Convective Storm Initiation Project Research Articles

A simple ensemble approach for more robust process‐based sensitivity analysis of case studies in convection‐permitting models

Case studies remain an important method for meteorological parameter sensitivity process studies. However, these types of study often use just a few case studies (typically up to three) and are not tested for statistical significance. This approach can be problematic at the convective scales, since uncertainty in the representation of an event increases, and the variability in the atmosphere arising from convective‐scale noise is not routinely taken into account. Here we propose a simple ensemble method for performing more robust sensitivity analysis without the need for an operational‐style ensemble prediction system and demonstrate it using a case study from the 2005 Convective Storm Initiation Project. Boundary‐layer stochastic potential temperature perturbations with Gaussian spatial structure are used to create small ensembles to examine the impact of increasing cloud droplet number concentration (CDNC) on precipitation. Whilst there is a systematic difference between the experiments, such that increasing the CDNC reduces the precipitation, there is also an overlap between the different ensembles implying that convective‐scale variability should be taken into account in case study process‐based sensitivity studies.

Comparison of Convective Boundary Layer Characteristics from Aircraft and Wind Lidar Observations

AbstractDuring the Convective Storm Initiation Project experiment, which was conducted in summer 2005 in southern England, vertical velocity in the convective boundary layer (CBL) was measured simultaneously with a research aircraft and a wind lidar. The aircraft performed horizontal flight legs approximately parallel to the prevailing wind direction and centered over the lidar. This measurement setup allows for the comparing of CBL characteristics (CBL depth zi, integral length scale lw, spectral peak wavelength λm, and vertical velocity variance ) from temporal (lidar) and spatial (aircraft) measurements. For this, the lidar time series are transferred into space using the mean wind. While the statistics of the aircraft data are all based on the 34-km flight legs, the averaging interval for the lidar is either 1 h or a longer period that corresponds to the 34-km leg. Although the lw and λm values from aircraft and lidar measurements are in the same range (100–200 and 500–2000 m) and agree well on the average, the correlation for individual legs is very low (R2 < 0.17). One possible explanation is the large uncertainty that arises from the transfer of the lidar time series to space. For , the agreement between aircraft and lidar is better for individual legs (R2 ≥ 0.63), but the mean absolute difference in is about 2.5 times as large as the statistical error. We examine the nonstationarity and heterogeneity for the lidar and aircraft samples and can exclude these as the major sources for the large differences between lidar and aircraft data.

Simulations of an observed elevated mesoscale convective system over southern England during CSIP IOP 3

Simulations of an elevated mesoscale convective system (MCS) observed over southern England during the Convective Storm Initiation Project (CSIP) provide the first detailed modelling study of a case of elevated convection occurring in the UK. The study shows that many factors can influence the maintenance of elevated deep convection, from large‐scale flow through to surface heating processes and diabatic cooling within the convective system. It is also shown that interactions and feedback mechanisms between a stable layer and the storm can act to maintain deep convection. The simulation successfully reproduced an elevated MCS above a low‐level stable undercurrent, with a wave in the undercurrent linked to a rear‐inflow jet (RIJ). Convection was fed from an elevated (840 hPa) source layer with CAPE of about 350 J kg−1. The undercurrent in the simulation was approximately 1 km deep, about half that observed. Unlike the observed MCS, a transition from elevated to surface‐based convection occurred in the simulation due to the combined effects of a pre‐existing large‐scale θe gradient, advection and surface heating causing the system to encounter increasingly unstable low‐level air and a shallower stable layer that was more susceptible to penetration by downdraughts. The transition to surface‐based convection was accompanied by the development of cold‐pool outflow and an increase in system velocity from about 6 to 10 m s−1. Diabatic cooling from microphysical processes in the simulation enhanced the undercurrent and strengthened the RIJ. This strengthened the wave in the undercurrent and led to more extensive convection. The existence of a positive feedback process between the convection, RIJ and stable layer is discussed. Uncertainty in the synoptic scale generating errors in the undercurrent is shown to be a major source of error for convective‐scale forecasts.

