Structure and Evolution of rainfall inNumerically Simulated Landfalling Hurricanes

Sytske Kimball, University of South Alabama, Mobile, AL

IntroductionThe Penn State/NCAR mesoscale model (MM5) is initialized with a southerly geostrophic wind of 8 m s-1. Embedded in this flow is a hurricane vortex with initial minimum surface pressure (PSMIN) of 970.6 mb and 42 km radius of maximum winds (RMW). The intensity and size properties of this vortex are based on the averaged properties of hurricanes making landfall in the north-central Gulf of Mexico during 1988 - 2002. The sea surface temperatures (SSTs) in the model are kept constant at 28°C and a straight, west-east oriented coastline is located at 30;N, 387 km north of the initial position of the vortex. The land surface is flat, has a height of 0.1 m above the sea surface, has an initial surface temperature of 28;C, and is covered by a single land-use category. A control simulation (noland) consisting of a water surface only is compared to simulations with different land-use categories.

Storm MotionWhen the nested domain is switched on, the storm center is located 195 km south of the coastline. The figure above shows the storm tracks in (part of) the nested domain for all 7 cases. Fifteen hours after the nested domain is turned on, the storm centers cross the coastline about 222 km east of their initial location in the nested domain. The presence of land does not seem to influence the speed or track of the storm. A small deviation in the tracks occurs after landfall, but it is never larger than a distance of about 20 km. The figure below shows the track speed of all cases. Speeds differ on slightly between cases, including the land case.

Storm IntensityThe figure to the left shows the evolution of storm intensity measured in terms of minimum surface pressure (PSMIN). The noland simulation weakens slowly until about t=18h when slow intensification begins. This storm remains a category 2 hurricane throughout the entire simulation. As expected, the other two storms continue to weaken after their centers cross the coastline at t=15h. Simulations with a larger roughness length (50 cm: urban, evergeen, needle) weaken more, as expected. Keeping moisture availability (MA) constant and increasing roughness length (RL) always leads to a weaker storm. However, keeping RL constant and lowering MA does not always imply a weaker storm. This is illustrated for example by needle (RL=50 cm; MA=30%) and evergreen (RL=50 cm; MA=50%); shortly after landfall evergreen is stronger, but towards the end of the simulation it is weaker in spite of having more surface moisture available. The largest difference in intensity between any two cases (urban and dryland) at the end of the simulation is 6 hPa, less than a Saffir-Simpson category.

AcknowledgementsThis works is supported by NSF Award No. ATM-0239492This work was made possible in part by a grant of high performance computing resources and technical support from the Alabama Supercomputer Authority.

Rainfall AccumulationThe figures above show the accumulated rainfall for the 34h simulation period for all 7 cases. Case noland (at left) displays a mostly symmetric pattern of accumulated rainfall around the storm center. With time, the rainfall amount and areal coverage increases as the storm intensifies around t=18h. In the earlier half of its life, the storm drops slightly more rain to the right of the track, in the second half of its life this is reversed. The landfalling storms display more pronounced asymmetries with rainfall maxima to the right of the track before landfall in all cases. As the storms approach the coastline, rainfall totals drop off. After landfall, rainfall accumulations are less than the amounts recorded to the right of the track while the storms were over water, and rainfall becomes more symmetrically distributed around the storm center. However, local areas of larger rainfall accumulations (≥ 14 cm) are seen just onshore to the left of the storm track.After the storm center has crossed the coastline, the distribution of rainfall accumulation over land varies considerably from case to case. As MA increases, so do rainfall accumulations and areal coverage of rainfall. Dry cases urban and savanna display smaller rainfall accumulations and smaller areal coverage than their moister counterparts. Rainfall becomes more symmetrically distributed around the storm center; dry cases savanna and, especially urban, form exceptions with higher rainfall accumulations on the right hand, or water, side. Also noticeable is a rainband feature that causes mostly rainfall off-shore and is least prominent in dry cases savanna and urban. A cross-shaped feature can be seen just off-shore to the right of the storm track in the accumulated rainfall totals in all cases except urban, this is associated with the rainband being orientated more perpendicular to the coastline around t=21 h and at a more acute angle around t=27h. In between those timeperiods the areal extent and rainfall rates of the rainband are reduced.

