How often does severe clear air turbulence occur over tropical oceans?

Transcription

1 train, generated by Arran and the Kintyre peninsula, which propagates towards the south-west. Both the horizontal wavelength and the orientation of the waves compare well with waves seen in the high-resolution satellite imagery (Fig. 2). The predicted horizontal wavelength of about 9 km corresponds to a value of 12 of around 0.49 km-2. As shown in Fig. 6, the variation of the Scorer parameter with height is consistent with such a wavelength being trapped in the lower troposphere. Given the occurrence of gravity waves, a layer of the atmosphere must be suitably moist for clouds to appear in the ascent regions of the waves (Fig. 3). In observing wave clouds of the kind seen on 15 March 2000, we are probably seeing only one (moist) level of a deeper wave field. Measuring, understanding and predicting the full 3D wave field remains an important challenge for meteorological research. References Bader, M. J., Forbes, G. S., Grant, J. R., Lilley, R. B. E. and Waters, A. J. (1995) Images in weather forecasting: A practical guide for interpreting satellite and radar imagey. Cambridge University Press Holton, J. R. (1992) An introduction to dynamic meteorology. Harcourt Publishers Ltd (a subsidiary of Harcourt International Ltd) Mobbs, S. D. and Darby, M. S. (1989) A general method for the linear stability analysis of stratified shear flows. Q. J. R. Meteorol. SOL., 115, pp Scorer, R. S. (1949) Theory of waves in the lee of mountains. Q. J. R. Meteorol. SOC., 75, pp (1986) Cloud investigation by satellite. Ellis Horwood, Chichester -(1990) Satellite as microscope. Ellis Horwood, Chichester Correspondence to: Dr S. Vosper, School of the Environment, University of beds, Leeds L S~JT. 0 Royal Meteorological Society, How often does severe clear air turbulence occur over tropical oceans? W. T. Roach' and C. E. Bysouth2 ' Crowthorne, Berkshire Met Office, Bracknell The motivation for writing this paper came from a report in Weather of prolonged and severe clear air turbulence (CAT) over the tropical Pacific by Blanchard (1999), followed shortly by a reply from Bysouth (2000) showing that this incident was forecast using the Met Office operational numerical forecast model - or Unified Model (Cullen 1993). This seemed anomalous, since CAT is normally associated with temperate (polar front) and subtropical jet streams usually found poleward of 30" latitude. However, examination of the wind field in Fig. 1 of Bysouth (2000) showed that the Northern Hemisphere subtropical jet stream (NHSTJ) had apparently buckled a few days before (rather like a giant breaking wave) to, form a very long disrupting trough (at the base of the 'breaker') slanting west-south-west to about 10"N. This trough coincided with the area of high CAT index used by the forecast model. Disrupting troughs occur quite often in temperate-latitude jet streams, and have long been known to contain (often severe) CAT. They sometimes form part of the process of detachment of separate vortices - or 'cold pools'. However, subtropical jets in both hemispheres appear to exhibit much more stable behaviour than temperate-latitude jets; subtropical jets rarely buckle and have not been seen to form cold pools in the study presented here. It was therefore decided to make briefcase- 8