High‐resolution observations of precipitation from cumulonimbus clouds

ABSTRACTMaking high‐resolution observations of heavy rainfall from cumulonimbus clouds involves choosing a good location where convection occurs frequently and with some degree of predictability so that aircraft do not have to chase clouds, and radars and other ground‐based instruments can be optimally placed. It also requires significant resources, international collaboration and a good deal of serendipity. This paper describes several observational campaigns that were designed to understand relatively small‐scale convective clouds that produce heavy precipitation, and presents results that have been found over the years. It complements the companion papers in this special issue by giving an observational perspective. It is by no means comprehensive and is limited to cumulonimbus clouds. The Convective Storm Initiation Project (CSIP) was designed to understand the phenomena responsible for the initiation of convection in the maritime environment of southern England. There was a similar objective in the Convective Orographically‐induced Precipitation Study (COPS), but for the complex terrain of the Black Forest mountains, Germany, and the Vosges Mountains, France. The recent Convective Precipitation Experiment (COPE) was designed to understand the entire life cycle of convective clouds that formed along sea‐breeze convergence lines in the southwestern peninsula of England. Its particular focus was on the dynamics and microphysics of the convective clouds. Examples of observations from these and other field campaigns designed to study the initiation of convection, cumulonimbus clouds and convective precipitation are presented, with a view to demonstrating how they contribute to a physical understanding of the phenomena and the processes involved.

The evolution of an MCS over southern England. Part 2: Model simulations and sensitivity to microphysics

AbstractSimulations using the Met Office Unified Model at 1 km horizontal grid spacing of a Mesoscale Convective System (MCS) with a cold pool which propagated across southern England on 25 August 2005 are validated using detailed observations from the Convective Storm Initiation Project (CSIP). Early organisation of the system is not especially well treated, but the model goes on to form a system which developed qualitative and quantitative features remarkably similar to the observations.A sensitivity study suggests that the initial linear system is driven by the position of a low‐level ‘lid’ and upper‐level instability, the linear organisation being promoted by a weak rear‐inflow jet forced by the upper‐level warm anomaly in the cloud anvil. A weak cold pool develops in the absence of ice‐phase processes, but this does not promote system propagation. Strengthening and descent of the rear‐inflow jet, and acceleration of the system, is promoted by the additional heating through glaciation and cooling through snow evaporation. The surface cold pool and gust front are further strengthened by snow melting and rainfall evaporation. With ice‐phase processes present, the cold pool strengthens as a result of the system development and its strength is broadly correlated with system propagation speed during the middle phase of the system's lifetime. Propagation enables the convective band to ‘sweep up’ any convective cells which trigger ahead of the system.The differing scales of microphysical processes means that it is difficult to form a steady‐state system. The observed transition phase corresponds to an increase in the slantwise nature of the flow in the storm and in the low‐level cooling by rainfall evaporation near the gust front, combined with a change in ambient conditions (advection over the sea) which eventually enable the cold pool to propagate ahead of the system.

The evolution of an MCS over southern England. Part 1: Observations

AbstractObservations are described of a Mesoscale Convective System (MCS) with a cold pool which propagated across southern England on 25 August 2005. The observations were made as part of the Convective Storm Initiation Project (CSIP).The observed MCS structure broadly followed well‐established theories, including the presence of a weak rear‐inflow jet. In detail, however, unsteady transitions occurred involving the formation of two distinct lines of showers ahead of the initial linear system. In each case the cold pool merged with cold downdraughts from the new showers leading to a discontinuous propagation of the system. One of these lines formed independently of the MCS, very probably on a sea‐breeze convergence line. The mechanism for formation of the other is unknown, but it is possible that it was triggered by ascent associated with gravity waves generated by the MCS. The merged cold pool was deeper and colder and propagated faster than the original system, eventually forming a bow echo and arc cloud as it propagated across the English Channel. Until completion of the merger, the propagation velocity of the overall system had been controlled by a combination of the above mechanisms rather than simply by cold pool dynamics.

