Forecasting convective rainfall:convective initiation, heavy precipitation and flash floodingRobert FovellUniversity of California, Los Angelesrfovell@ucla.edu

Heavy precipitation at a location = intensity + longevity

Common sources of heavy precipitation in U.S.Mesoscale convective systems and vorticesOrographically induced, trapped or influenced stormsLandfalling tropical cyclones

Mesoscale Convective Systems (MCSs)

MCSs & precipitation factsCommon types: squall-lines and supercellsLarge % of warm season rainfall in U.S. and flash floods (Maddox et al. 1979; Doswell et al. 1996)Initiation & motion often not well forecasted by operational models (Davis et al. 2003; Bukovsky et al. 2006)Boundary layer, surface and convective schemes Achilles heels of regional-scale modelsImproved convective parameterizations help simulating accurate propagation (Anderson et al. 2007; Bukovsky et al. 2006)Supercells often produce intense but not heavy rainfallForm in highly sheared environmentsTend to move quickly, not stay in one place

Seasonality of flash floods in U.S.Maddox et al. (1979)Contribution of warm season MCSs clearly seenNumber of events

Linear MCS archetypes(e.g., squall-lines)Parker and Johnson (2000)58%

19%

19%

Squall-lines usually multicellular

The multicell stormBrowning et al. (1976)Four cells at a single timeOr a single cell at four times:unsteady

The multicell stormBrowning et al. (1976)Fovell and Tan (1998)Unsteadiness =episodic entrainment owingto local buoyancy-inducedcirculations.

Storm motion mattersDoswell et al. (1996)How a storm moves over a specificlocation determines rainfallreceived

Storm motion mattersDoswell et al. (1996)

Forecasting MCS motion(or lack of motion)

19980714 - North Plains

Some common rules of thumb ingredientsCAPE (Convective Available Potential Energy)CIN (Convective Inhibition)Precipitable waterVertical shear - magnitude and directionLow-level jetMidlevel cyclonic circulations

Some common rules of thumbMCSs tend to propagate towards the most unstable air1000-500 mb layer mean RH 70%MCSs tend to propagate parallel to 1000-500 mb thickness contoursMCSs favored where thickness contours divergeMCSs back-build towards higher CINDevelopment favored downshear of midlevel cyclonic circulations

70% layer RHJunker et al. (1999)# = precip. categoryRH > 70%70% RH rule of thumb

Implication:Relative humiditymore skillful thanabsolute humidity

MCSs tend to follow thickness contoursImplication: vertical shear determinesMCS orientation and motion.Thickness divergence likely impliesrising motion

Back-building towards higher CINLifting takes longer where thereis more resistance

Corfidi vector methodPropagation is vector differenceP = S - CTherefore, S = C + P

Example

Schematic exampleWe wish to forecast system motionSo we need to understand what controlscell motion and propagation

Individual cell motionGo with the flowAgrees with previous observations (e.g, Fankhauser 1964) and theory (classic studies of Kuo and Asai)Cells tend to move at850-300 mb layer wind speed*Corfidi et al. (1996)*Layer wind weighted towards lower troposphere,using winds determined around MCS genesis.Later some slight deviation to the right often appears

Individual cell motionCells tend to move at850-300 mb layer wind speedCell direction comparableTo 850-300 mb layerwind directionCorfidi et al. (1996)

Composite severe MCS hodographBluestein and Jain (1985)

Composite severe MCS hodographLow-level jets (LLJs)are common

NoteP ~ -LLJBluestein and Jain (1985)

Propagation vector and LLJ Many storm environments have a low-level jet (LLJ) or wind maximum

Propagation vector oftenanti-parallel to LLJCorfidi et al. (1996)Propagation vector directionP ~ -LLJ

Forecasting system motionusing antecedent informationCell motion ~ 850-300 mb windPropagation ~ equal/opposite to LLJS = C - LLJ

Evaluation of Corfidi methodMethod skillful in predictingsystem speed and directionCorfidi et al. (1996)

