Near-surface winds and wind shear at four airports during the St Jude s day storm

Transcription

1 Near-surface winds and wind shear at four airports during the St Jude s day storm 288 M. P. de Villiers 1 and D. White 2 1 Weather Services International Ltd, Birmingham 2 Eurocontrol, Brussels, Belgium Introduction During Monday, 28 October 2013, a small, but deep, low-pressure cell with a centre of ~980hPa (Met Office, 2013a) moved rapidly from west of southwestern England at 0000 UTC to the Baltic Sea 24 hours later, deepening to approximately 970hPa in the process (Met Office, 2013b). During the low s passage across southern England to the North Sea, airport observations in the Meteorological Aviation Report (METAR) format show that when the low-pressure cell passed near Heathrow Airport to the west of London at 0620 UTC, the surface pressure was down to 978hPa, with the mean wind speed around 41kn and gusting to 51kn. Seven aircraft executed missed approaches and no landings occurred for 45 minutes. Such is the pressure of demand on the aviation infrastructure at this airport that even the consecutive loss of a small number of arrival slots can cause a backlog. When the low passed Amsterdam s Schiphol Airport at approximately 0955 UTC, the wind speed reached 46kn, gusting to 60kn, and arrival rate was drastically reduced. Ten hours after passing London it had a wind speed of 43kn, with a 64kn gust, at Copenhagen s Kastrup Airport, where movements were temporarily suspended due to wind-blown debris. It is not generally the practice to name storms that occur in the British Isles, however, the name was popularised in the media when it was brought to their attention that the expected day of the storm would coincide with St Jude s Day: St Jude being the patron saint of desperate cases and lost causes. The strong winds associated with the St Jude s Day storm presented an opportunity to evaluate the 3000ft (900m) wind and wind-shear forecasts of the 12km-resolution Weather Research and Forecasting (WRF) numerical weather prediction (NWP) model that were issued to a client at four major airport hubs: Amsterdam (EHAM), Frankfurt (EDDF), London Heathrow (EGLL) and Paris (LFPG). The WRF model is a software system designed to be used for operational weather forecasting and atmospheric research. Version of the WRF-ARW is used, with an important enhancement provided by the 4km domain, which is nested within the 12km domain, that greatly influences the output for the first 36h. The 12km forecast within the first 36h is an interpolation of the 4km forecast onto the 12km grid. Importance of 3000ft winds Winds at 3000ft are important to air traffic controllers in order for them to assess the impact of wind direction with respect to the prevailing runway direction, ground speed and degree of drift of aircraft as they are sequenced and turned onto the Instrument Landing System and, importantly, once locked onto the glide slope, the ground speed on the descending approach to the runway. The stronger the headwind component, the slower the approach, which results in a reduced landing rate on the runway in question and the introduction of a capacity management measure known as a regulation. For example, a regulation may be instituted at London Heathrow when the headwind or the surface/3000ft wind differential is 15kn or more. Strong winds aloft and weakening winds to ground level can lead to severe wind shear. Wind shear can be defined as the rate of change of the wind in space, considered as a vector (Meteorological Office, 1994). Rate of change can be the result of simultaneous or independent changes in wind speed and direction, vertically or horizontally. Wind shear is particularly noticeable below 2000ft (600m) where frictional drag on the air closest to the Earth s surface causes changes in both wind speed and direction with height (ICAO, 1987). Strong wind shear can be dangerous and has been a cause of accidents (de Villiers, 2009). Aircraft rely on the flow of air over the wings to remain airborne, and stronger wind aloft than nearer to the ground results in decreased air flow over the wings during descent and air-speed reduction. Should the speed of the air over the wing decrease below a