High-resolution representation of the mechanisms responsible for the initiation of isolated thunderstorms over flat and complex terrains: analysis of CSIP and COPS cases

Two important measurement campaigns took place in Europe in the last years, the Convective Storm Initiation Project (CSIP) and the Convective and Orographically-induced Precipitation Study (COPS) to gain a better understanding of why deep convection develops. In both campaigns, a dense network of instruments was deployed for detailed observation of the boundary layer characteristics. This paper investigates the usefulness of the combination of high-resolution surface, GPS, and radiosonde measurements to ascertain the likelihood of deep convection in particular for the complex terrain of the COPS domain. Two convection episodes were analysed for this purpose, one from the CSIP campaign and one from the COPS experiment. This study shows that despite the high spatial resolution of the radiosonde network in comparison to current observations, it was necessary to ascertain higher-resolution data sets illustrating the spatial variability of humidity, atmospheric stability, and convective inhibition. GPS and radiosonde data were used to determine high-resolution humidity fields, yielding high-resolution convection-related parameters’ fields representing atmospheric instability and inhibition in the area. The surface data provided a high-resolution representation of the near-surface convergence zones, which acted as triggering mechanism. The use of GPS measurements and surface stations, in addition to radiosonde measurements, turned out to be especially helpful to determine the regions of increased likelihood of deep convection.

A case study of a large patch of billows surmounted by elevated convection

AbstractA region of large‐amplitude Kelvin‐Helmholtz billows, of area 2000 km2, was observed in a region of moderate rain at the rear of a mesoscale convective system on a day during the Convective Storm Initiation Project in southern England. The mesoscale convective system (MCS) was characterized by elevated convection above a layer of cool air, separated by a sloping statically stable layer characterized by strong wind shear. Doppler radar observations showed that the large patch of billows was situated within this shear layer and that it persisted over the 2 h period of observation, maintaining its position with respect to the MCS and producing surface pressure perturbations of ±0.3 hPa. Potentially unstable air above the sloping shear layer reached its level of free convection while ascending above this layer. The resulting elevated convection was in the form of cells whose spacing appeared to be influenced by the underlying billows. Results from large‐eddy model simulations are shown to support the hypothesis that, although the large‐scale ascent alone would have triggered the elevated convection, the billows exerted a secondary forcing effect on the convection. Copyright © 2012 Royal Meteorological Society

Lidar observations of flow variability over complex terrain

AbstractThis paper examines one way in which remote sensing instrumentation can be used to advance our understanding of the interactions between complex terrain and the atmospheric boundary layer (ABL). When mean flow speed is of moderate strength and the ABL is stable, mechanical effects will dominate thermal effects in modifying flow speed and direction. Boundary layer measurements were made using the scanning Salford 10 µm pulsed CO2 Doppler lidar during the 2005 Convective Storm Initiation Project (CSIP), above the heterogeneous orography that surrounds Faccombe, Hampshire, UK. A new method of detecting boundary layer flow perturbations was developed and successfully applied to the lidar data, giving a clearer insight into flow modification that occurs above complex terrain. Copyright © 2011 Royal Meteorological Society

Multi-sensor observations of a wave beneath an impacting rear-inflow jet in an elevated mesoscale convective system

Observations of a wave generated below, and propagating with, an elevated mesoscale convective system (MCS) are presented. The MCS and the wave tracked directly across the 90 km extent of the observational network deployed for the Convective Storm Initiation Project (CSIP) and maintained a coherent structure and velocity during this 1.5 hour period. The rear-inflow jet of the MCS was observed to impact upon, but not penetrate, the stable undercurrent that was located beneath the MCS, apparently constricting the depth of the undercurrent, which did not participate in the main convective updraught. The wave crest was located approximately 15 km ahead of this constriction. The overall wave led to between 200 and 1000 m of ascent and approximately 1500 m of descent. However, undulations were superimposed on the overall wave, with a wavelength of approximately 7 km and a crest-totrough amplitude of approximately 1000 m. These undulations were associated with double maxima in surface pressure and the elevation of the observed low-level wind anomaly, where low-level winds were backed by more than 90 ◦ .T hese winds were never at rest relative to the system and the wave was therefore a ‘gravity wave without stagnation’. The hydrostatic effect expected from the observed lifting in the wave was broadly consistent with the observed surface pressure perturbation of 2 hPa. No cold-pool outflow was ever observed; instead the observed surface cooling of approximately 4 ◦ C was shown to be consistent with reduced solar insolation. The generation of the observed wave structure and its implications for the maintenance of the MCS and numerical modelling of the system are discussed. Copyright c

Observations of dual slantwise circulations above a cool undercurrent in a mesoscale convective system

