TUE pm Michaelides 4 32

ACTIVITIES ON SHORT-RANGE FORECASTING METHODS OF FOG, VISIBILITY AND LOW CLOUDS IN EU COST ACTION 722Silas Chr. Michaelides  Meteorological Service, Nicosia, Cyprus: 

ACTIVITIES ON SHORT-RANGE FORECASTING METHODS OF FOG, VISIBILITY AND LOW CLOUDS IN EU COST ACTION 722 Silas Chr. Michaelides Meteorological Service, Nicosia, Cyprus WSN05 – Toulouse 5-9 September, 2005

Cost Action 722 is a consortium of scientists from twenty-two Institutions in fifteen countries, coordinating their national research on forecasting methods of fog, visibility and low cloud.: 

Cost Action 722 is a consortium of scientists from twenty-two Institutions in fifteen countries, coordinating their national research on forecasting methods of fog, visibility and low cloud. WSN05 – Toulouse 5-9 September, 2005

Participating Institutions in Cost Action 722Zentralanstalt für Meteorologie und Geodynamik (Austria)National Institute of Meteorology and Hydrology (Bulgaria)Meteorological Service of Canada (Canada)Cyprus Meteorological Service (Cyprus)Danemarks Meteorologiske Institute (Denmark)Finnish Meteorological Institute (FMI, Finland)Météo-France (France): 

Participating Institutions in Cost Action 722 Zentralanstalt für Meteorologie und Geodynamik (Austria) National Institute of Meteorology and Hydrology (Bulgaria) Meteorological Service of Canada (Canada) Cyprus Meteorological Service (Cyprus) Danemarks Meteorologiske Institute (Denmark) Finnish Meteorological Institute (FMI, Finland) Météo-France (France) WSN05 – Toulouse 5-9 September, 2005

Deutscher Wetterdienst (DWD, Germany)University of Marburg (Germany)University of Bonn (Germany)Hungarian Meteorological Service (Hungary)Norwegian Meteorological Institute (Norway)Institute of Meteorology and Water Management (Poland)Instituto Nacional de Meteorologia (INM, Spain): 

Deutscher Wetterdienst (DWD, Germany) University of Marburg (Germany) University of Bonn (Germany) Hungarian Meteorological Service (Hungary) Norwegian Meteorological Institute (Norway) Institute of Meteorology and Water Management (Poland) Instituto Nacional de Meteorologia (INM, Spain) WSN05 – Toulouse 5-9 September, 2005

Sveriges Meteorologiska och Hydrologiska Institut (SMHI - Sweden)Analysen & Konzepte (Switzerland)University of Basel (Switzerland)MeteoSwiss (Switzerland)University of Leeds (United Kingdom)University of Manchester, Institute of Science and Technology (United Kingdom)U.K. Met. Office (United Kingdom): 

Sveriges Meteorologiska och Hydrologiska Institut (SMHI - Sweden) Analysen & Konzepte (Switzerland) University of Basel (Switzerland) MeteoSwiss (Switzerland) University of Leeds (United Kingdom) University of Manchester, Institute of Science and Technology (United Kingdom) U.K. Met. Office (United Kingdom) WSN05 – Toulouse 5-9 September, 2005

Official Web site: 137.248.191.94/cost/ : 

Official Web site: 137.248.191.94/cost/ WSN05 – Toulouse 5-9 September, 2005

The main objective of the EU Cost Action 722 is:To develop advanced methods for very short-range forecasts of fog, visibility and low clouds, adapted to characteristic areas and to user requirements. : 

The main objective of the EU Cost Action 722 is: To develop advanced methods for very short-range forecasts of fog, visibility and low clouds, adapted to characteristic areas and to user requirements. WSN05 – Toulouse 5-9 September, 2005

This overall objective includes: the development of pre-processed methods of the necessary input data, the development of the appropriate forecast models and methods and the development of adaptable application software for the production of the forecasts. : 

This overall objective includes: the development of pre-processed methods of the necessary input data, the development of the appropriate forecast models and methods and the development of adaptable application software for the production of the forecasts. WSN05 – Toulouse 5-9 September, 2005

The duration of the Action is five years and the effort has been split into four Phases. : 

The duration of the Action is five years and the effort has been split into four Phases. WSN05 – Toulouse 5-9 September, 2005

The major tasks during the First Phase were a) to document the existing forecast techniques and ongoing projects in the field of forecasting fog, visibility and low clouds within European countries, and b) to investigate the needs by different groups of users, based upon the evaluation of questionnaires filled out by customers and forecasters in many European countries. : 

