|
||||||||||||||||||||||||||||||||||||||||||
|
2.6 RAINFALL
Tropical cyclones contain extremely warm, moist air, cover an area of up to 1 million square kilometres, and move relatively slowly. Hence, they are capable of producing very heavy rain. Especially difficult forecast situations arise when a cyclone is moving inland and redeveloping as a flood depression, or during extratropical transformation, when a new, hybrid system of the type described in Section 2.5 may rapidly develop and produce heavy rain.
If above are generally evident, upper support is FAVOURABLE, otherwise UNFAVOURABLE |
||
|
||
|
||
|
||
| Likelihood of transformation resulting in a STRONG extratropical low | ||
|
||
| Transformation Type | Upper Support |
|
| FAVOURABLE | UNFAVOURABLE | |
| COMPOUND | PROBABLE | POSSIBLE Especially if existing frontal low is deepening |
| COMPLEX | PROBABLE | NOT LIKELY but watch for:"baroclinic leaf"
forming in frontal zone "self-development" of trough upstream of tropical cyclone outflow ridge |
Figure 2.22: A suggested decision strategy for analysing and predicting extratropical transformation of tropical cyclones.
Quantitative prediction of tropical cyclone rainfall is very difficult for three reasons:
1. Rainfall itself is difficult to measure accurately, which hinders both operational analysis of rainfall and the development of improved forecasting aids;
2. Current errors in track prediction mean that accurate rainfall estimates cannot necessarily be transformed into precise predictions, this is especially a problem when a cyclone is moving near regions of significant orography;
3. Interactions between tropical cyclones and other weather systems (Section 2.5) are themselves complicated and poorly understood, so that heavy rain in areas of large-scale ascent and high humidity are difficult to predict;
4. Even within clearly defined threat areas, mesoscale processes, which are poorly understood and difficult to monitor, may determine the distribution of heavy rainfall.
As with other aspects of tropical cyclone structure forecasting, operational numerical models generally lack the resolution and physical processes to predict rainfall accurately and explicitly but they are rapidly being improved. They may be especially useful in determining threat areas in complicated situations.
Figure 2.23: Three-hourly isohyets (mm) centred on 1200 UTC 21 June 1972 (left) and 1200 UTC 22 June 1972 (right) during the extratropical transformation of Hurricane Agnes. These times coincide with the composite analyses in Figs. 2.17 and 2.18 respectively. Note the complex pattern and extreme spatial variability (DiMego and Bosart, 1982a).
Research on tropical cyclone rainfall has tended towards intensive examination of a few cases. Improvements in forecasting ability, especially of regional peculiarities, would be well served by the development of a simple archive of the relevant synoptic features and resulting rainfall for a wide variety of cases. Were this available, forecasters would be able to classify each new situation within the range of typical patterns and perhaps make a more accurate prediction of the heavy rain threat area.
Rainfall is very difficult to measure accurately, especially for small areas, heavy rain, and short periods of time. Three methods are commonly used:
Rain Gauges: are very simple and direct. Unfortunately, convective rainfall is extremely variable in the horizontal (Fig. 2.23), so a rain gauge network must be very dense. Otherwise a local extreme can be misinterpreted as the amount for an entire region. High winds such as found in tropical cyclones may also cause turbulence around the gauge and lessen its catch unless special shielding is used. Rain gauge networks are of most value in providing the ground truth, however limited, for indirect radar and satellite estimates.
Radar: can continuously cover a 200 km radius circle over all conditions, unless blocked by terrain. It is less prone to sampling problems than gauges because of its continuous spatial coverage and implicit averaging over an area determined by the pulse length and beamwidth. Radar measures the strength of radio pulses scattered back to the radar by precipitation particles, which is related to their size and type (rain, snow, hail) by a rather complex equation. The size and type of particles is in turn related to rain rate by a less clearly known relationship, based on empirical 'Z-R relationships' (Batten, 1973) determined by comparing radar and rain gauge measurements. The relationship varies according to the radar and type of weather system.
Satellite Imagery: can be used to estimate rainfall by empirical relationships based on the shape, texture and (infrared) black-body temperature of the tops of clouds. Although the relationships are not overly accurate, the large area and frequent time coverage make this a useful initial estimate of tropical cyclone rainfall over the ocean.
Satellite Microwave Measurements: use either radiometers measuring upwelling microwave radiation, or active "radars" in space, which work on the same principle as ground-based radars. Several systems are currently being tested and show significant promise for quantitative determination of tropical cyclone rainfall. Their operational use is untested at this stage, however.