AbstractA series of mesoscale convective systems (MCSs) was observed in the UK during one of the Intensive Observation Periods (IOP 3) in the Convective Storm Initiation Project (CSIP). A detailed case‐study of one of the MCSs, involving some novel interpretation of single Doppler radar data, reveals the evolution of stacked slantwise circulations associated with elevated upright convection. A feature of this occasion was the existence near the surface of a strong undercurrent of cool air flowing against the direction of travel of the storm. The air feeding both the upright and slantwise ascent originated not from near the surface but, rather, from layers between 1 and 3 km. The upright convection was intense at first and was accompanied by a region of dry subsidence surrounding it, together with the alternating layers of slantwise ascent and descent within the storm's precipitation area. When the elevated upright convection weakened, vigorous slantwise motions continued for some time, but their slope decreased from the initial 1 in 4 to 1 in 9. The lowermost layer of slantwise descent corresponded to a moderately intense rear‐inflow jet. The rear‐inflow jet did not penetrate the cool undercurrent and reach the surface; instead, beneath where it impacted the warm air capping the undercurrent, the undercurrent took on the structure of a gravity wave without stagnation. The raised head of the undercurrent lifted overlying air of high wet‐bulb potential temperature sufficiently to overcome the convective inhibition. Copyright © 2010 Royal Meteorological Society

Impact of Data Assimilation on Forecasting Convection over the United Kingdom Using a High-Resolution Version of the Met Office Unified Model

Abstract A high-resolution data assimilation system has been implemented and tested within a 4-km grid length version of the Met Office Unified Model (UM). A variational analysis scheme is used to correct larger scales using conventional observation types. The system uses two nudging procedures to assimilate high-resolution information: radar-derived surface precipitation rates are assimilated via latent heat nudging (LHN), while cloud nudging (CN) is used to assimilate moisture fields derived from satellite, radar, and surface observations. The data assimilation scheme was tested on five convection-dominated case studies from the Convective Storm Initiation Project (CSIP). Model skill was assessed statistically using radar-derived surface-precipitation hourly accumulations via a scale-dependent verification scheme. Data assimilation is shown to have a dramatic impact on skill during both the assimilation and subsequent forecast periods on nowcasting time scales. The resulting forecasts are also shown to be much more skillful than those produced using either a 12-km grid length version of the UM with data assimilation in place, or a 4-km grid-length UM version run using a 12-km state as initial conditions. The individual contribution to overall skill attributable to each data-assimilation component is investigated. Up until T + 3 h, LHN has the greatest impact on skill. For later times, VAR, LHN, and CN contribute equally to the skill in predicting the spatial distribution of the heaviest rainfall; while VAR alone accounts for the skill in depicting distributions corresponding to lower accumulation thresholds. The 4-km forecasts tend to overpredict both the intensity and areal coverage of storms. While it is likely that the intensity bias is partially attributable to model error, both VAR and LHN clearly contribute to the overestimation of the areal extent. Future developments that may mitigate this problem are suggested.

Convection forced by a descending dry layer and low-level moist convergence

A narrow line of convective showers was observed over southern England on 18 July 2005 during the Convective Storm Initiation Project (CSIP). The showers formed behind a cold front (CF), beneath two apparently descending dry layers (i.e. sloping so that they descended relative to the instruments observing them). The lowermost dry layer was associated with a tropopause fold from a depression, which formed 2 d earlier from a breaking Rossby wave, located northwest of the UK. The uppermost dry layer had fragmented from the original streamer due to rotation around the depression (This rotation was also responsible for the observations of apparent descent—ascent would otherwise be seen behind a CF). The lowermost dry layer descended over the UK and overran higher θw air beneath it, resulting in potential instability. Combined with a surface convergence line (which triggered the convection but had less impact on the convective available potential energy than the potential instability), convection was forced up to 5.5 km where the uppermost dry layer capped it. The period when convection was possible was very short, thus explaining the narrowness of the shower band. Convective Storm Initiation Project observations and model data are presented to illustrate the unique processes in this case.