The major tasks during the First Phase were a) to document the existing forecast techniques and ongoing projects in the field of forecasting fog, visibility and low clouds within European countries, and b) to investigate the needs by different groups of users, based upon the evaluation of questionnaires filled out by customers and forecasters in many European countries. WSN05 – Toulouse 5-9 September, 2005

The First Phase has been Completed and the results have been published: 

The First Phase has been Completed and the results have been published WSN05 – Toulouse 5-9 September, 2005

The second Phase (Research and Development) is in its final year of completion and the three broad research areas: : 

The second Phase (Research and Development) is in its final year of completion and the three broad research areas: WSN05 – Toulouse 5-9 September, 2005

1st) initial data acquisition, techniques and requirements, 2nd) initialization of models and testing of the microphysics of different models, 3rd) the development and testing of statistical methods. : 

1st) initial data acquisition, techniques and requirements, 2nd) initialization of models and testing of the microphysics of different models, 3rd) the development and testing of statistical methods. WSN05 – Toulouse 5-9 September, 2005

The aim of the Third Phase (Development and Application) is to ensure that the methods can be used at different Institutes. During the last Fourth Phase (Dissemination) the results and methods will be distributed.: 

The aim of the Third Phase (Development and Application) is to ensure that the methods can be used at different Institutes. During the last Fourth Phase (Dissemination) the results and methods will be distributed. WSN05 – Toulouse 5-9 September, 2005

In the following, a brief review is made of some of the tasks undertaken under different research areas of the current second Phase in order to display the spectrum of the research efforts in Cost 722: 

In the following, a brief review is made of some of the tasks undertaken under different research areas of the current second Phase in order to display the spectrum of the research efforts in Cost 722 WSN05 – Toulouse 5-9 September, 2005

1st Research Area: INITIAL DATA The objective of this research area is to use satellite data and in-situ measurements for the derivation of a climatology of visibility, fog and low stratus and for a better understanding of relevant processes for the formation and dissipation of fog and low stratus. : 

1st Research Area: INITIAL DATA The objective of this research area is to use satellite data and in-situ measurements for the derivation of a climatology of visibility, fog and low stratus and for a better understanding of relevant processes for the formation and dissipation of fog and low stratus. WSN05 – Toulouse 5-9 September, 2005

The two main activities under this research area are:The climatology of fog, andthe detection and determination of fog properties: 

The two main activities under this research area are: The climatology of fog, and the detection and determination of fog properties WSN05 – Toulouse 5-9 September, 2005

Climatology of fog: 

Climatology of fog WSN05 – Toulouse 5-9 September, 2005

The Instituto Nacional de Meteorologia (INM), reports on the development of a conceptual model for fog forecasting in Central Spain.: 

The Instituto Nacional de Meteorologia (INM), reports on the development of a conceptual model for fog forecasting in Central Spain. WSN05 – Toulouse 5-9 September, 2005 Climatology of fog: Instituto Nacional de Meteorologia (INM)

Climatology of fog: Instituto Nacional de Meteorologia (INM)Through a systematic subjective mesoscale analysis, the coexistence of two mesoscale mechanisms in most of fog events has been shown: cold drainage winds converging at the bottom of the valleys and a warm advection of air from the Mediterranean or the Atlantic: 

Climatology of fog: Instituto Nacional de Meteorologia (INM) Through a systematic subjective mesoscale analysis, the coexistence of two mesoscale mechanisms in most of fog events has been shown: cold drainage winds converging at the bottom of the valleys and a warm advection of air from the Mediterranean or the Atlantic WSN05 – Toulouse 5-9 September, 2005

Under these circumstances, nights with clear skies, katabatics winds are overflown by warm air.: 

Under these circumstances, nights with clear skies, katabatics winds are overflown by warm air. WSN05 – Toulouse 5-9 September, 2005 Climatology of fog: Instituto Nacional de Meteorologia (INM)

Slide22: 

WSN05 – Toulouse 5-9 September, 2005 Methodology for fog forecasting is based on a combination of Teledetection (using Meteosat channel 6, centered at 7.3 m), and Numerical Weather Pediction (single-column version of HIRLAM) Climatology of fog: Instituto Nacional de Meteorologia (INM) WV image HIRLAM 00+06 T and Wind 850 hPa

In the Finnish Meteorological Institute, work is in progress, regarding the development of climatological forecasting tools. : 