critical speed for the aircraft, it will stall, and unless corrected by the pilot, at the final approach and landing stage, this can result in at least a hard landing or at worst a crash landing short of the runway (de Villiers, 2009). Criteria for hazardous low-level wind shear and turbulence can be defined as the mean surface wind being 20kn, while the vector windspeed difference from the surface to 2000ft is 40kn. Other criteria to be considered are thunderstorms within 10km and wind shear previously reported by pilots in the vicinity (Meteorological Office, 1997). In practice, forecast alerts are issued when the 3000ft wind speed is expected to exceed 25kn. Using Heathrow again as an example, aircraft fly an indicated airspeed of 160kn to 4 nautical miles (7.4km) from the runway and then gradually reduce this to landing speeds between 115 and 130kn. At an air speed of 160kn and headwind of 0kn, an aircraft would have an approach ground speed of 160kn, whereas a 25kn headwind would reduce the ground speed to 135kn. If the surface wind speed is 10kn, that is 15kn differential, the landing ground speed would be kn, which would result in fewer aircraft landing per hour, lower capacity and inability to maintain the full schedule, leading to delays, diversions and cancellations. Table 1 provides a guide for when the headwind is 25kn for different wind directions and speeds in relation to the runway headings of 090 and 270 at London Heathrow. Table 1 Wind velocity required for 25kt headwinds at London Heathrow. Wind direction (degrees) Speed ( kn) Infinity

2 Table 2 Forecast wind direction/velocity at 3000ft and surface to 3000ft wind shear (in parentheses) issued at 1217 UTC on 27 October 2013 for the 28th with comments below. Forecast criteria: wind velocity >25kn and/or wind shear >25kn. 27th/1800 UTC 27th/2100 UTC 28th/0000 UTC 28th/0300 UTC 28th/0600 UTC 28th/0900 UTC 28th/1200 UTC EGLL 240/45 (25) 190/35 (20) 190/55 (30 35) 230/65 (35) 270/30 (15 20) 250/30 (15) 250/25 30 (15 20) Expected maximum 3000ft winds between 02 UTC and 04 UTC at /65 70kt (35 40) EHAM 240/50 (30) 240/50 (30) 210/45 (30) 210/50 (40) 210/65-70 (35) 240/65 (30) 240/40 (20) Expected maximum 3000ft period UTC LFPG 240/45 (25 30) 240/60 (35) 210/60 (35) 230/65 (40) 230/60 (35) 240/55 (35) 250/40 (20) A longer expected period of 3000ft winds 60kt from 21 UTC (today) to 06 UTC (28th) EDDF 240/40 (30) 240/45 (25 30) 240/45 (30) 240/50 (30 35) 230/55 (35 40) 230/50 (35 40) 230/50 (30) Expected maximum 3000ft winds between 04 and 09 UTC EGLL, London Heathrow; EHAM, Amsterdam; EDDF, Frankfurt; LFPG, Paris. Model data and analysis Hourly wind-velocity and wind-shear graphics from the 12km-resolution WRF NWP model for 3000ft above mean sea level (MSL) are routinely available for the issue of forecasts. On the morning of the 27th, 6h forecast snapshots were created and saved from the 0000 UTC 12km WRF run, which were valid from 1800 UTC on the 27th to 2400 UTC on the 28th. At the surface, 6h 12km- resolution WRF forecast wind velocities were created and saved. Analyses of the surface air-pressure and surface wind-velocity conditions (in ms 1 ) were made using the NCEP/NCAR reanalysis. Two geopotential-metre height and wind-velocity levels from the NCEP/NCAR reanalysis are available for evaluation of the 3000ft WRF forecast winds (in ms 1 ): 925hPa (~2500ft) and 850hPa (~5000ft). At first glance the 925hPa level appears to be the obvious choice, however, the 2500ft and 5000ft levels are close approximations to actual altitudes under International Civil Aviation Organization (ICAO) standard atmospheric conditions with a sea-level pressure of 1013hPa. During the passage of the storm, sea-level pressure fell to approximately 980hPa, therefore, allowing for an approximate difference of 30ft per hpa, the 2500ft (925hPa) and 5000ft (850hPa) levels were in fact down to around 1500 and 4000ft above MSL, respectively. (For the height difference in feet between 1013 and 980hPa: = 990). Further from the centre of the low, at Paris and Frankfurt, the surface pressure never fell to such an extent, being 999 and 1002hPa at its lowest, respectively, and the altitude error was less, at ft. Furthermore, towards the end of the day at