The skill of seasonal mean precipitation and temperature predictions by multi-model ensemble (MME) schemes has been examined using hindcast data from the APEC Climate Network (APCN). The hindcast data of the member models have been collected from several institutions in the APEC region. Skill metrics, namely, anomaly correlation coefficient (ACC), Hanssen-Kuipers skill (HKS), and economic value were used to assess the APCN MME predictions. The member models have a range of skills in predicting the seasonal mean conditions: however, the MME schemes consistently produce improved predictions. Though the weighted multi-model ensemble schemes have better skill scores in terms of ACC, the unbiased multi-model ensemble mean provides better and more useful categorical predictions as seen from the HKS score and the economic values. Prediction skills show distinct seasonality in different regions. Precipitation prediction skill is the lowest in the Asian monsoon region in June-July-August. The APCN MME predictions have positive economic values in most regions and examination of monthly values suggests that the MME predictions are reasonably good and useful even with three-month lead times. An uncalibrated multi-model probabilistic prediction scheme gives an indication of threshold probability for which the seasonal predictions are more beneficial to users. The extreme climatic conditions are predicted well by the MME schemes. However, user sectors for which the extreme climate predictions are beneficial are different to those of regular three-category predictions. These results can be used in planning and preparedness for the extremes of climate in the future, and when developing economic decision support models.

Meteosat infra-red imagery for the Great Storm of October 1987 is analysed to show a series of very shallow arc-shaped and smaller chevron-shaped cloud features that were associated with damaging surface winds in the dry-slot region of this extra-tropical cyclone. Hypotheses are presented that attribute these low-level cloud features to boundary-layer convergence lines ahead of wind maxima associated with the downward transport of high momentum from overrunning, so-called sting-jet, flows originating in the storm's main cloud head.

This study measured the amount of hoar-frost on the road surface with the help of copper plates. The amounts were compared to data collected at a station in the Road Weather information System (RWiS) together with additional measurements of the temperature and humidity profile above the road surface. The data were used to specify the weather parameters when hoar-frost forms on the road surface and relate the measured amount of hoar-frost and resulting friction to measured weather parameters. The aim was also to see whether there was an increased possibility of predicting the amount of hoar-frost with additional measurements of the temperature and humidity profiles. The results of this study confirm that the amount of hoar-frost deposited on the road surface increases with increasing average wind speed, increasing maximum difference between dew point (Td) and road surface temperature (Tr) and increased duration when Tr < Td, measured at the RWiS station. However, the difference between Td and Tr failed to indicate hoar-frost on several occasions due to the location and poor accuracy of the measuring equipment. Measurements of temperature and humidity in a profile up to 2.5 m above the road surface showed that the air layer up to 0.4 m reflected conditions at the road surface whereas the surrounding areas influenced the measurements at higher levels. A forward multiple regression performed on the data showed that measurements of temperature and humidity at lower levels improved predictions of deposition of hoar-frost on the road surface. Measurements of friction on the road surface when hoar-frost was present showed that friction decreased with increasing difference between Td and Tr and it also increased with the duration of Tr < Td. There was no correlation between the amount of hoar-frost and friction at the road surface. The friction is probably influenced not only by the amount of hoar-frost but also by the structure of the hoar-frost, which has not been determined in this study.

Atmospheric dispersion models are now being used by some local authorities in the United Kingdom as part of their management of urban air quality. Local authorities are also involved in using air pollution model output in environmental impact assessments for industrial projects, assessing the impact of road improvement or traffic management schemes and environmental health studies. The study, based in Northampton, UK, is concerned with the limitations of the meteorological data available to local authorities. The results indicate that while there is no major problem in the case of long-term average pollution concentration, meteorological data limitations pose problems when modelling individual events, particularly on occasions when higher pollution levels are expected.