In the Finnish Meteorological Institute, work is in progress, regarding the development of climatological forecasting tools. WSN05 – Toulouse 5-9 September, 2005 Climatology of fog: Finnish Meteorological Institute (FMI)

30 years of SYNOP observations for selected stationsObservations should be classified in groups of typical low visibility situationsIt would help forecasters to learn the climatologies of new areas they are assigned to. It might even show new dependencies, not thought of before. Clustering algorithms can do this.: 

30 years of SYNOP observations for selected stations Observations should be classified in groups of typical low visibility situations It would help forecasters to learn the climatologies of new areas they are assigned to. It might even show new dependencies, not thought of before. Clustering algorithms can do this. WSN05 – Toulouse 5-9 September, 2005 Climatology of fog: Finnish Meteorological Institute (FMI)

Clustering of fog “situations”: 

Clustering of fog “situations” WSN05 – Toulouse 5-9 September, 2005 Climatology of fog: Finnish Meteorological Institute (FMI)

Example:An advection fog cluster using NCEP data. Mean and the standard deviation of geopotential and temperature: 

Example:An advection fog cluster using NCEP data. Mean and the standard deviation of geopotential and temperature WSN05 – Toulouse 5-9 September, 2005 Climatology of fog: Finnish Meteorological Institute (FMI)

The Institute of Meteorology and Water Management of Poland (IMGW) developed a climatology on small-scale advection for Warsaw and Krakow Airports based upon 30 years of synoptic data.: 

The Institute of Meteorology and Water Management of Poland (IMGW) developed a climatology on small-scale advection for Warsaw and Krakow Airports based upon 30 years of synoptic data. WSN05 – Toulouse 5-9 September, 2005 Climatology of fog: Institute of Meteorology and Water Management of Poland (IMGW)

Slide28: 

WSN05 – Toulouse 5-9 September, 2005 Representation of a cluster of conditions favorable for fog formation Climatology of fog: Institute of Meteorology and Water Management of Poland (IMGW)

The Bulgarian National Institute of Meteorology and Hydrology works on fog climatologyBy using multiple linear regression (data sequences with a length of 92 terms), statistical relationships between visibility and several predictors were established: 

The Bulgarian National Institute of Meteorology and Hydrology works on fog climatology By using multiple linear regression (data sequences with a length of 92 terms), statistical relationships between visibility and several predictors were established WSN05 – Toulouse 5-9 September, 2005 Climatology of fog: Bulgarian National Institute of Meteorology and Hydrology

Several parameters have been examined: relative humidity, temperature, wind direction and speed, at four levels at Sofia station, 700, 850, 925hPa, and ground level together with SYNOP data for several stations in Bulgaria: 

Several parameters have been examined: relative humidity, temperature, wind direction and speed, at four levels at Sofia station, 700, 850, 925hPa, and ground level together with SYNOP data for several stations in Bulgaria WSN05 – Toulouse 5-9 September, 2005 Climatology of fog: Bulgarian National Institute of Meteorology and Hydrology

Encouraged by the results of this study, Aladin and HRM are employed to obtain visibility values in the beginning and in the end of the fog formation: 

Encouraged by the results of this study, Aladin and HRM are employed to obtain visibility values in the beginning and in the end of the fog formation WSN05 – Toulouse 5-9 September, 2005 Climatology of fog: Bulgarian National Institute of Meteorology and Hydrology

Detection and determination of fog properties : 

Detection and determination of fog properties WSN05 – Toulouse 5-9 September, 2005

In the Finnish Meteorological Institute, tests were carried out with the wind profiler LAP3000 (Vaisala) at Rovaniemi Airport. This instrument showed a good performance during situations with fog and low stratus.: 

In the Finnish Meteorological Institute, tests were carried out with the wind profiler LAP3000 (Vaisala) at Rovaniemi Airport. This instrument showed a good performance during situations with fog and low stratus. WSN05 – Toulouse 5-9 September, 2005 Detection and determination of fog properties : Finnish Meteorological Institute (FMI)

In Météo-France, work is progress to review the potentials and shortcomings of different LIDAR and SODAR instruments. Both methods are good for fog, however, LIDAR cannot penetrate dense fog and SODAR hardly detects thin fog layers. Heavy air traffic causes problems which limits the use at airports. : 

In Météo-France, work is progress to review the potentials and shortcomings of different LIDAR and SODAR instruments. Both methods are good for fog, however, LIDAR cannot penetrate dense fog and SODAR hardly detects thin fog layers. Heavy air traffic causes problems which limits the use at airports. WSN05 – Toulouse 5-9 September, 2005 Detection and determination of fog properties : Météo-France