London and Amsterdam the surface pressure had risen to around 1002hPa and the pressure differential decreased to around 330ft, making the 925hPa (~2500ft) level more representative at approximately 2200ft, rather than 4700ft at the 850hPa level. Therefore, the 925 and 850hPa levels have been taken into consideration for evaluation in Tables 3 6 and in Figures 1 4: to limit the number of Table 3 Comparison between observations and reanalysis with model forecasts at London Heathrow. Approximate wind directions in degrees (true north). Wind speeds and windshear values in knots. London Heathrow 0000 UTC 0600 UTC 1200 UTC 1800 UTC METAR MSL pressure METAR surface observation 190/08 210/23G36 240/22 270/13 WRF surface forecast 170/25 260/15 240/20 240/15 NCAR/NCEP 2500ft/5000ft wind 210/47/50 260/40/44 230/40/44 230/42/40 WRF 3000ft forecast 180/50 270/40 240/30 250/45 Surface 2500/5000ft wind shear 39/42 17/21 18/22 29/27 WRF 3000ft wind shear Table 4 As Table 3, but for Amsterdam Schiphol. Amsterdam Schiphol 0000 UTC 0600 UTC 1200 UTC 1800 UTC METAR MSL pressure METAR surface observation 200/16 210/30G44 250/28G39 240/23G34 WRF surface forecast 190/15 190/30 230/20 200/15 NCAR/NCEP 2500ft/5000ft wind 230/42/48 230/46/56 260/44/60 260/42/48 WRF 3000ft forecast 210/50 210/70 230/35 230/45 Surface 2500/5000ft wind shear 26/32 16/26 16/32 19/25 WRF 3000ft wind shear Table 5 As Table 3, but for Paris Charles de Gaulle. Paris Charles de Gaulle 0000 UTC 0600 UTC 1200 UTC 1800 UTC METAR MSL pressure METAR surface observation 210/21G31 200/24G38 240/19 230/16 WRF surface forecast 180/25 190/25 240/20 210/15 NCAR/NCEP 2500ft/5000ft wind 230/48/56 230/48/60 250/36/44 250/42/40 WRF 3000ft forecast 210/55 210/60 230/40 230/45 Surface 2500/5000ft wind shear 27/35 24/36 17/25 26/24 WRF 3000ft wind shear figures only the surface and 850hPa levels have been provided. It should be noted, however, that the evaluation process is subjective and that there is a degree of error. The evaluation times selected were restricted to the most critical 6h forecast times of 0000, 0600, 1200 and 1800 UTC on the 28th, which coincided with the available NCEP/NCAR reanalysis intervals. In addition, hourly METAR observations were used for surface comparison and evaluation. The storm s progress At 0000 UTC on 28 October 2013 the NCEP/ NCAR reanalysis shows surface low pressure below 985hPa west of England and Wales and south of Ireland, with a marked Near-surface winds and wind shear at four airports 289

3 Near-surface winds and wind shear at four airports pressure gradient along the southern periphery and a weaker pressure gradient towards the north, where it was adjacent to a larger low-pressure system (Figure 1). Maximum southwesterly to westsouthwesterly surface winds over 40kn (>20ms 1 in the figure) to the south of the low increased to over 60kn (30ms 1 ) at 850hPa from a similar direction and where the low pressure had evolved into a trough associated with the low to the north. By around 0600 UTC (Figure 2) the low had moved to central eastern England. This is a reasonable fit with the low- pressure cell passing to the north of London nearby Heathrow Airport between 0550 and 0720 UTC, when the pressure in the 0520 UTC METAR was down to its lowest value at Table 6 As Table 3, but for Frankfurt Main. No WRF surface-wind velocities are available at Frankfurt due to the way the data were saved in Figure 5, with no recourse to recreate them. Frankfurt Main 0000 UTC 0600 UTC 1200 UTC 1800 UTC METAR MSL pressure METAR surface observation 190/11 200/14 230/29G46 220/16 WRF surface forecast NCAR/NCEP 2500ft/5000ft wind 230/30/44 230/40/60 240/40/64 260/30/40 WRF 3000ft forecast 230/40 210/45 210/50 240/40 Surface 2500/5000ft wind shear 19/33 26/46 11/35 14/24 WRF 3000ft wind shear hPa. In the reanalysis the westsouthwesterly surface wind in the London area was ~28kn (~14ms 1 ). At Heathrow the observed METAR surface wind at 0550 UTC was 23kn at 210, gusting 36kn, which is a good match. According to the METAR, at 0620 UTC the 250 surface wind peaked at 41kn, gusting 51kn, and by 0720 UTC it was at 260 and down to 29kn, gusting 39kn. Meanwhile, at 850hPa in the reanalysis a wind of ~44kn (~22ms 1 ) was present in the