Observations of the development of convection through a series of stable layers during the Convective Storm Initiation Project

AbstractThis article describes the development of convection through a series of stable layers on 20 July 2004 during the pilot project of the Convective Storm Initiation Project (CSIP). Convective clouds developed in south central England as a result of diurnal heating. They formed near the south coast and grew gradually as they moved northwards, eventually producing light precipitation. Several stable layers and associated dry layers (or lids) were observed by sequential radiosondes launched every two hours. Most of these lids were also detected by the Chilbolton L‐band radar and with a wind profiler as prominent layers of Bragg echo. One particular lid at 820 mb (1.7 km) inhibited the early development of the convection for about an hour, delaying the production of precipitation. The uppermost lid at 650 mb stopped all convection from intensifying in the CSIP domain.Estimates of the growth of convection through the lids were sensitive to the values of temperature and humidity chosen to represent the ascending parcels. The development of the thermals in the boundary layer was best explained by using the surface‐layer average values. Parcels would not have been sufficiently buoyant to reach the lifting condensation level (LCL) unless these values were used. However, such parcels in the absence of mixing would then have been too buoyant to be capped by the main lid at 1.7 km, as was observed by the radars. In fact the development of the cumulus clouds above the LCL is best represented by using 950‐‐1000‐mb average thermodynamic values.The results highlight the importance of predicting the altitude, strength and persistence of such lids in order to forecast the intensity, location and timing of convective precipitation more accurately. Copyright © 2008 Royal Meteorological Society

Application of the Aventech AIMMS20AQ airborne probe for turbulence measurements during the Convective Storm Initiation Project

Abstract. The Convective Storm Initiation Project (CSIP) took place during the summers of 2004 and 2005, centred on the research radar at Chilbolton, UK. Precursors to convective precipitation were studied, using a comprehensive and broad-based range of fieldwork and modelling. The principal aim of CSIP was the detection of the primary and secondary initiation of convective cells. The Universities Facility for Atmospheric Measurements (UFAM) Cessna 182 was used to map temperature and humidity fields over a broad area within and beyond the Chilbolton radar beam. Additionally, air motion was measured using a new turbulence probe, the AIMMS20AQ. The performance of the probe is critically appraised, based on calibrations, test flights and data flights flown during CSIP intensive operating periods. In general, the probe performed well, although some aspects require more careful data interpretation which we describe in detail.

Convective inhibition beneath an upper‐level PV anomaly

AbstractUpper‐level potential‐vorticity (PV) anomalies reduce the convective stability of the troposphere through their impact on the vertical potential‐temperature profile, thus reducing convective inhibition (CIN) and increasing convective available potential energy. Here, by contrast, we show the impact of a layer of stable air that was intrinsically linked with an upper‐level PV anomaly and that increased CIN. This layer descended and tracked beneath the small upper‐level PV anomaly, which in this case was a shallow upper‐level trough. This low‐humidity, relatively high‐PV layer originated from the tropopause fold, generated by a breaking Rossby wave, which also produced the upper‐level PV anomaly two days later. Despite conditions favourable for deep convection (as demonstrated by the development of a single storm), the CIN produced by this dry layer or lid was largely responsible for capping convection over much of southern England at around 2.5 km during the case presented here, which comes from the Convective Storm Initiation Project. Copyright © 2008 Royal Meteorological Society

The Convective Storm Initiation Project

The Convective Storm Initiation Project (CSIP) is an international project to understand precisely where, when, and how convective clouds form and develop into showers in the mainly maritime environment of southern England. A major aim of CSIP is to compare the results of the very high resolution Met Office weather forecasting model with detailed observations of the early stages of convective clouds and to use the newly gained understanding to improve the predictions of the model. A large array of ground-based instruments plus two instrumented aircraft, from the U.K. National Centre for Atmospheric Science (NCAS) and the German Institute for Meteorology and Climate Research (IMK), Karlsruhe, were deployed in southern England, over an area centered on the meteorological radars at Chilbolton, during the summers of 2004 and 2005. In addition to a variety of ground-based remote-sensing instruments, numerous rawinsondes were released at one- to two-hourly intervals from six closely spaced sites. The Met Office weather radar network and Meteosat satellite imagery were used to provide context for the observations made by the instruments deployed during CSIP. This article presents an overview of the CSIP field campaign and examples from CSIP of the types of convective initiation phenomena that are typical in the United Kingdom. It shows the way in which certain kinds of observational data are able to reveal these phenomena and gives an explanation of how the analyses of data from the field campaign will be used in the development of an improved very high resolution NWP model for operational use.