At the University of Marburg, Germany, research is done on the retrieval of fog areas and properties from satellite imagery. : 

At the University of Marburg, Germany, research is done on the retrieval of fog areas and properties from satellite imagery. WSN05 – Toulouse 5-9 September, 2005 Detection and determination of fog properties : University of Marburg, Germany

Using parameterizations based on radiative transfer and microphysics, fog and low stratus can be separated with some degree of certainty. : 

Using parameterizations based on radiative transfer and microphysics, fog and low stratus can be separated with some degree of certainty. WSN05 – Toulouse 5-9 September, 2005 Detection and determination of fog properties : University of Marburg, Germany

2nd Research Area: MODELS The main activities under this research area are: the initialization of models, the investigation of microphysics, the development and testing of models: 

2nd Research Area: MODELS The main activities under this research area are: the initialization of models, the investigation of microphysics, the development and testing of models WSN05 – Toulouse 5-9 September, 2005

In University of Basle Switzerland, by using the NCEP (National Centers for Environment Prediction) model (1km), it was found that topography is important for forecasting fog and low clouds. First trials of ensemble forecasting are in progress by varying the initial conditions slightly.: 

In University of Basle Switzerland, by using the NCEP (National Centers for Environment Prediction) model (1km), it was found that topography is important for forecasting fog and low clouds. First trials of ensemble forecasting are in progress by varying the initial conditions slightly. WSN05 – Toulouse 5-9 September, 2005 Initialization of models: University of Basle, Switzerland

A fog and low stratus forecast system was developed for Zurich Unique airportin Switzerland. Fog prediction is done using an ensemble of 1D forecasts. Necessary initial conditions are obtained from variational data assimilation.: 

A fog and low stratus forecast system was developed for Zurich Unique airport in Switzerland. Fog prediction is done using an ensemble of 1D forecasts. Necessary initial conditions are obtained from variational data assimilation. WSN05 – Toulouse 5-9 September, 2005 Initialization of models: University of Basle, Switzerland

Investigation of microphysics : 

Investigation of microphysics WSN05 – Toulouse 5-9 September, 2005

In order to improve the understanding of the microphysical processes within the clouds and to get a better understanding of the phenomenon “fog”, several measurement campaigns (field and aircraft) are planned for winter 2005/2006 (Finland, Toronto, New York Airport). These experiments will also serve to inter-compare ground-based fog property measurements with satellite retrievals.: 

In order to improve the understanding of the microphysical processes within the clouds and to get a better understanding of the phenomenon “fog”, several measurement campaigns (field and aircraft) are planned for winter 2005/2006 (Finland, Toronto, New York Airport). These experiments will also serve to inter-compare ground-based fog property measurements with satellite retrievals. WSN05 – Toulouse 5-9 September, 2005 Investigation of Microphysics : Finnish Meteorological Institute and Met. Service of Canada

Development and testing of models : 

Development and testing of models WSN05 – Toulouse 5-9 September, 2005 Development and testing of models

Physical processes involved in the formation of fog interact in a complex and highly non-linear manner. Such interactions are not resolved by operational mesoscale models. Therefore, the development of a 3-D fog forecasting model calls for a finer horizontal and vertical resolution. : 

Physical processes involved in the formation of fog interact in a complex and highly non-linear manner. Such interactions are not resolved by operational mesoscale models. Therefore, the development of a 3-D fog forecasting model calls for a finer horizontal and vertical resolution. WSN05 – Toulouse 5-9 September, 2005 Development and testing of models

For the development and testing of models in forecasting low visibilities and low cloud, several models are under examination by various Institutions within Cost 722.: 

For the development and testing of models in forecasting low visibilities and low cloud, several models are under examination by various Institutions within Cost 722. WSN05 – Toulouse 5-9 September, 2005 Development and testing of models

The University of Bonn works on the development of a nested 50km 3-D-version of the LM (Lokal modell of DWD) with a very fine vertical resolution near the surface, at 2, 6, 10 m from the ground. The parameterizations are taken from the parameterized fog microphysics model – PAFOG : 

The University of Bonn works on the development of a nested 50km 3-D-version of the LM (Lokal modell of DWD) with a very fine vertical resolution near the surface, at 2, 6, 10 m from the ground. The parameterizations are taken from the parameterized fog microphysics model – PAFOG WSN05 – Toulouse 5-9 September, 2005 Development and testing of models : University of Bonn, Germany

