonly nlyzed in opertionl forecsting with the intention of mintining the pplicility of the reserch to the forecsting environment. An ojective definition of wedge-front convection (WFC) ws used to ssemle dtset of ctive WFC cses from specific sptil, temporl, nd intensity requirements from temperture, rdr, nd renlysis dt ssocited with pst CAD events. The CAD events used to determine WFC cses were compiled from (i) list generted from the previous reserch of Tom Green, Wendy Sellers, nd Chris Biley, (ii) list generted y CAD lgorithm from the NWS WFO in Columi, SC (CAE), nd (iii) from dditionl cses suggested y NWS forecsters. Together the lists provided 241 totl dys of CAD tht could e nlyzed for WFC with resources of consistent dt, spnning from 01 April 2001 to 15 June 2007 (pproximtely 6 yers). Once dtset of ctive WFC cses ws ssemled, the cses were nlyzed for representtiveness in (i) cold dome strength, (ii) severity of convection, nd (iii) mount of mient moisture nd instility. Dt from cse determined to e est representtive of convection ssocited with reltively strong cold dome ws used to initilize numericl simultions for testing. 2.2 Numericl simultions When properly conducted, numericl simultion comprisons cn provide n efficient method for isolting key physicl processes in vrious tmospheric phenomen nd cn serve s roust tests of scientific hypotheses. Successful isoltion of the CAD cold dome s effect on the lowertropospheric convective environment ws chieved through modeling experiment in which version 2.2 of the Wether Reserch nd Forecsting (WRF) model ws initilized with North Americn Regionl Renlysis (NARR) gridded dt from representtive rel cse. A simultion of CAD cold dome serves s the control run (CAD run) for testing ginst, nd second simultion with the sme initil conditions s the CAD run, ut with the Applchin mountins removed serves s the experimentl run (NoCAD run). Removl of the mountins, while keeping ll other tmospheric fetures nerly identicl to the CAD run, resulted in similr pttern evolution ut without the presence of CAD. An nlysis of the differences in the lower-tropospheric environment etween the CAD nd NoCAD runs ws conducted to isolte the effect of the ctul cold dome presence in the model. 3. CASE REPRESENTATIVENESS Results from the identifiction process indicte tht there were 17 CAD events identified with ctive wedge-front convection. Of these CAD events, there were 24 ctive WFC dys, which is pproximtely 10% of the totl CAD dys. After the nlysis of CAD cold dome strength, 6 ctive WFC dys were found to e ssocited with reltively wek CAD cold domes, 9 with moderte domes, nd 9 with strong domes. Archived storm reports from the Storm Prediction Center were nlyzed to see if ny mtched the time nd loction with identified wedge-front convection on rdr. Upon completion of the nlysis, 6 out of 17 WFC events included some severe convection (pproximtely one third of the events). Of the 6 severe events, 4 were ssocited with reltively strong CAD domes. Bsed on surfced-sed CAPE vlues, the WFC occurrences were seprted into ins of reltively high or low CAPE vlues. A cse occurring on 20 Mrch, 2003 ws overll the est exmple of strong CAD dome convection (Fig. 1), nd mkes it pproprite for initilizing simultions. 4. MODEL EXPERIMENTS 4.1 Initil simultions Simultions were initilly performed to nlyze the effect of the CAD dome on the lower troposphere in reltively lrge-scle environment. Horizontl gridspcing ws kept t 12 km nd the Kin- Fritsch convective prmeteriztion scheme ws used to ccount for su-gridscle convective processes. NARR gridded dt from the rel cse of 20 Mrch, 2003 were used to initilize the CAD run, since it ws representtive of convective development ssocited with wedge front long reltively strong nd definite CAD cold dome. The simultions were run for 72 forecst hours from efore CAD strted to occur (0600Z 18 Mrch, 2003) until fter convection ws ctive long the wedge front (0600Z 21 Mrch, 2003).