2 studies of the synoptic patterns associated with the Blanchard and two other reports of severe CAT, and use the (provisional) conclusions from these studies to see whether anything useful could be inferred about the incidence of severe CAT in tropical latitudes - i.e. equatorward of about 30 latitude - in any season or longitude. The paper concludes with an Appendix discussing the forecast model used, and some verification aspects. However, before presenting this evidence, it is appropriate to give a brief outline of what is currently known about CAT, and its causes and maintenance. What is CAT? With the birth of the jet aircraft age soon after the end of World War 11, pilot reports of bumpy rides (like riding over cobblestones) in cloudless conditions at high levels (-1 0 km) soon began to be received. This bumpiness was soon named clear air turbulence, or CAT. CAT was usually light and intermittent, but very occasionally became severe enough to lift passengers from their seats, and even cause injuries. Although the level of severity attained was below that encountered within severe thunderstorms (which can break up aircraft), the unexpected nature of severe CAT made it a significant hazard. It was soon found that CAT usually occurred in layers of strong wind shear below and above the cores of temperate-latitude and subtropical jet streams. Therefore, our interest was aroused by the Blanchard report of severe CAT over the tropical Pacific referred to above. What is the cause of CAT? The atmosphere is, on average, thermally stable - more so in the stratosphere than the troposphere - but this stability is not uniform; the atmosphere has a laminated structure of layers of varying thickness and stability. If a parcel of fluid within a stable layer is vertically displaced, it will oscillate about its original position and in so doing will radiate the energy of the initial displacement away as gravity waves. The frequency of oscillation, known as the Brunt-Vaisala frequency, N, will depend upon the degree of stability, and its wavelength will depend on the layer thickness. As in the sea and on its surface, gravity waves in the atmosphere are ubiquitous; they are occasionally marked (usually at high levels if no low cloud is present) by billow clouds if the air in the gravity wave is near saturation. In addition, the wind, V, also varies with height, z, and a stable layer will usually have some change of wind through it. If this wind change with height (or wind shear, S=AvlAz) becomes comparable to A!, the wind will amplify any existing gravity waves (or create them if initially absent) and overturn them into rollers (or vortices) which then break into general turbulence which, if not maintained, decays after a few minutes. This is known as Kelvin-Helmholtz instability (KHI).* An aircraft encountering large KHI billows in the initial act of overturning may experience large upward or downward displacements similar to updraughts or downdraughts in a thunderstorm. What maintains CAT? Many encounters with light or moderate CAT may be due to the sporadic and local formation of KHI billows - analogous to the formation of white horses by a strong breeze at sea independently of the synoptic-scale wind patterns. However, if the wind shear is being increased over a considerable volume by large-scale synoptic development, KHI billows will start to form at many locations within this area (an analogy is the increase of white horses at sea in a rising wind). The local disturbances caused by these billows will generate further gravity waves, which then develop further billows, eventually merging into a chaotic (and probably unbroken) layer of moderate or ~ * If the Richardson number, Ri(=Nz/p), decreases below %, KHI will set in. In other words, the time, AzlAK taken for an overturning particle within a KHI event to move from top to bottom of the layer is comparable to the natural oscillation period, 1/N, of the layer, and is a form of resonance. Note that two wave-breaking phenomena have been introduced in this paper: (i) the buckling of the subtropical jet stream, which occurs in a roughly horizontal plane on a scale of thousands of kilometres, and (ii) local overturning of a thermally stable layer thick enough to disturb the flight of an aircraft, which occurs in a vertical plane on a scale of kilometres. 9

3 severe turbulence of thickness of 1 km or more (see later case-studies). This turbulence is probably being maintained against dissipation by continuing large-scale development (e.g. as in frontogenesis in temperate latitudes, or the cases under study in this paper). An attempt to develop a theory of CAT maintenance was made by Roach (1970). He identified two principal dynamical factors that reduce Richardson number, Ri, following the flow?: (i) stretching and deformation of the horizontal flow, (ii) shear advection in which change of wind with height modified both the vertical wind shear and thermal stability. When CAT set in, frictional stresses caused by the turbulence had to be taken account of. This was tackled by deriving an expression for the rate of turbulent energy dissipation required to maintain Ri at a roughly constant value (of about 0.5) against the dynamical processes reducing it. The value of energy dissipation derived v2/24, where A Vis the velocity difference across the turbulent layer. A more recent study by Keller (1990) developed Roach's theory further by considering the relative contributions to changing Ri in isentropic flow made by dynamical processes and turbulent mixing, once this had set in. Keller also used some case-studies that showed significant coupling between observed fields of Ri and forecast fields of changing Ri in regions away from turbulence. This gave some confidence that the theory was on the right lines. Owing to the layered structure of the atmosphere, there were considerable difficulties in adapting Roach's theory for practical use as a CAT index in numerical forecast models. This task was attempted by Brown (1973), who developed a modified form of CJ - the 'Brown t The parameter chosen was the fractional rate of reduction of Ri, defined by: CP = -D( In Ri]lDt. Typical background values of Ri of 3 or less are found in the vicinity of strong jet streams. Therefore, a reduction of Ri by about a factor of 3-10 (or of In Ri by 1-2) will probably be enough to trigger CAT. This will take a time of order At = -A(ln Ri]lCP -1-3 hours using typical values of CP yielded by the model. During At, the air will have travelled of order 300 km, which is small compared with the horizontal dimensions of CAT discussed in the case-studies below. 10 indicator' - used in this paper, and defined in 'Forecast model' in the Appendix. Synoptic background: some case-studies (i) The Bfanchard report A survey of the 6-hour forecasts of the 250 mbar wind field a few days before the Blanchard report showed that the subtropical jet had started to buckle near 30"N, 175"E shortly before 1800 GMT on 18 February 1999 (Fig. 1). This buckling developed quickly into a giant fold in the NHSTJ, bodily displacing that part of the NHSTJ lying east of the trough to about 10"N. The trough itself extended from near 180" longitude to the Californian coast, with CAT forecast along much of its length. During the period 18 February-2 March 1999, the trough migrated eastwards for about km, on a trajectory taking it first well into tropical latitudes, and back into temperate latitudes by 28 February (Fig. 2). The trough fluctuated in sharpness during its life, but at no time showed any tendency to calve the cold pool familiar in temperate latitudes - possibly due to the weak Coriolis force in tropical latitudes. The maximum development of the trough appeared to occur the day before Blanchard's report. Values of 2-3 x s-l were forecast to occur at 1800 GMT on 23 February 1999 within a vast sloping sheet containing a total vertical shear of about 100 kn and having dimensions of about 3000 km long by 200 km wide by 5 km deep (Fig. 3). A north-south cross-section marked by. the white line through the forecast CAT patch in Fig. 3 is shown in Fig. 4 (see back cover). This cross-section shows the main wind trough line curving downwards from south to north (white dashed line) through the centre of the CAT area with an average slope of order 1 in 100. Also shown in Fig. 4 are the parts of the 340, 345 and 350K isentropes (lowest white line is 340K) through the trough zone. The vertical spreading of the isentropes through the trough is probably due to local convergence of air of lower thermal stability than is characteristic of temperate upper-frontal zones. The