PAFOG is also used in another modeling approach which involves the Universities of Bonn and Basle. In this approach, the NOAA/NCEP Nonhydrostatic Mesoscale Model (NMM) is coupled with PAFOG.: 

WSN05 – Toulouse 5-9 September, 2005 NMM_PAFOG PAFOG is also used in another modeling approach which involves the Universities of Bonn and Basle. In this approach, the NOAA/NCEP Nonhydrostatic Mesoscale Model (NMM) is coupled with PAFOG. NMM-4 NMM_PAFOG Development and testing of models : Universities of Bonn, Germany and Basle, Switzerland

The representation of low cloudiness (Stratus) with the Aladin model is studied in Zentralanstalt für Meteorologie und Geodynamik (Austria) by extending their system: Integrated Nowcasting through Comprehensive Analysis (INCA) : 

The representation of low cloudiness (Stratus) with the Aladin model is studied in Zentralanstalt für Meteorologie und Geodynamik (Austria) by extending their system: Integrated Nowcasting through Comprehensive Analysis (INCA) WSN05 – Toulouse 5-9 September, 2005 Development and testing of models : Zentralanstalt für Meteorologie und Geodynamik, Austria

The HIRLAM 1-D model (initialized with 3-D) is tested in the Sveriges Meteorologiska och Hydrologiska Institut (SMHI).: 

The HIRLAM 1-D model (initialized with 3-D) is tested in the Sveriges Meteorologiska och Hydrologiska Institut (SMHI). WSN05 – Toulouse 5-9 September, 2005 Development and testing of models : Sveriges Meteorologiska och Hydrologiska Institut (SMHI)

A numerical forecast method integrating dedicated observations and the COBEL-ISBA high resolution numerical model has been developed in Météo-France.: 

A numerical forecast method integrating dedicated observations and the COBEL-ISBA high resolution numerical model has been developed in Météo-France. WSN05 – Toulouse 5-9 September, 2005 Development and testing of models : Météo-France

Experience with the modelling approach shows:Promising first resultsComputationally very efficient and feasible in todays operational frameworkHowever, more cases and ‘verification’ needed: 

Experience with the modelling approach shows: Promising first results Computationally very efficient and feasible in todays operational framework However, more cases and ‘verification’ needed WSN05 – Toulouse 5-9 September, 2005

Model IntercomparisonWithin Cost 722, one very important component of the effort for the development and testing of models is the model inter-comparison. : 

Model Intercomparison Within Cost 722, one very important component of the effort for the development and testing of models is the model inter-comparison. WSN05 – Toulouse 5-9 September, 2005

Participating InstitutionUniversity of BonnAnalysen & KonzepteUniversity of BaselInstituto Nacional de Meteorologia: 

Participating Institution University of Bonn Analysen & Konzepte University of Basel Instituto Nacional de Meteorologia WSN05 – Toulouse 5-9 September, 2005 This task is currently under deployment but some preliminary results are already available for Lindenberg in Germany (where three 1-D models are tested on common fog periods) 1st Model Intercomparison

One more inter-comparison task is also under deployment in France for CDG Airport. Participating InstitutionInstituto Nacional de MeteorologiaAnalysen & KonzepteDanemarks Meteorologiske InstituteMétéo-FranceUniversity of BaselUniv. de les Illes Balears: 

One more inter-comparison task is also under deployment in France for CDG Airport. Participating Institution Instituto Nacional de Meteorologia Analysen & Konzepte Danemarks Meteorologiske Institute Météo-France University of Basel Univ. de les Illes Balears WSN05 – Toulouse 5-9 September, 2005 2nd Model Intercomparison

Slide54: 

WSN05 – Toulouse 5-9 September, 2005 3rd Research Area: STATISTICAL METHODS The following two issues are considered in this research area: a. Improvement and innovation (statistical deterministic and probabilistic forecasting of visibility, fog and low clouds are investigated) b. Selection of Predictors and evaluation of methods (The most important parameters for high quality forecasts are to be selected)

Improvement and innovation of statistical forecasting: 

Improvement and innovation of statistical forecasting WSN05 – Toulouse 5-9 September, 2005

The Norwegian Met. Service and the Cyprus Met. Serviceco-operate in developing an Artificial Neural Networkforecasting sytem for Larnaca Airport in Cyprus: 

The Norwegian Met. Service and the Cyprus Met. Service co-operate in developing an Artificial Neural Network forecasting sytem for Larnaca Airport in Cyprus WSN05 – Toulouse 5-9 September, 2005 Improvement and innovation : Norwegian Met. Service – Cyprus Met. Service