Two case studies in which elevated thunderstorms played an important role in enhancing precipitation totals are discussed. During the period 20–22 February 1993, elevated thunderstorms over Iowa produced a mesoscale band of snow with amounts in excess of 25 cm across north central Iowa. Diagnosis of the environment revealed an elevated layer of convective instability between 700 and 540 hPa above a well-defined frontal zone. Elevated, upright convection resulted from the release of the convective instability as air parcels, ascending isentropically over the frontal zone, reached saturation. A second case, in which heavy rain (greater than 50 mm in 24 h) fell over parts of Oklahoma, Kansas and Missouri during the period 27–28 April 1994, is also examined. Once again, elevated thunderstorms resulted from the ascent of an elevated layer of convective instability over a strong baroclinic zone. Positive CAPE values are found by lifting air parcels having the maximum equivalent potential temperature in the lower portion of the troposphere, whereas no available energy is diagnosed for surface parcels. While both cases strongly resemble the climatology of elevated thunderstorms, these case studies suggest that convective instability aloft released by isentropic ascent, rather than frontogenetical forcing in the presence of weak symmetric stability, can result in heavy precipitation.

A three-day mesoscale numerical simulation has been performed over the narrow Salento peninsula (south-eastern Italy) during summer conditions characterised by weak synoptic forcing. These atmospheric conditions favour the development of complex sea-breeze systems and convergence zones on the peninsula. The aim of this work is to investigate the ability of an atmospheric mesoscale model to reproduce the surface fields of meteorological variables in the presence of local-scale forcing and breeze circulations, which are fundamental in applications such as air pollution modelling and nowcasting.

The modelled fields have been compared with available surface measurements and sodar data. Results indicate that the model can simulate the general mean wind field in a realistic way. The diurnal evolution of the wind is well reproduced and the maximum deviations mostly occur during the night, being associated with calm conditions.

Statistical analysis indicates that the typical mean bias is found to be about 1 m s-1 for hourly averaged wind speed, less than 20° for wind direction and about 1°C for temperature. The root mean square error (rmse) varies from 1 to 3 m s-1 for wind speed, from 50° to 70° for wind direction, and is about 2.4°C for temperature. All the values of the numerical indexes are within ranges which are characteristic of those found for other state-of-the-art models applied to similar cases studies.

Despite a good overall agreement between predictions and observations, some discrepancies were found in the individual profiles due both to the limited spatial representation of the local details and to the complex wind field which makes the space–time matching between the model and the observations quite critical.

The structures of the thermal mixed layer and the breeze convergence zone are similar to numerical studies relative to more idealised conditions.

The impact of vegetation on the microclimate has not been adequately considered in the analysis of temperature forecasting and modelling. To fill part of this gap, the following study was undertaken.

A daily 850-700 mb layer mean temperature, computed from the National Center for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR) reanalysis, and satellite-derived greenness values, as defined by NDVI (Normalised Difference Vegetation Index), were correlated with surface maximum and minimum temperatures at six sites in northeast Colorado for the years 1989-98. The NDVI values, representing landscape greenness, act as a proxy for latent heat partitioning via transpiration. These sites encompass a wide array of environments, from irrigated-urban to short-grass prairie. The explained variance (r2 value) of surface maximum and minimum temperature by only the 850-700 mb layer mean temperature was subtracted from the corresponding explained variance by the 850-700 mb layer mean temperature and NDVI values. The subtraction shows that by including NDVI values in the analysis, the r2 values, and thus the degree of explanation of the surface temperatures, increase by a mean of 6% for the maxima and 8% for the minima over the period March-October. At most sites, there is a seasonal dependence in the explained variance of the maximum temperatures because of the seasonal cycle of plant growth and senescence. Between individual sites, the highest increase in explained variance occurred at the site with the least amount of anthropogenic influence. This work suggests the vegetation state needs to be included as a factor in surface temperature forecasting, numerical modeling, and climate change assessments.