4 60N 30N C Fig. 1 Forecast fields ofthe 250mbar wind vector and CATusing Brown s indicator (see section on Forecast mo&l for definition) in units of lo-. s-, valid at 1800GMTfLom data at 1200GMTon 18 February Areas ofnegative absolute voniciry are marked in white with a hard edge, and ure distinctfroni the shaded white areas of the CAT index. Thin curved white line is the great circle arc from Ftji to Los Angeles. JON 0 Fig. 2 Forecast areas of CATat 250mbar at 0600GMTon the days shown (1812 is 18 February 1999 etc.), associated with the moving disrupting trough in the subtropicaljet stream. This includes Blanchard s (1 999) report of severe CATon 24 February Dotted lines show the axes of the disrupting trough where it extends bevond the main areas of nrrbulence. Dashed line is the great circle track from Fiji to LAX Aiigeles. 11

5 Fig. 3 Forecast as in Fig. 1, but valid for 1800 GMTfrom data at 1200 GMTon 23 Februa y 1999 isentropes over the rest of the chart run horizontally, showing no evidence of a change of airmass; the trough after all is the same airmass folding upon itself. The field of vertical velocity (not shown here) exhibits a large zone of subsidence north of the trough line with a maximum of about 7cms-' (with respect to a constant pressure surface) near 300 mbar, about 500 km north of the trough line. There is also a region of weaker ascent (-3cmi') above 200mbar to the south of the trough, indicating that air ahead of the advancing trough line is being pushed upwards. A disruption of the whole NHSTJ may generate CAT on a scale and intensity over the north tropical Pacific not found anywhere else on this planet away from mountainous areas. Therefore, an aircraft flying through the 'white' area in Fig. 3 is quite likely to encounter severe CAT, although we have not found any aircraft reports from that area at the chart time - about a day before the Blanchard report. (ii) Severe CAT causing a fatality east of Tokyo This incident occurred on 28 December approximately 2 hours after the validity time of the forecast field shown in Fig. 5. The location of the incident is marked with an asterisk. The United Airlines flight had travelled about miles (1 760 km) from Tokyo and a meal was being served when the aircraft experienced an upward load of 1.8g for 6 seconds followed by -0.8g for the next 0.5 seconds; 74 passengers and 9 flight attendants were taken to hospital and one woman died. Figure 5 shows that the maximum value of coincides with buckling of the NHSTJ which had just started (see Fig. 2 for similar event). The site of the incident (in our view) is close. enough to the site of this buckling to be associated with the resultant CAT. (Note that the value of the maximum is about 25% less than in Fig. 1 but it should be noted that the model used in 1997 had a lower spatial resolution than the current model.) (iii) A private report Mr Folland (personal communication) reports that he took off from Los Angeles (USA) for Auckland (New Zealand) at local time on 11 February Turbulence started about 20 minutes after take-off and lasted for