Combination of Mesoscale and Synoptic Mechanisms for Triggering an Isolated Thunderstorm: Observational Case Study of CSIP IOP 1

Abstract An isolated thunderstorm formed in the southern United Kingdom on 15 June 2005 and moved through the area where a large number of observational instruments were deployed as part of the Convective Storm Initiation Project. Earlier, a convergence line had formed downstream of Devon in the southwest of the United Kingdom in a southwesterly airflow, along which a series of light showers had formed. The depth of these showers was limited by a capping inversion, or lid, at around 2.5 km. The deep thunderstorm convection developed from one of these showers when the convection broke through the lid and ascended up to the next inversion, associated with a tropopause fold at around 6 km. A series of clear-air reflectivity RHIs are used to map the height of the capping inversion and its lifting resulting from the ascent along the convergence line. The origins of the lid are tracked back to some descent from the midtroposphere along dry adiabats. The strength of the lid was weaker along a northwest-to-southeast-oriented region located behind an overrunning upper cold front. The transition from shallow to deep convection occurred where this region with a weaker lid intersected the region with a raised lid, oriented southwest to northeast, downstream of Devon. A very high resolution forecast model that is being developed by the Met Office predicted the isolated thunderstorm successfully. This success depended on the accurate representation of the following two scales: the synoptic-scale and the surface-forced mesoscale convergence line. The interaction between these scales localized the convection sufficiently in space and time for the initiation and subsequent development to be highly predictable despite the relatively poor representation in the model of processes at the cloud scale.

Variable cirrus shading during CSIP IOP 5. I: Effects on the initiation of convection

AbstractObservations from the Convective Storm Initiation Project (CSIP) show that on 29 June 2005 (Intensive Observation Period 5) cirrus patches left over from previous thunderstorms reduced surface sensible and latent heat fluxes in the CSIP area. Large‐eddy model (LEM) simulations, using moving positive surface‐flux anomalies, show that we expect the observed moving gaps in the cirrus cover to significantly aid convective initiation. In these simulations, the timing of the CI is largely determined by the amount of heat added to the boundary layer, but weak convergence at the rear edge of the moving anomalies is also significant.Meteosat and rain‐radar data are combined to determine the position of convective initiation for all 25 trackable showers in the CSIP area. The results are consistent with the LEM simulations, with showers initiating at the rear edge of gaps, at the leading edge of the anvil, or in clear skies, in all but one of the cases. The initiation occurs in relatively clear skies in all but two of the cases, with the exceptions probably linked to orographic effects.For numerical weather prediction, the case highlights the importance of predicting and assimilating cloud cover. The results show that in the absence of stronger forcings, weak forcings, such as from the observed cirrus shading, can determine the precise location and timing of convective initiation. In such cases, since the effects of shading by cirrus anvils from previous convective storms are relatively unpredictable, this is expected to limit the predictability of the convective initiation. Copyright © 2007 Royal Meteorological Society

Secondary initiation of multiple bands of cumulonimbus over southern Britain. II: Dynamics of secondary initiation

AbstractThe Convective Storm Initiation Project (CSIP) pilot field programme took place in the summer of 2004 in southern England. During this field campaign a case of a cold‐pool outflow initiating an arc of convection downwind of the generating storm (the ‘primary storm’) was observed. Three further arcs were initiated further downwind of this first arc. These arcs all later gave significant rainfall over south‐east England. Results from the modelling studies described in this paper show that gravity waves, generated by the ‘primary storm’, may have been responsible for initiating the further three arcs of convective showers that were observed.The modelled primary storm generates waves with a range of vertical wavelengths. These are separated by dispersion, with the higher‐order modes (with the largest vertical wavelengths) travelling fastest (approximately 30 m s−1). The fastest two modes suppressed convection and later modes increased boundary‐layer depth and so initiated convection. The multiple peaks (from non‐hydrostatic effects) in the third mode may have been responsible for the two main arcs observed and the fourth mode may have been responsible for the third arc. The amplitude of the modelled waves at low levels was much larger downwind of the primary storm than upwind and only initiated convection there, which agrees with the observations of the arcs forming downwind of the primary storm. This asymmetry appears to be caused by the moving‐wave source and asymmetric wave trapping. Finally we discuss the implications for numerical weather prediction (NWP). Reducing the resolution of the model, and varying the magnitude of the wind speed used, shows that a non‐hydrostatic high‐resolution ( ≃1 km) NWP model should be able to capture such wave processes, if the generating storm is forecast or analysed correctly. Copyright © 2006 Royal Meteorological Society.