4.1.1 Control run (unmodified terrin) The control CAD simultion successfully represented convection long the wedge front (Fig. 2). Much of the totl convective precipittion occurred ner the loction of the reltively strong temperture grdient long the cold dome periphery. The development of mountin-prllel jet led to enhnced frontogenesis nd convergence long nd ner the cold dome periphery shortly efore significnt convective precipittion mounts were present (not shown). The cold dome ws nlyzed to e in region of limited moisture nd instility (with higher mounts ner the periphery reltive to the dome interior) nd verticl wind sher nd 0-1-km storm reltive helicity (SRH) vlues cple of producing strong to severe convection (not shown). 4.1.2 Experimentl run (flttened terrin) When compred to the CAD run, the NoCAD run hd significnt convection closer to the Atlntic costline (Fig. 2), suggesting tht CAD influenced the preferred region for convective development to e locted more inlnd. There ws incresed moisture nd instility nd decresed 0-1-km SRH vlues in the NoCAD run cross the sme geogrphicl region where the cold dome ws nlyzed in the CAD run (not shown). Difference plots of 0-1-km SRH etween the two runs indicte ~600 m 2 /s 2 higher vlues in the CAD dome region (not shown). While these simultions suggest tht the cold dome lters the lower-tropospheric sher nd thermodynmic environment nd tht the region ner the wedge front serves s lifting mechnism cple of triggering convective development, the specific influence of the wedge front on convective chrcter nd intensity cnnot e determined from the corse 12-km grid used to this point. Therefore, nested domins with finer sptil nd temporl resolution were utilized in order to explicitly resolve convective development nd ehvior. This will llow determintion of the extent to which convective storms were le to drw on the helicity-rich ir found in the vicinity of the wedge front. Also, comprison of convective cells in these high-resolution simultions will revel whether storms were more likely to exhiit rottion when forming in the vicinity of the cold dome. Figure 2. 2-m potentil temperture (contours, intervl 2 K with cooler colors representing lower vlues) nd 10- hr convective precipittion mounts (fill, intervl 10 mm strting t 10 mm) ending t 1200Z 20 Mrch 2003 (forecst hour 57) for () the CAD run nd () NoCAD run. South Crolin is outlined in red to provide visul reference frme for the loction of convective precipittion mounts. 4.2 Convective-scle simultions Additionl simultions were performed t 1.3-km horizontl gridspcing nd 5-minute output intervl to llow for nlysis t the convective scle nd cpture the ehvior nd evolution of convective storm development. 4.2.1 Control run (unmodified terrin) The control run (herefter referred to s CAD-CS run for the convective scle) used the initil 12- km domin s the outermost domin nd included two nested domins t 4-km nd 1.3-km to mintin 3:1 gridspcing rtio etween domins. The innermost domin t 1.3-km gridspcing ws
centered over the region of significnt convective precipittion nlyzed in the initil CAD run, therefore this domin ws used for the convective scle nlysis. Composite reflectivity vlues indicted significnt region of convection cross centrl Georgi nd South Crolin (Fig. 3). Convective development initilly ecme pprent long the wrm side of the wedge front t 0505Z 20 Mrch, 2003, while susequent convective development occurred long region of enhnced frontogenesis nd convergence locted long the estern extent of the mountin-prllel jet s it developed within the cold dome nd pushed frther southwrd. The convection hd discrete cellulr structure nd some cells split into rightnd left-movers with the right-movers hving incresed longevity ehvior common in sher environments with incresed low-level positive SRH vlues. After severl hours of development the convective region formed into squll-line nd propgted estwrd (not shown). 4.2.2 Experimentl run (flttened terrin) The experimentl run (herefter referred to s NoCAD-CS run for the convective scle) employed the sme domin gridspcing nd initiliztions s the CAD-CS run, ut hd the mountins removed to simulte the sme tmosphere without the effect of CAD (nlogous to the initil NoCAD run). The nested domins were centered more over the South Crolin costline to llow for nlysis in the region of significnt convective precipittion nlyzed in the initil NoCAD run. A significnt region of convective development ws nlyzed in the composite reflectivity where initil development occurred out n hour fter wht ws nlyzed in the CAD-CS run nd in loction over 200 km frther to the northest (in centrl South Crolin, Fig. 3). Without the presence of the cold dome, the convection developed initilly in more liner structure (less discrete) nd did not pper to involve cell splits. The convective region lso formed into squllline (with more rpid trnsition thn in the CAD- CS run) which propgted estwrd. Figure 3. 2-m potentil temperture (contours, intervl 2 K with cooler colors representing lower vlues) nd composite reflectivity (dbz) cross centrl GA nd SC from CAD-CS run t 0640Z 20 Mrch, 2003 () nd cross centrl SC from NoCAD-CS run t 0755Z 20 Mrch, 2003 (). White rectngles outline res shown in Fig. 3 nd 3 respectively. 4.2.3 Averged model hodogrphs In order to nlyze the sher environment in which the convective development occurred, verged hodogrphs were clculted cross vrious loctions nd times in the simulted tmosphere. Figure 4 shows the verge hodogrphs long the region where convection occurred t time immeditely prior to initil development in oth convective-scle runs. While the NoCAD-CS run did hve some positive low-level SRH present, the CAD-CS run hd nerly twice s much low-level SRH present, suggesting more environmentl stremwise vorticity ville for tilting into the verticl y ny developing convective updrfts.