6 ia 21 Fig. 5 Forecast $el& of the 300 nibar wind vector and CAT using Brown's indicator in units of 10-5s-', valid at 1200 GMT~~OIP~ data at 0000 GMTon 28 December 1997 about 2 hours at a severe level. The 6-hour forecast chart for 0600 GMT on 12 February 2000 was extracted (ie. roughly the same time as the Folland CAT report) and is shown in Fig. 6. The approximate location of the report is shown by the dotted white line in Fig. 6, which appears to be close enough to the white crescent of to be regarded as in good agreement with the forecast. The white crescent lies along a sharp trough in the wind field, apparently associated with buckling of the subtropical jet stream much further east of the sites of other jet-buckling events noted in this paper. How often does CAT occur in tropical latitudes? It appears from the above case-studies that severe CAT in tropical latitudes is principally (if not entirely) associated with disrupting upperair troughs generated by the buckling of the NHSTJ. These disrupting upper troughs appear to be well forecast by existing models. It therefore seems not unreasonable to construct a climatology of CAT based on short-term numerical forecasts of CAT since it is not possible to construct an observationally based climatology of CAT in tropical latitudes - or anywhere else - as the distribution of aircraft reports is highly non-uniform in space and time, and non-existent in many regions, particularly over southern oceans. A climatology of CAT based on numerical forecasts is attempted here using a brief survey of the distribution of 6-hour forecasts of CAT made with the Met Office Unified Model This indicator was chosen in preference to others, as it appeared to perform well in the case-studies given above, and also in an earlier comparison of the performance of several CAT indicators made by Bysouth (1998). The results of the survey presented here are summarised in Figs These figures were based on the red (a> 1.5 x 1 O'4 s") and yellow (@>2.5 x s") coloured areas of the original coloured forecast charts. (Figs. 1, 3, 5 and 6 of this paper were also in colour but have been reproduced in black and white.) The red and yellow areas were taken, for colloauial Dur-

7 Fig. 6 Forecast fields of the 250mbar wind vector and (:AT using Brown s indicator in units of 10-5s-, valid at 0600 GM7 from data at 0000 GMTon 12 February E W 9OW Fig. 7 Composite diagram of all patches of moderatelsevere (:AT in the Pacifc hemisphere at 250mbar equatorrvard of about 40 latitude forecast every 84 hours from 0600~M~ on 4 January 1999 to 0600GMTon 2 March Thick dashed line is the great circle arc from Fii; to 1-0s Angeles. See text forjitrther explanation. poses, as indicating the likelihood of moderate and severe turbulence respectively. The lines on the diagrams indicate the axes of the areas of forecast CAT (not the widths to avoid cluttering the diagram). The thick lines correspond to severe CAT, the thin lines to moder- 14

8 30 N 0 30 S 135 E W 90 W Fig. 8 Forecast areas of CAT in the Pacific hemisphere at 250mbar at I-weekb intervakfrom 0600 GMTon 18Januay 1999 to 0600GMTon 21 December Thick dashed line is the great circle arc from F$' to Los Angeles. See text for further explanation. 30N 0 30 S 45w 0 45 E 90E Fig. 9 Forecast areas of CAT in the Afican hemisphere at 250mbar at l-weekly intervals from ~ ~ 18 ~ o n January 1999 w 0600GMTon 21 December See text for further explanation. ate' CAT and the dashed lines to extensions of sharp troughs in the wind field not indicated as turbulent in the forecast. The thick dotted line is the great circle track from Fiji to Los Angeles. Figure 7 shows that in the first two months of 1999 (1 7 forecasts for the Pacific hemi- sphere) there was a heavy concentration of CAT forecasts down to at least 10"N in longitudes 120 W-1800, with a marked decrease to the east and a complete absence to the west. CAT forecast poleward of about 30" latitude is usually associated with the wind-shear zones along the poleward boundaries of the subtrop- 15