London area and the peaking strong winds were found to coincide with the missed approaches and temporary suspension of landing at Heathrow. At 1200 UTC (Figure 3) a maximum strength westerly wind of about 32kn was found to be present over the southern North Sea. This compares favourably with the 1155 UTC METAR at Amsterdam s Schiphol Airport, which reported a 250 wind at 28kn, gusting 39kn. Earlier, at 0955 UTC, the wind was at 230 with a speed of 46kn, gusting 60kn. These strong winds coincided with hourly arrival rates of at Schiphol. At 850hPa westerly winds >60kn (>30ms 1 ) extended (a) 290 Figure 1. NCEP/NCAR reanalysis of (a,b) 850hPa height (geopotential metres) and wind speed (ms 1 ) and (c,d) mean sea-level pressure (Pa) and mean wind speed (ms 1 ) at 0000 UTC on 28 October 2013.

4 (a) Near-surface winds and wind shear at four airports Figure 2. As Figure 1, but at 0600 UTC on 28 October over The Netherlands and across northern Germany to northwestern Poland. By 1800 UTC (Figure 4), the reanalysis places the low over southern Sweden, with ~32 36kn (16 18ms 1 ) surface westerly winds across Denmark and on into the southern Baltic Sea. Once again the surface wind compares favourably with the mean wind speed of 37kn (48kn gust) in the 1750 UTC METAR at Copenhagen s Kastrup Airport. Earlier, at 1620 UTC, the observed (METAR) wind had been 43kn with a 64kn gust. At 850hPa the wind was approximately 46kn (23ms 1 ) over Denmark to >60kn (>30ms 1 ) over northern Poland. Wind forecasts and regulation Advance warning of impending strong wind was brought to the attention of Eurocontrol Network Managers in a separate from the routine 5-day forecast on the 22nd, which read At this time, the 28th looks like being another day with 30 50kn winds over Britain and later the NW part of mainland Europe due to a low of about 962hPa west of Ireland. Strong to gale force at the surface as well, and this was changed to gales and strong gales at the surface on the 23rd. The forecast was based on longer range information from the NWP models of the 1 NCEP Global Forecast System (not shown). Subsequent 5-day forecasts continued to draw attention to gale-force winds until the day prior to the expected disruption, when the required detailed forecast was issued at 1217 UTC on the 27th for the designated major airports (Table 2). The text forecast was based on 6h post-processed diagram products derived from the 0000 UTC run of the 12km-resolution WRF NWP model (28/0000 to 28/1800 UTC; Figure 5), which was sent to Eurocontrol in a separate . Further figures were created at 0200, 0400, 0800, 1000 and 2200 UTC (not shown). The forecast severity of the winds enabled air-traffic control and airport operators to assess the potential impact and to plan mitigation strategies, including the prior-warning regulation for strong winds, which were applied at the airports listed in Table 3. Early cancellation of regulations occurred at Gatwick and London City because of earlier than anticipated easing of the wind, but the longer application period required at Heathrow can be attributed to continued 25kn headwind on the approach and until demand decreased sufficiently for it to be cancelled. In spite of strong winds there was no traffic disruption at Frankfurt and Paris: four runways aligned east west at Paris and three runways aligned east west at Frankfurt allowed these airports to cope with the excessive winds from a southwesterly direction and thus avoid the implementation of regulations. Many airfields were only slightly affected and the management of those airfields that were more severely affected was very effective: it was only when the high level of gusts appeared that the managementof-demand plan was tested to its limits. From 0505 to 0636 UTC, seven flights into Heathrow executed missed approaches due 291