A three-dimensional numerical model for mountain waves is described and validated against radiosonde observations over the Isle of Arran, south-west Scotland. The model is based on the Boussinesq equations of motion linearised about a background profile of wind, potential temperature and density. The equations are discretised using finite-difference approximations and the time dependence is retained. Solutions are then computed by integrating forward the equations in time until a stationary mountain-wave field is obtained. Data from radiosondes, launched during fieldwork on Arran during February and March 2000, are used to represent the background flow and model simulations of the wave field are compared directly with fluctuations in the ascent rates of the balloons. On average, the model reproduces the observed maximum peak-to-peak vertical velocity to within about 0.6 ms-1. In many cases the accuracy of the simulations is even better than this: the model vertical velocity along the path of the radiosondes agree well with fluctuations in the balloon ascent rate. The sensitivity of the model results to the wave ‘launching height’ is discussed. The model was also used to provide forecasts of wave activity during the field campaign. In this case, rather than using radiosonde data to represent the background flow, data were extracted from a global forecast model. The accuracy of these forecasts is examined and the viability of using such a model to provide routine operational high-resolution mountain-wave forecasts for specific sites is discussed.

An assimilation of radar reflectivity into a numerical weather prediction (NWP) model with a horizontal resolution of 2.8 km is presented and applied to three severe convective events. The suggested assimilation method takes into account differences between the model and radar-derived precipitation in modifying vertical profiles of water vapour mixing ratio in each model time step by the nudging approach. Version 3.9 of the LM COSMO (Local Model COSMO) – NWP model used in this study includes the explicit formulation of the cloud and rain processes involved. Two variants of the assimilation technique are designed and outputs of their implementation are compared. The first variant makes use of the ground data only, while the second utilises vertical profiles of precipitation water. Both variants provide an improvement of precipitation forecast in comparison with outputs of the control run without assimilation procedures applied. When the assimilated radar data indicate initial precipitation near an expected storm, the NWP model is capable of forecasting basic features of the storm development two to three hours ahead. Three case studies are presented. In one, the assimilation method that takes into account the vertical structure of the precipitation water yields better results than the others which utilise ground data only. However, for the remaining two case studies both types of the assimilation method produce comparable results.

A recent requirement charged to the Aviation Weather Center (AWC) is to produce significant weather charts for aviation users with, among other forecast products, forecast locations of significant fronts. To increase forecaster productivity, the AWC decided to evaluate the possibility that fronts should be first drawn objectively. Hewson (1998) describes the basic technique which uses a variation on the Renard & Clarke (1965) frontal locator function to find the fronts. The AWC had to overcome many problems in implementing Hewson's techniques. This paper illuminates the problems and describes the AWC solutions. As a result of the AWC's success, objective frontal analyses and forecasts are now a reality, and the productivity of forecasters increased.

A volcanic ash detection product using AVHRR data has been developed for use in the London Volcanic Ash Advisory Centre, operated by forecasters at the Met Office. The product is an image that shows the brightness temperature difference between two infrared channels, which can be used to discriminate between clouds containing volcanic ash particles and those containing water droplets or ice crystals. Factors such as water vapour, water droplet and ice crystal content, opacity, cloud top temperature, ash particle size, surface emissivity properties and instrumental effects all have an effect on the spectral signal represented in the image. AVHRR imagery has been used to study the volcanic clouds ejected from eruptions of Mount Etna (Italy) in July 2001, Hekla (Iceland) in February 2000 and Mount Spurr (USA) in September 1992 to demonstrate the application of this product during volcanic events. The volcanic ash detection product provides essential information that can help forecasters locate volcanic ash and give appropriate guidance through advisory statements to the aviation industry and thus help avoid dangerous and expensive encounters between aircraft and volcanic ash.

Analysis of observational data of convective storms in Northern England suggests that the particular combination of effects such as sea breezes, elevated terrain and the presence of large cities has an influence on the initiation and development of convective storms. To study the importance of these effects for the development of convection, a cloud physics model has been initialised with the topography and surface characteristics of Northern England. The model enabled inclusion of three principal influences: sea breezes, topography and urban heat islands. The results suggested that the presence of the Pennines, a north-south orientated ridge, could influence the initiation of convection due to its long sun-facing slopes, and to a lesser degree forced lifting along the slopes. Inclusion of urban heat island effects focused convection downwind of the principal urban areas.