9 ical jet streams in both hemispheres. The maximum CAT incidence lies right across the principal air routes from California to AustraliaNew Zealand. In the Southern Hemisphere there is roughly even distribution of forecast CAT extending to about 15"S, and having a much lower frequency than north of the equator. Figure 8 shows that for the whole of 1999 (12 forecasts for the Pacific hemisphere) the 'lobe' of CAT forecasts in the Northern Hemisphere had shifted west to the longitude belt 150"E- 140"W, and does not extend quite so far south as in midwinter. The Southern Hemisphere shows a comparable density of CAT forecasts throughout the year, but these rarely extend north of about 20"s. Figure 9 is as Fig. 8 but for the African hemisphere. This shows that forecasts of CAT in the Northern Hemisphere rarely extend south of about 20"N, but in the Southern Hemisphere the extension north is to about 15"s. This limited investigation suggests that severe CAT may occur equatorward of the mean position of the main subtropical jet streams in both hemispheres at any time of the year, but most often over the tropical Pacific in winter in the zone 10-30"N and 140"W to 180, with Hawaii located near the centre of this zone. During this period, the NHSTJ buckles, and in so doing generates a disrupting trough near the International Date Line at intervals of several days. These troughs develop and then sweep fairly steadily through important tropical air routes during the next week or so, and there should be adequate time for warnings of severe CAT to be issued for aviation. It is well known that over Japan in winter the NHSTJ develops the highest wind speeds (often in excess of 200kn) on this planet, probably because the pole-equator temperature contrast in these longitudes is largely concentrated within N, and the polar front jet stream has merged with the NHSTJ. It is not intuitively surprising that some form of dynamical instability occurs in the diffluent area downstream of Asia where the meridional temperature gradient associated with the NHSTJ diminishes markedly over the Pacific Ocean. The nature of this instability cannot be estab- 16 lished in a paper of this type, but the authors suspect that the local trigger is inertial instability. * Ellrod (1993) published a climatology of a CAT index for the Northern Hemisphere based on a parameter TI (= deformation x vertical wind shear) developed by Ellrod and Knapp (1992). Ellrod highlights seven areas of high TI, including the subtropical east Pacific near Hawaii, particularly in the winter season. This result is consistent with our Figs. 7 and 8. He also comments that "The lower latitude regions of large mean TI values are locations where the subtropical jet stream interacts with the polar jet stream". Conclusions The purpose of this paper is to present preliminary evidence that widespread, severe CAT sometimes occurs in latitudes down to about 10" from the equator. Equatorward of about 40' latitude, this turbulence appears to be associated with (and is probably due to) buckling, then folding of the subtropical jet streams which generates extensive fields of severe CAT along the disrupting trough. The principal site for these disruptions appears to be in the NHSTJ in winter downstream of the (climatological) maximum wind speeds over Japan and south-east China, but may occur in both hemispheres in any season, although more rarely, particularly in the African hemisphere. Once formed, these sharp troughs associated with jet stream disruption appear to move fairly steadily east. The forecasting of the initial buckling, trough formation and subsequent movement appears to be well handled by modern operational numerical forecast models. This paper does not address the question of which CAT indicator is best; CP appears to iden- Areas of slightly negative absolute vorticity (see Fig. 1) had developed to the south of the jet axis just upstream of the growing instability. In simple terms, the jet stream flow had acquired a strong anticylonic curvature. In this situation on a rotating earth, the Coriolis force was balancing the associated centrifugal force in a region where the horizontal pressure gradient had nearly disappeared. Therefore, in nonrotating coordinates, horizontal forces constraining the airflow had vanished, allowing the airflow to become unstable.