Because of the meteorological complexity, measurement limitations, and lack of objective aids, analysis and forecasting of heavy rain associated with tropical cyclones can at best be indicative of likely outcomes. A suggested mode of operation is to first classify the situation as uncomplicated or complicated.
Uncomplicated situations satisfy the following conditions:
1. The tropical cyclone is relatively well developed;
2. The tropical cyclone is a day or less from landfall and is moving rapidly enough such that its precipitating region will pass over a given point completely within a day or less;
3. There are no topographic features within the path of the tropical cyclone which are significant enough to appreciably alter the rainfall;
4. There are no significant nearby weather systems, including frontal zones, jet streams, or upper-level cut-off lows, which are likely to interact with the tropical cyclone during its passage inland.
Unfortunately, the majority of forecast situations near landfall involve rapid changes in the character and structure of the precipitation as the system moves inland and interacts with orography and other weather systems. Simple extrapolation procedures will not work very well and the situation is therefore complicated. About the best the forecaster can do in advance is identify a general threat area based on the locations of the tropical cyclone and surrounding weather systems. The actual locations of heavy rain must then be identified as the event proceeds in order to identify areas which are accumulating dangerous amounts of rainfall. In the absence of dominating terrain, mesoscale processes such as the development of new convective cells at the merger of old convective outflow boundaries generally determine where within the threat area the heavy rain actually falls. If these mesoscale focusing mechanisms are quasi-stationary, extremely heavy rain may fall even though the convective elements are moving quickly.
2.6.2.1 Uncomplicated Situations
Under these conditions, a reasonable precipitation forecast can be made by estimating the present spatial distribution of rainfall rates within the tropical cyclone and assuming that they will persist throughout its passage over a given point. A technique for making such forecasts using geostationary satellite imagery is summarised in Fig. 2.24 and a simple example is shown in Fig. 2.25 (Spayd and Scofield, 1984). Note that the rainfall rates vary by a factor of 5 or more for a given region of the satellite signature (precipitation class), so that precision in calculations is not required. The dominant factors are the translation speed of the tropical cyclone, the size of the region of active precipitation, and the overall character of the satellite signature (expanding and cooling or contracting and warming).
The example in Fig. 2.25 is included to indicate the method of making time and distance calculations. The example uses average rain rates for all regions and does not have a leading-edge band (LEB) or embedded cloud tops (ECT). Total rainfall in LEB and ECT regions can be very significant, especially if a band is oriented roughly parallel to the tropical cyclone motion. In this case one region may be effected by the entire length of the cloud band. Because the actual locations of ECTs vary, it is acceptable to use a percent-coverage approach. For example, the passage of an outer band area 200 km wide at a speed of 5 ms-1 (10 kt, 20 km h-1)) with embedded cloud tops covering 10% of its area would result in an expected exposure of 9 h of OBA and 1 h of ECT.
Figure 2.24: Determining rainfall rates for an uncomplicated situation following Spayd and Scofield (1984). This technique is illustrated in Fig. 2.25. It should not be applied for periods longer than 24 h. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Precipitation Class | Rain Rate (mm/h) | Distance (km) | Time (hours) (distance/speed) | Precipitation Contribution | Storm Total |
|---|---|---|---|---|---|
| periphery | 8 | 225 | 11025 | 90 | |
| 90mm | |||||
| Precipitation Class | Rain Rate (mm/h) | Distance (km) | Time (hours) (distance/speed) | Precipitation Contribution | Storm Total |
|---|---|---|---|---|---|
| periphery (front) | 8 | 100 | 5 | 40 | |
| core (front) | 25 | 30 | 1 | 30 | |
| eye wall (front) | 50 | 10 | 0.5 | 25 | |
| eye | 20 | 1 | |||
| eye wall (rear) | 50 | 10 | 0.5 | 25 | |
| core (rear) | 25 | 30 | 1 | 30 | |
| periphery (rear) | 8 | 50 | 2.5 | 20 | |
| 170mm | |||||
Figure 2.25: Example of the technique used in Fig. 2.24. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
2.6.2.2 Complicated Situations
The task in complicated rainfall situations is twofold: 1. determine in advance of onset the general areas where heavy rain is probable; and, 2. monitor the event as it develops for areas that accumulating dangerous precipitation totals. As an aid to the first task, a checklist is provided in Fig. 2.26, together with some typical synoptic-scale patterns associated with mid-latitude flash floods in Fig. 2.27 (adapted from Maddox et al., 1979).