Numerical simulations are performed using the Penn State University/ National Center for Atmospheric Research Mesoscale Model (MM5) to study the impact of initial conditions on the super cyclone which hit the coast of Orissa in 1999. Because analysis of the cyclone's circulation was inadequate in the initial fields owing to the coarse resolution of the operational analysis systems and sparse oceanic data coverage, synthetic vortex data were generated using empirical relations and used in the analysis. Four-dimensional data assimilation is performed in order to assimilate the synthetic vortex in the initial stage to the model.

Considerable improvement in the track is obtained by using the synthetic vortex. With better specification of the initial vortex structure, the model successfully simulated the typical tropical cyclone characteristics, such as asymmetries in the wind field: the strongest winds occurred in the east and close to the cyclone's centre, strong wind gradients were found between the centre and the maximum wind region, and there was a slow decrease in wind speed up to the middle troposphere. Despite failing to produce the intense pressure drop observed for this cyclone, the model shows much better cyclone development with enhanced initial condition than the analysis.

This paper reports the results of the aerosol lidar experiments that have been performed at the Indian Institute of Tropical Meteorology (IITM), Pune (18.54°N, 73.85°E, 559 m amsl), a tropical station in India. The lidar-observed cloud macro-physical parameters (cloud-base and cloud-ceiling heights, vertical thickness, etc.) and polarisation characteristics and their association with surface-generated aerosols at the experimental site are presented and discussed. The correspondence among the lidar-derived aerosol distributions, meteorological parameters and south-west (SW) monsoon (June-September) activity over Pune during 12 successive SW monsoon seasons (1987-98) including two pairs of contrasting seasons of 1987-8 and 1993-4 is also examined. The results indicate an association between variations in aerosol loading in the boundary layer during the pre-monsoon season (March-May) and precipitation intensity during the ensuing monsoon season. Moreover, the decrease in aerosol content from pre-monsoon to monsoon season is found to follow the SW monsoon season total precipitation. Thus the results suggest that (i) the IITM lidar can also be a useful remote sensor for aerosol characterisation studies from polarisation measurements, and some important physical properties of clouds in the lower atmosphere over the station, and (ii) there exists a correspondence between boundary-layer aerosol content and SW monsoon precipitation over Pune, which is explained in terms of the type of aerosols and the environmental and meteorological processes, particularly during pre-monsoon and monsoon months prevailing over the experimental station.

A method of identifying sea-breeze fronts by remote sensing (visible and infrared data from the GOES-8 satellite) was applied to north-eastern Brazil. The aim of the study was to estimate the frequency of observed sea-breeze fronts and the average distance they penetrate inland at 1800 UTC. During the dry season (between September and December) in 2000, the highest frequency of sea-breeze fronts was recorded on the northern fringe of the study area. The sea-breeze fronts penetrated furthest inland between September and November, with a maximum distance from the coast at Ceará of 100 km.

An automated contrail detection algorithm has been applied to AVHRR data to obtain the position of contrails. Nearly simultaneous data from the Europe Model of the Deutscher Wetterdienst have been used to find typical weather conditions associated with 742 persistent contrails. Further, a flow pattern analysis identifies typical regions where the occurrence of contrails was above average. These regions are in the upper atmosphere: (a) ahead of a surface warm front either in moist warm layers before the cirrus clouds arrive or more likely with the cirrus in a warm conveyor belt and (b) ahead of a surface cold front in rapidly moving cold air in the turbulent regions near a band of strong wind (though the speed is not necessarily as high as in a jet). Usually, the atmosphere is baroclinic in the contrail region. Most of the detected contrails occur in divergent flow patterns in the upper troposphere in slowly rising warm or locally turbulent cold air masses.