10 Weather Vol. 57 Januw 2002 the disrupting trough situation well, but any CAT indicator ought to pick out such an extreme situation. The evidence presented here of the ability of numerical models to forecast such extreme conditions is most important for aviation, as the troughs forming over the tropical North Pacific in winter tend to be aligned with, and to pass across, the main air routes from California to Australia, New Zealand and Japan. It is to be regretted if forecasters rely increasingly on model output without occasional reference to the synoptic background to the production of severe CAT. Study of the synoptic signature associated with severe tropical CAT should be used to develop guides to pilots caught in an area of tropical CATon how to escape from it as soon as possible. Nevertheless, the conclusions of this brief study fall some way short of convincing verification. There are three approaches to this problem: (0 (ii) Collect aircraft reports of severe turbulence over tropical oceans, and compare them with fields of the wind vector and Q, or other CAT indicator as has been done in the three cases reported above. This sample is too small, but there are probably many other similar reports if they can be found. Blanchard (1999) said that after landing, the aircraft captain confirmed that the turbulence had been unusually bad, but was not unusual on this route at that time of year. Analysis of routine Aircraft to Satellite DAta Relay (ASDAR) and Aircraft Meteorological DAta Relay (AMDAR) reports as described in Verification aspects in the Appendix. This method generates large samples for statistical analysis, but caught only one severe CAT (TURB~) report (over the North Atlantic) in several thousand reports. However, if AMDAR reports had been available for the three case-studies presented above, it is possible that the statistical analysis would have rejected all of them as being associated with a high CAT index because the reports were (iii) slightly displaced from areas of high forecast CAT index due to small errors in their forecast location. This demonstrates the value of complementing statistical studies with individual casestudies. Mount aircraft research flights (possibly based from Hawaii) to fly cross-sectional patterns through sharp troughs in the upper windflow while in the vicinity of the research aircraft base. This should establish (if establishment were still needed) that sharp (or disrupting) troughs generated severe CAT. Therefore, a sharp trough will be associated with severe CAT in the same sense that heavy rain and high winds are associated with a humcane. Since numerical forecasts appear to handle sharp troughs, it should be easier to forecast severe CAT. There may also be meteorological interest in studying the incidence of subtropical jet disruption, and its relationship with other meteorological topics, such as El Niiio, estimates of the rates of CATenergy dissipation, etc. Finally, the purpose of this paper will be achieved if it raises awareness in meteorological and aviation circles that severe CAT occurs quite often over tropical oceans, particularly over the north tropical Pacific. Appendix Forecast model Brown s is given by the following equation: where u and w are the components of the horizontal wind in the x and y directions respectively and f is the Coriolis parameter. The terms in brackets on the right-hand side are, respectively: absolute vorticity, shear deformation and stretch deformation. The absolute vorticity is a crude approximation to 17

11 the shear advection term in Roach's original equation, and the horizontal derivatives in CP have been calculated by taking Au, Av, Ax and Ay over a distance of two grid boxes. The factor (A v2/24 in Roach's original expression was omitted as no satisfactory way of deriving it was found. This is to be regretted as its inclusion gives an estimate of the rate of turbulent energy dissipation, which in turn is related to the intensity of turbulence experienced by the aircraft. The data used in this study are from the global version of the Met Office Unified Model (um). This model now has a horizontal resolution of 519" latitude and 516" longitude. The vertical coordinate of the UM is the hybrid eta level. The levels are pure sigma levels near the surface and pure pressure levels above 40 mbar and a hybrid in between. Mathematically, eta = A/105Pa+B where A and B satisfy pressure = A + B x surface pressure in Pascals. Verification aspects Many studies have attempted to compare the skill of CAT predictors by objective verification against aircraft reports. One such study is summarised in Turner and Bysouth (1999). Turbulence reports from commercial aircraft reporting via the World Meteorological Organization ASDAR and AMDAR were used. These turbulence reports are an integer between 0 and 3 based on the vertical acceleration of the aircraft. The reports were filtered to eliminate faulty recorders and reports for which there was cloud forecast. Reports below ft (6096 m) were also eliminated to avoid reports of turbulence during ascent or descent. Thresholds were chosen for each predictor so that CAT was predicted for approximately 1.5% of the reports. Predictors based on horizontal and vertical derivatives of wind showed the greatest skill in the extratropics; CP had the greatest probability of detection in the Northern Hemisphere (5.3%) and Dutton's index (Dutton 1980) showed even greater skill in the Southern Hemisphere. The performance of these indicators in the tropics is poor, with greater skill shown by predictors based on Ri. Comparison with satellite imagery suggests that the skill of the latter is due to prediction of 18 tropical cumulonimbus in locations where the model (used in the filtering) forecasts clear air. Given Fig. 8 and Fig. 9, it is not surprising that predictors such as CP fail to detect many events in the tropics. The above CAT verification procedures are standard practice but, in effect, assess an area forecast by a point method. For example, suppose that scattered showers are forecast for southern England on a given afternoon. This could also be expressed as a probability forecast of, say, 20% that showers will be experienced at any given site. Suppose it is then found (e.g. from a study of rainfall radar echoes) that 18% of the area had experienced showers, then the forecast would be judged as being very accurate as an area forecast, but not much good as a point forecast, since 82% of the area did not experience showers. However, if a belt of continuous rain was forecast to move through the area, and did, then the forecast would be judged as accurate on both a point and an area basis - although, of course, its timing might be in error. Slight to moderate CAT is generally patchy in nature, and is analogous to the shower situation, whereas severe CAT occupies a considerable volume of atmosphere and could scarcely be avoided by an aircraft flying through the volume. This case might be considered as analogous to the continuous rain situation. Acknowledgement The authors are indebted to Mr R. Lunnon for encouraging this study, and for his support, while the study was being executed. References Blanchard, W. (1999) Clear air turbulence. Weather, 54, pp Brown, R. (1973) New indices to locate clear air turbulence. Meteorol. Mag., 102, pp Bysouth, C. E. (1998) A comparison of clear air turbulence predictors. Forecasting Research Technical Report No. 242, Meteorological Office, Bracknell -(2000) Clear air turbulence - a reply. Weather, 55, p. 147 Cullen, M. J. I? (1993) The unified forecastlclimate model. Meteorol. Mag., 122, pp