High RLLow RL

Dry

Moi

st

Rainfall AsymmetryThe figures to the right show various quantities as a function of azimuth and radius or height. All quantities are averaged over t= 5-10h, before the storm center crosses the coastline but when a strong left-right (L-R) asymmetry in rainfall can be seen (see figures at left). As moisture availability (MA) decreases (case urban, column furthest right) more rain falls on the right side of the storm and less on the left than in moist cases (evergreen, center column). Urban has less rainfall on the left, despite the fact that it has higher latent heat fluxes on the left than evergreen. Hence, there may be more dry air on the left of the dry case storm, causing the rain to evaporate and the fluxes to increase. All cases, including no-land, display a L-R asymmetry in the wind fields. This is partly caused by the storm motion. However, for the same forward motion (bottom left of poster) three different tangential and radial wind fields are seen: tangential and radial winds are stronger on the right of the storm for lower MA cases. This explain the stronger fluxes on the right of the storm, but there is no logical explanation for the stronger winds. The windspeed shows no correlation with RL (not shown).It is possible, therefore, that the convective asymmetry is forced by dry air intrusion and that the configuration of the convection drives the structure of the low-level wind field and not vice versa. In other words, stronger rainfall accompanied by stronger updrafts, forces stronger low-level inflow (i.e. radial wind). This, in turn, causes inward advection of angular momentum and, via angular momentum conservation, stronger tangential winds. Since all fields (rainfall, upward motion (not shown), radial winds, tangential winds) are co-located, this seems a likely sequence of events. Before concluding that dry air intrusion drives the convection, dynamic forcing mechanisms must be considered. The tangential divergence field (white contours, 3rd row at right) shows that maximum convergence, as expected, is located just downstream from the tangential wind and rain maxima. If anything, the rainfall maximum occurs in an area of low-level speed divergence in all cases. This further strengthens the case for dry air intrusion as the forcing mechanism behind the observed asymmetries.

Dry air intrusionDetailed examination of the 3-dimensional equivalent potential temperature (θe) and wind fields in the eyewall annulus reveals that all cases experience dry air intrusion from the drier (by definition, since a hurricane core is moist) environment. The dry air paths are depicted schematically below. In each case, including no-land, the moisture is pushed into the vortex by the southerly environmental flow which is slightly stronger than the forward motion of the vortex. The dry air gets entrained into the vortex circulation and wraps cyclonically from the southeast around the outer edge of the front half of the eyewall annulus. The upper portion (above about 850 hPa) of the inflowing dry air experiences outflow in the front half of the vortex and, therefore, continues to encircle the vortex until it reaches the rear. Here, the vortex experiences inflow up to about 600 hPa at the outer edge of eyewall annulus. The dry air enters the outer edge of the eyewall annulus in the southwest quadrant and moves cyclonically and inward to reach the inner edge of the eyewall in the northeast or right front (RF) quadrant, this is depicted by the black arrow. A patch of low θe air can clearly be seen in the right front quadrant extending upto almost 650 hPa (see θe plots in 5th row at right). The lower part (below 850 hPa) of the dry environmental flow that entrains into the vortex’ circulation from the southeast, experiences inflow when it reaches the front of the vortex and, hence, enters the eyewall annulus in the left front quadrant of the eyewall, and establishes itself as the dry tongue seen in the θe field (dark grey arrow) extending into the left half of the vortex. The lower portion of the RR quadrant of the vortex remains relatively unaffected by dry air intrusion and with the pocket of low θe aloft in the RF quadrant creates a slantwise unstable situation. As a result, large amounts of cloud and rainwater form just downstream of the RR quadrant. In the left-front (LF) quadrant a similar slantwise instability can be seen but it is capped by high eyewall θe air and less convective activity occurs than in the RF quadrant. The convection in the left half of the no-land storm increases the low-level inflow of dry air into that side of the storm, explaining why more dry air is seen there than in the land cases.In the land cases, the situation is more complex. In addition to the environmental source, another source of dry air enters the vortex from the landmass to its north (light grey arrow). One channel comes off land to the left of the vortex and entrains into the outer edge of the annulus in the LR quadrant. Here the air rises and merges with the environmental dry air channel (black arrow). This reinforces the higher level dry air, and associated instability, on the right side of the vortex. Again slantwise instability allows the formation of convection but this time with a maximum between the RR and RF quadrants. More hydrometeors form than in the no-land case (bottom row at right). At the same time, dry air from the land surface enters the outer edges of the front of the eyewall at very low levels, any rain being advected around the vortex from the right side by the tangential winds evaporates in this dry air.

Moist Dry

Rai

nfal

l (cm

)La

tent

hea

t flu

x (W

/m2 )

Tan

gent

ial w

ind

(m/s

)R

adia

l win

d (m

/s)

The

tae

(K)

Hyd

rom

eteo

rs (

g/kg

)

NO-LAND