5 Near-surface winds and wind shear at four airports to wind shear. At about 0620 UTC a gust of 64kn was recorded and because of the wind conditions no landings occurred from 0640 to 0725 UTC, at which time a flight landed in a 250 wind at 25kn, gusting 45kn. A regulation for wind (Table 4) was applied at 0600 UTC for a capacity rate of 28, raised at 0830 UTC to 32 and at 0915 UTC to 40, with cancellation at 1637 UTC, once the residual traffic had been absorbed: the usual arrival capacity is 45. The arrival rate at Schiphol at 0600 UTC was 28, varying between 30, 15 and 40 until 1330 UTC, when it was raised throughout the afternoon to 65. Kastrup planned in advance to reduce their normal arrival rate from 42 to 30 from 1640 UTC due to the forecast severe winds. At 1800 UTC the rate dropped to zero because of unexpected severe gusts that caused debris to be blown about the airport. Despite the severe wind conditions air- traffic control could have delivered 30 aircraft to the runway as the wind was aligned with the runway direction. At 1930 UTC the rate was raised again to 30. As a result of the sudden unavailability of the airport 23 flights en route to Copenhagen diverted to other airports, and three aircraft taxiing at other airports bound for Copenhagen were delayed. These capacity-management figures do not include that made available for flights entering the European Civil Aviation Conference (ECAC) region from other parts of the world. Forecast evaluation Visual inspection of the forecast parameters for surface and upper air produced by various models showed that they were in general agreement with the reanalysis with respect to the evolving wind changes. The different formats, however, made an effective evaluation difficult, and it was therefore decided to make subjective spot readings of wind velocity and wind shear at the four airports: at the surface and 3000ft and the surface and 2500ft and 5000ft. Vector differences due to differing wind direction were ignored when estimating wind shear. The results are presented in the comparison Tables 3 6. Unfortunately, WRF forecast surface winds are not available at Frankfurt due to the format in which the post-processed graphics were saved, with no recourse to recreation (Figure 5 and Table 6). Nevertheless, in spite of aforementioned drawbacks and discrepancies, the tables do confirm the initial impression that the models gave a good indication of the impending strong winds and significant wind shear, although there were notable exceptions. At London (Table 3) the 0000 UTC surfacewind observation of 8kn was much lower than the WRF forecast, this resulted in excessive wind-shear calculations at 2500ft and 5000ft. In fairness the observed surface wind did increase to 17kn, gusting 28kn, at 0220 UTC, so the WRF model was not too inaccurate. The WRF forecast of the surface wind direction at 0600 UTC was also premature in forecasting veering of the wind to westerly behind the passing low, however, once again, this was not too innaccurate, because at 0620 UTC the METAR wind was 250 at 41kn, gusting 51kn, and at 0650 UTC (a) 292 Figure 3. As Figure 1, but at 1200 UTC on 28 October 2013.

6 (a) Near-surface winds and wind shear at four airports Figure 4. As Figure 1, but at 1800 UTC on 28 October Table 7 Regulation application (date and time), duration and cancellation where known (times UTC). Frankfurt cancelled before implementation. Airport Activation Regulation London Heathrow 27/ cancelled at 1637 Amsterdam Schiphol 27/ Amsterdam Schiphol 28/ London Gatwick 27/ cancelled 1019 London City 27/ cancelled 0913 Brussels National 27/ cancelled 0834 Copenhagen Kastrup 28/ Frankfurt Main 28/ cancelled 1741 it was 270 at 34kn, gusting 45kn, although the forecast speed was too low at 15kn. All in all there was good agreement between WRF forecasts of wind velocity and windshear and the observed/analysed values. It is worth noting that during the period of strongest winds at London, runway 27L (left) was used for landing and 27R (right) for take-off, prior to switching runways at 1400 UTC, until the wind eased. This avoided known increased turbulence in strong winds on short finals to 27R and the landing point on 27R due to the location of terminal buildings and maintenance hangers at the eastern end between the two runways 27R and 27L. There was also good agreement between the observations at Amsterdam and the model data (Table 4). Although there were exceptions around the 3000ft level, most notably the 3000ft wind of 70kn by the 0600 UTC WRF was considerably stronger than the reanalysis values of 46 and 54kn, and the 60kn value of the 