12 Dutton, J. A. (1980) Probability forecasts of clear-air turbulence based on numerical model output. Meteorol. Mag., 109, pp Ellrod, G. P. (1992) A Northern Hemisphere clear air turbulence climatology. In: Proceedings of 5th International Conference on Aviation Weather Systems, August 2-6, 1993, Vienna, Virginia, USA, pp Ellrod, G. P. and Knapp, D. (1992) An objective clear-air turbulence forecasting technique: verification and operational use. Wea. Forecasting, 7, pp Keller, J. L. (1990) Clear air turbulence as a response to meso- and synoptic-scale dynamic processes. Mon. Wea. Rev., 118, pp Change of political climate Roach, W. T. (1970) On the influence of synoptic development on the production of high level turbulence. Q. J. R. Meteorol. SOC., 97, Turner, J. A. and Bysouth, C. E. (1999) Automated systems for predicting clear air turbulence in global aviation forecasts. In: Proceedings of 8th Conference on Aviation, Range and Aerospace Meteorology, Dallas, Texas, American Meteorological Society Correspondence to: Dr W. T. Roach, 14 Priors Wood, Crowthorne, Berkshire RG45 6BZ. 0 Royal Meteorological Society, C. E. Bysouth s contribution is Crown copyright. 0. M. Ashford Blewbury, Oxfordshire Politicians will never take any action about climate change until meteorologists can tell them what will happen rather than what might happen. This was the advice given in 1977 by a scientist/politician at an informal meeting convened by the World Meteorological Organization (WMO) as part of its preparation for the First World Climate Conference (FWCC). His view was accepted and the Conference was in fact attended almost exclusively by scientists. Less than 20 years later, however, climate change had risen to the top of the political agenda, even though climate predictions were still rather uncertain. Was the advice unsound? Or had circumstances, such as the political climate, changed significantly in the meantime? Before attempting to answer these questions it may be useful to look back briefly to the 1970s and 1980s. Since the publication of Rachel Carson s Silent spring in 1962, there had been growing public concern about man s impact on the environment. Following the Study of critical environmental problems in 1970 by a group of some 70 scientists, the need was felt for an indepth examination of man s impact on climate. The resulting report, Inadvertent climate mod$cabon - known as the SMIC report (SMIC 1972) - highlighted the major and serious gaps in our understanding of the complex systems that determine climate which made it difficult to identify any man-made effect. More research was called for. The SMIC report formed part of the input to the UN Conference on the Human Environment in 1972 which in turn recommended that WMO, in co-operauon With the International Council of Scientific Unions (ICSU), should undertake activities aimed at improving our understanding of the causes of climatic change whether... natural or the result of man s activities (Ashford 1972, p. 208). It had been known for more than a century that carbon dioxide (C0.J released by the combustion of fossil fuels could raise the temperature of the atmosphere by the so-called greenhouse effect. But although a rise of about 0.5degC had been recorded in the fust half of the twentieth century, few scientists paid much attention to this possibility. In the early 1970s many climatologists were becoming alarmed about something quite different. The average surface temperature had in fact decreased by about 0.3degC in the preceding 20 years, and they warned of the likelihood of further cooling in the decades ahead. These 19