A ten-year climatology has been prepared of dust days for central eastern Australia, from January 1995 to February 2004, using reports from the Australian Bureau of Meteorology's observational network. It was found that there had been a total of 55 dust days during this ten-year period. Of these, 43 dust days were associated with cold fronts: 24 were classified as embedded in the zonal westerlies in the Great Australian Bight (GAB) or in westerlies derived from low pressure systems in the GAB, whereas the remaining 19 were associated with cold fronts over eastern Australia where high pressure systems in the GAB generated strong post-frontal south to southeast winds. Three detailed case studies of dust storm generation and transport were modelled using a sophisticated coupled wind erosion prediction system. The model predictions in all three cases were found to be broadly consistent both with satellite images highlighting dust and the synoptic observations that reported dust. Antecedent land surface conditions are found to be critically important in generating dust, and their representation in the integrated prediction system is subsequently important in providing skilful model predictions of dust transport.

Some meteorological and geo-orographical factors related to cloud-to-ground flashes in Castilla-Leon (Spain) are studied. The relationship between flash density and the number of days with thunderstorms (thunderdays) is analysed, and a clear dependence and geographical variability is found. It is also shown that flashes have an average delay of four hours with respect to the time of the beginning of convection. The effect on flashes of dry static instability and moisture content is analysed by classifying days into four types according to the average values of the 850-700 hPa temperature difference and the 850 hPa dew-point temperature. A similar classification using the 700 hPa wind direction is used to analyse the effect of the mean-level flow on flashes. The results suggest that the daily number of flashes is mainly affected by the moisture content. Finally, a relationship between the cloud-to-ground flash density and geo-orographical factors (altitude, latitude and longitude) is deduced using multiple linear analysis.

In Part II of this study the ECMWF Ensemble Prediction System (EPS) is used to study the probabilistic predictability of three major European storms of the twentieth century. The storms considered are the Dutch storm of 1 February 1953, the Hamburg storm of 17 February 1962, and the British/French storm of October 1987 (Great October storm). Common to all these storms is their severity that caused large loss of life and widespread damage.

In Part I of this study it has been found that deterministic predictability of the Dutch and Hamburg storms amount to 48 and 84 hours, respectively. Here, it is shown that the ensemble forecasts supplement the deterministic forecasts. The large number of members in the 48 and 84 hour ensemble forecasts of the Dutch and Hamburg storms, respectively, suggest that at this forecast range and for these storms the sensitivity of the forecasts to analysis and model uncertainties is rather small. From these results, therefore, it is argued that reliable warnings (i.e. low probability for the occurrence of a forecast failure) for the Dutch and Hamburg storms could have been issued 48 and 84 hours, respectively, in advance, had the current ECMWF EPS been available.

For the Great October storm it has been found in Part I of this study that short-range and medium-range forecasts of the intensity and track of the storm were very skilful with a high-resolution model of the ECWMF model. The actual timing of the storm, however, was difficult to predict. Here, it is shown that the EPS is capable of predicting large forecast uncertainties associated with the timing of the Great October storm up to 4 days in advance. It is argued that reliable warnings could have been issued at least 96 hours in advance had the ECMWF EPS been available.

From the results presented in this study it is concluded that an Ensemble Prediction System is an important component of every early warning system for it allows an a priori quantification of the probability of the occurrence of severe wind storms.

Extensive studies of extra-tropical cyclones over the past sixty years have led to a plethora of conceptual models that describe the structure and evolution of storms. These conceptual (or structural) models blend observational, experimental and mathematical studies so as to represent the physical phenomena within a simple, yet flexible framework. The first three sections of this report review these conceptual models by stepping through the evolution of a mid-latitude cyclone and aim to illustrate how at each stage, conceptual models give a description of the physical processes occurring within the system. The fourth section reviews a recent cyclone classification scheme in the context of these conceptual models and attempts to group the cyclone classes into those arising through similar mechanisms. Conceptual models are then combined with this classification scheme so as to offer a unified perspective of cyclogenesis out of which a cyclone continuum develops.