60 50 40 30 20 10 0-40 -20 0 20 40 60 80 100 120 14-10 -20 Figure 4. Averged hodogrphs (kts) for the CAD-CS run (grey) nd NoCAD-CS run (lck). The hodogrphs extend from the closest plotted pressure level to the surfce (lower left endpoints) to 125 m (right endpoints). 4.2.4 Updrft rottion A cross-sectionl nlysis of right-moving cell following split in the CAD-CS run showed the presence of counterrotting verticl vortex pirs, with the cyclonic vortex co-locted with the storm s updrft (not shown) lso suggesting the influence of stremwise vorticity on the developing convection. The product of verticl velocity nd verticl vorticity within the convection ws nlyzed cross multiple cells during development in oth convective-scle runs nd indicted rotting updrfts (Fig. 5). Frequencies of significnt updrft rottion cross grid points in similr geogrphicl res for oth convective-scle simultions re currently eing clculted to etter quntify the convective ehvior nd intensity etween the runs. Preliminry results suggest tht the significnt convection in the NoCAD-CS run reches updrft rottions of moderte intensity more frequently thn in the CAD-CS run. Verticl velocities within the convection overll pper to e stronger in the NoCAD-CS s well. The incresed updrft strength ppers to e ttriutle to the higher moisture nd instility vlues in the NoCAD environment. Figure 5. Convective cells outlined y white rectngles in () Figure 3 nd () Figure 3. Updrft rottion (fill, intervl 200 x 10-4 m/s 2 t vlues greter thn or equl to 600 x 10-4 m/s 2 ) t the 700 m level nd composite reflectivity (contours, intervl 5 dbz t vlues greter thn or equl to 40 dbz). 5. CONCLUSIONS After identifying cses of ctive WFC nd conducting n nlysis of representtiveness, the 20 Mrch, 2003 cse ws found to e representtive of convection ssocited with reltively strong nd well defined CAD cold dome. Initil CAD nd NoCAD WRF simultions initilized with NARR dt from this cse suggested tht the CAD cold dome primrily stilized the environment (with higher moisture nd instility mounts long the periphery reltive to the dome interior) nd incresed lowlevel SRH. The development of mountinprllel jet led to n re of enhnced frontogenesis nd convergence ner the wedge front nd serves s lifting mechnism nd preferred region for convective development.
Additionl simultions on the convective scle indicted tht convection developed ner the wedge front in region of incresed low-level SRH, wheres the storms without the influence of CAD cold dome developed in region of incresed moisture nd instility. Although oth simulted regions of convection eventully evolved into squll lines, the pre-squll-line period of development ner the wedge front included discrete cell structures nd cell splits with incresed longevity, while more liner reflectivity structure with less discreteness occurred in the sence of the wedge front. Rotting updrfts of moderte intensity were nlyzed in the convection for oth convective-scle runs, indicting the tilting of environmentl stremwise vorticity into the verticl. Quntifictions of convective chrcter nd intensity re currently eing conducted in the reserch to properly compre the convective regions etween the CAD-CS nd NoCAD-CS runs nd isolte the role of the wedge front on the convection. Preliminry results suggest tht lthough significnt updrft rottion within the convection ws present in oth runs, the frequency of modertely strong updrft rottion ws notly higher without the presence of CAD. While more low-level environmentl stremwise vorticity is ville for tilting into the verticl y convection long the CAD cold dome periphery, the convection without the presence of CAD ppers to hve more overll updrft rottion likely due to the incresed moisture nd instility. 7. REFERENCES Biley, C. M., G. Hrtfield, G. M. Lckmnn, K. Keeter, nd S. Shrp, 2003: An ojective climtology, clssifiction scheme, nd ssessment of sensile wether impcts for Applchin cold-ir dmming. We. Forecsting, 18, 641-661. Bell, G. D., nd L. F. Bosrt, 1988: Applchin cold-ir dmming. Mon. We. Rev., 116, 137-161. Crr, J. A., 1951: The Est Cost ckdoor front of My 16-20, 1951. Mon. We. Rev., 79, 100-110. Fores, G. S., R. A. Anthes, nd D. W. Thomson, 1987: Synoptic nd mesoscle spects of n Applchin ice storm ssocited with cold-ir dmming. Mon. We. Rev., 115, 564-591. Keeter, K., S. Businger, L. G. Lee, nd J. S. Wldstreicher, 1995: Winter wether forecsting throughout the estern United Sttes. Prt III: The effects of topogrphy nd the vriility of winter wether in the Crolins nd Virgini. We. Forecsting, 10, 42-60. 6. ACKNOWLEDGEMENTS Support for this reserch ws provided y the Collortive Science, Technology & Applied Reserch Progrm (CSTAR, Grnt #: NA07NWS4680002). We thnk the Ntionl Wether Service, specificlly Trish Plmer (WFO Pechtree City, GA), Mike Cmmrt (WFO Columi, SC), nd WFO Rleigh, NC for providing CAD event dt, suggestions, nd reserch feedck. We would like to cknowledge Drs. Mtthew Prker nd Annth Aiyyer for helpful insights nd discussions concerning this prolem nd Dr. Roert Fovell for model code tht llowed for esier terrin flttening in the numericl simultions.