Limitations to Corfidi methodWind estimates need frequent updatingInfluence of topography on storm initiation, motion ignoredSome storms deviate significantly from predicted direction (e.g., bow echoes)P ~ -LLJ does not directly capture reason systems organize (shear) or move (cold pools)Beware of boundaries!Corfidi (2003) modified vector method

http://locust.mmm.ucar.edu/episodes

5 June 2004X = Hays,Kansas, USA

Mesoscale Convective Vortices (MCVs)

Cyclonic vortex following squall lineNot a cleanMCV case

Potential vorticity (PV) anomaliesPV anomaly shown drifting in westerly sheared flowRaymond and Jiang (1990)

Potential vorticity (PV) anomaliesAscent occurs on windward (here, east) side destabilizationRaymond and Jiang (1990)

Potential vorticity (PV) anomaliesCyclonic circulation itself results in ascent on east sideRaymond and Jiang (1990)

Potential vorticity (PV) anomaliesCombination: uplift & destabilization onwindward side AND downshear sideRaymond and Jiang (1990)

Composite analysis of MCV heavy rain eventsSchumacher and Johnson (2008)600 mb vorticity (color), heights and winds.Map for scale only Based on 6 cases poorly forecasted by models

Composite at time of heaviest rain (t = 0h)

Heaviest rain in early morning

Heaviest rain south of MCV in 600 mb trough

Schumachers situationHairpin hodograph:Sharp flow reversal above LLJ

Schumachers situationSouth side of MCV is windward at low-levelsand downshear relative to midlevel vortex

Back-buildingSchumacher and Johnson (2005)Doswell et al. (1996)Ground-relative system speed ~ 0

Evolution of the heavy rain event600 mb vorticity, 900 mb winds & isotachsAt t - 12h (afternoon): - MCV located farther west - 900 mb winds fairly lightSchumacher and Johnson (2008)

Evolution of the heavy rain event600 mb vorticity, 900 mb winds & isotachsAt t - 6h (evening): - MCV drifted west - 900 mb winds strengthening (LLJ intensifying)Schumacher and Johnson (2008)

Evolution of the heavy rain event600 mb vorticity, 900 mb winds & isotachsAt time of heaviest rain (midnight): - 900 mb jet well developed - LLJ located east, south of MCVSchumacher and Johnson (2008)

Evolution of the heavy rain event600 mb vorticity, 900 mb winds & isotachsAt t + 6h (morning):rain decreases as LLJ weakensSchumacher and Johnson (2008)

Episodes of MCSs& predictabilityCarbone et al. (2002)Hovmoller diagrams reveal westward- propagating MCSsNote envelope of several systems with connections

MCV role in predictabilityCarbone et al. (2002)

Training lines of cellsSchumacher and Johnson (2005) In Asia, stationary front could be the Mei-Yu (China), Baiu (Japan) or Changma (Korea) front

Motion along the front and/or continuous back- building

Sun and Lee (2002)Record 619 mm in 15 h at Ganghwa, KoreaXLee et al. (2008)shear

2-3 April 2006

Why did new cells appearahead of the mature line?

New cell initiation ahead of squall-linesThe waves themselves disturb the storm inflow Fovell et al. (2006)

New cell initiation ahead of squall-linessome of which can develop intoprecipitating, even deep, convection Fovell et al. (2006)

New cell initiation ahead of squall-lines150 km14 kmFovell et al. (2006)

Importance of antecedentsoil moisture conditions(Generally not captured well by models)

Tropical Storm Erin (2007)http://en.wikipedia.org/wiki/Image:Erin_2007_track.png

Erins redevelopmentover OklahomaEmanuel (2008)http://www.meteo.mcgill.ca/cyclone/lib/exe/fetch.php?id=start&cache=cache&media=wed2030.ppt

Erin inland reintensificationHot and wet loamy soil can rapidly transfer energy to atmospherePrevious rainfall events left Oklahomas soil very wetNeed to consider antecedent soil moisture and soil typeEmanuel (2008)see also Emanuel et al. (2008)

Soil T as Erin passedEmanuel (2008)

end

*Note in passing*Also show multiple cells in conceptual model

*Also show multiple cells in conceptual model

Here they work together