Similar configurations to those in Fig. 2.27 occur for tropical cyclones in subtropical latitudes with complicated rainfall situations. The general threat area is between the upstream 500 hPa trough and the downstream ridge, in the vicinity of the 850 hPa low-level jet and the cool (downstream) side of any low-level frontal zones (even weak ones) or convergence lines. Surface boundaries are an especially important factor in localising the heavy rain. Such boundaries do not have to be synoptic-scale fronts; it is quite common that an area of heavy convection will move away but leave a pool of cool air and weak high pressure at the surface. The boundary between this cool region and the warm, moist air flowing into the large-scale system is a preferred area for repeated convective development, with each episode renewing the boundary for the next. Such a situation is shown in Fig. 2.27(c).
The series of composite charts for the remains of Hurricane Agnes (Figs. 2.17-2.19) illustrate some of the ingredients of a complicated situation involving a tropical cyclone remnant, orography, and pre-existent upper-level waves and low-level frontal boundaries. At 1200 UTC 20 June (Fig. 2.17), the center of the remnants of Agnes was just onshore near 32oN 84oW. A 500 hPa trough extended north-northwestward from the center with a 500 hPa ridge along 75oW. Within the general threat area defined by the trough and ridge, other features were highly dispersed. The 200 hPa jet was well to the north and surface boundaries and warm advection regions were near the north and east edges of the map.
By 21 June (Fig. 2.18), Agnes had moved northeastward along the coast. The frontal zone had advanced from the west, warm-air advection had developed over a large area to the northeast of Agnes' center, and the 500 hPa ridge and 200 hPa jet had redeveloped and rotated counter-clockwise, bringing nearly all of the heavy rain factors together in the region between 35-40oN and 77-83oW. Heavy rain was falling in the areas marked with the four-dot symbols (see Fig. 2.23 for the rainfall accumulation). The separate area of heavy rain near 42oN and 80oW was apparently associated with the northern portion of the advancing trough in the westerlies. The overall distribution of heavy rain relative to the trough and tropical cyclone remnant resembles the schematic in Fig. 2.27d.
On 22 June (Fig. 2.19) the region of juxtaposed heavy rain threat factors had shifted northward by about 5o lat, associated with the movement of Agnes. The southern end of the heavy rain area was determined by the 500 hPa trough axis (westward from Agnes' centre) and a transition from warm to cold air advection.
| This is intended to be an example of how a real case
appears in a composite chart sequence. Variability is the rule, and considerable work
remains to be done to pool the experiences of forecasters in different basins by making
case summaries available using common symbology and notation. Determining threat areas: Heavy rain threat areas should be revised at least every 12 hours. Threat areas can change. New threat areas can develop. This procedure requires a set of composite charts prepared as described in Fig. 2.15. The heavy rain threat area for the time (analysis or prognosis) covered by the composite chart is defined as the intersection of areas defined by surface and 850, 500, and 200 hPa features as listed below:
The more of the above features present in a region, the greater the threat of heavy rain. The surface features should receive the maximum emphasis. Monitoring the Event: Use the rainfall rates for OBA and ECT from Fig. 2.24 to estimate rainfall rates every 1-2 hours or as often as imagery is available. Be especially alert for small, rapidly expanding cells as they typically produce much higher rain rates than large, impressive cloud shields which are often mostly stratiform. Maintain a single map with positions of active cells from each estimate as this will indicate:
|
Figure 2.27: Typical patterns associated with mid-latitude flash floods (a-c; adapted from Maddox et al., 1979) and with the remnants of Hurricane Agnes approaching the baroclinic westerlies (d; adapted from Carr and Bosart, 1978).
| Bureau Home || BMRC Home || Search || Contact BMRC Webmaster |
Home | About Us | Learn about Meteorology | Contacts | Search | Help | Feedback Weather and Warnings | Climate | Hydrology | Numerical Prediction | About Services | Registered Users | SILO |
|
© Copyright Commonwealth of Australia 2008, Bureau of Meteorology (ABN 92 637 533 532) Please note the Copyright Notice and Disclaimer statements relating to the use of the information on this site and our site Privacy and Accessibility statements. Users of these web pages are deemed to have read and accepted the conditions described in the Copyright, Disclaimer, and Privacy statements. Please also note the Acknowledgement notice relating to the use of information on this site. No unsolicited commercial email. |