1200 UTC 5000ft reanalysis appears excessive. Another exception is the wind-shear value at 1800 UTC, which could be anywhere between 20 and 25kn due to difficulty in visually calculating the wind shear in Figure 5. There was similar good agreement at Paris and Frankfurt (Tables 5 and 6), although both the NCEP/NCAR 0600 and 1200 UTC 5000ft wind analyses of 60 and 64kn, in Table 6, are too high, which markedly increases the wind shear, or the WRF values of 45 and 50kn are too low. Taking into consideration the information in Table 3, Figure 5 and the METARs, the main effect of the winds at the surface and 3000ft at London was experienced in the early hours to around 0900 UTC, especially UTC. This was slightly longer than the suggested time of UTC in Table 2, which was based on the intermedi- 293

7 Near-surface winds and wind shear at four airports ate forecast snapshots of the WRF model at 0200 and 0400 UTC (not shown). Similarly, at Amsterdam and using Table 4, the period of maximum effect was UTC, which is again longer than suggested by the WRF model data and indicated in Table 2, as well as the wind being stronger and more sustained than at London. Conclusions A small but deep low passed from southern England, across Belgium, The Netherlands and Denmark on its way to the Baltic Sea, (a) causing air traffic disruption at the main airports in these countries. Of the four major airports in this study, operations at Paris and Frankfurt were not disrupted. Although the winds eased considerably from about 0800 UTC at London Heathrow, the period of regulation was prolonged and can be attributed to >25kn headwinds aloft and capacity limitation. During the worst wind conditions, numerous missed approaches occurred and landing was suspended for about 50min. London was also fortunate in that the strongest winds passed to the south across the English Channel to Amsterdam, which experienced the stronger winds for a longer period. NWP models gave sufficient early indication of the storm and strong winds for a preliminary warning to be issued 6 days prior to the event, which was maintained, in finer detail, until 24h before the storm. At this time the finer resolution data from the NWP model of the 12km WRF instilled sufficient confidence for a significantly more accurate forecast to be issued. This provided air-traffic control and airport authorities with a level of accuracy to assess the impact and, in advance, plan traffic management strategies to shape demand to the ability of the aviation infrastructure to cope with the severe-weather conditions. The NCEP/NCAR reanalysis was shown to be an effective evaluation tool in spite of the differences between the 925hPa (2500ft) and 850hPa (5000ft) wind levels and the 3000ft WRF level. Acknowledgements Thank you to Weather Services International for use of their computer facilities; to Leon Brown and Martin Leeson for the WRF 3000ft wind and wind-shear graphics; to the anonymous reviewers for their useful comments. A special thank you to Eurocontrol for the valuable air-traffic information provided. References De Villiers MP Aviation Meteorological Forecasts and Warnings: An Airport Management Tool. Nova Science Publishers, Inc.: New York, NY. Internat ional Civil Aviation Organization (ICAO) Wind Shear, ICAO Circular 186-AN/122. International Civil Aviation Organization: Montreal, Canada, p 11. Meteorol ogical Office Handbook of Aviation Meteorology. Her Majesty s Stationary Office: London, p 77. Meteorol ogical Office Source Book to the Forecasters Reference Book, Meteorological Office Meteorological Office College: Reading, UK, pp 5. Met Of fi ce. 2013a. Surface analysis. uk/public/weather/surfacepressure/#?tab=surfacepressurebw (accessed 28 October 2013). Met Offi ce. 2013b. Surface analysis. ww.metoffice.gov. uk/public/weather/surfacepressure/#?tab=surfacepressurebw (accessed 29 October 2013). 294 Figure 5. Surface winds (left), 3000ft winds and wind shear (right) at (a) 0000, 0600, 1200 and 1800 UTC (top to bottom) on 28 October 2013 from the 12km resolution WRF NWP model 0000 UTC on 27 October Full feather 10kn and half feather 5kn. Correspondence to: M. P. de Villiers mdevilliers@wsi.com 2014 Royal Meteorological Society doi: /wea.2338