List of predictors used for forecasting Meteorological Optical Range (predictand). MonthHourVisibility (MOR)Height of cloud base (HCB)Wind speedWind directionTemperature Dew point temperatureRelative humidityPressure: 

List of predictors used for forecasting Meteorological Optical Range (predictand). Month Hour Visibility (MOR) Height of cloud base (HCB) Wind speed Wind direction Temperature Dew point temperature Relative humidity Pressure WSN05 – Toulouse 5-9 September, 2005 Improvement and innovation : Norwegian Met. Service – Cyprus Met. Service Data: From Automatic Weather Observing System

Graphical presentation of probabilistic forecasts.: 

Graphical presentation of probabilistic forecasts. WSN05 – Toulouse 5-9 September, 2005 Improvement and innovation : Norwegian Met. Service – Cyprus Met. Service

Hungarian Met. Service – Cyprus Met. Service co-operate in comparing different visibility forecast methods for Larnaca Airport- Cyprus: 

Hungarian Met. Service – Cyprus Met. Service co-operate in comparing different visibility forecast methods for Larnaca Airport- Cyprus WSN05 – Toulouse 5-9 September, 2005 Improvement and innovation : Hungarian Met. Service – Cyprus Met. Service

Three methods were tested:1. Artificial Neural Network 2. Genetic alghorithm.3. Decision tree (Hungarian).Data: From Automatic Weather Observing System and ECMWF: 

Three methods were tested: 1. Artificial Neural Network 2. Genetic alghorithm. 3. Decision tree (Hungarian). Data: From Automatic Weather Observing System and ECMWF WSN05 – Toulouse 5-9 September, 2005 Improvement and innovation : Hungarian Met. Service – Cyprus Met. Service

Slide61: 

WSN05 – Toulouse 5-9 September, 2005 Improvement and innovation : Hungarian Met. Service – Cyprus Met. Service 15-hour visibility forecast by using ECMWF forecast 1-7 March 2003 – Larnaca Airport

Selection of predictors and evaluation of methods : 

Selection of predictors and evaluation of methods WSN05 – Toulouse 5-9 September, 2005

Météo-France proposed a method for determining how to evaluate the potential of existing methods. The proposed methodology can be applied to any forecasting method, statistical or based on numerical modelling : 

Météo-France proposed a method for determining how to evaluate the potential of existing methods. The proposed methodology can be applied to any forecasting method, statistical or based on numerical modelling WSN05 – Toulouse 5-9 September, 2005 Selection of predictors and evaluation : Météo-France

Slide64: 

WSN05 – Toulouse 5-9 September, 2005 Selection of predictors and evaluation : Météo-France Pseudo-ROC diagram (which shows Hit Rate versus False Alarm Ratio, the bias of deterministic forecasts and the ability of probabilistic forecasts to discriminate between events and non-events) First aspect: detection and false alarms

Slide65: 

WSN05 – Toulouse 5-9 September, 2005 Selection of predictors and evaluation : Météo-France Reliability diagram (which shows observed frequency versus forecast probability and the probability bias of probabilistic forecasts). Second aspect: reliability (of prob forecasts)

In MeteoSwiss, work is under way for defining a methodology for the selection of predictors in statistical forecasting of fog and low visibilities.Two methodologies are under consideration:: 

In MeteoSwiss, work is under way for defining a methodology for the selection of predictors in statistical forecasting of fog and low visibilities. Two methodologies are under consideration: WSN05 – Toulouse 5-9 September, 2005 Selection of predictors and evaluation : MeteoSwiss

Slide67: 

WSN05 – Toulouse 5-9 September, 2005 Selection of predictors and evaluation : MeteoSwiss a) The variables are selected before the multivariate model is set up

Slide68: 

WSN05 – Toulouse 5-9 September, 2005 Selection of predictors and evaluation : MeteoSwiss b) The selection of variables is included in the whole modeling process

This presentation was just an overview of the activities in Cost 722 showing the wide spectrum of tasks undertaken by several researchers.Conclusion: Experience has shown that developing forecasting methods for such complex phenomena such those of low visibilities, fog and Stratus requires an extensive multi-disciplinary cooperation.: 

This presentation was just an overview of the activities in Cost 722 showing the wide spectrum of tasks undertaken by several researchers. Conclusion: Experience has shown that developing forecasting methods for such complex phenomena such those of low visibilities, fog and Stratus requires an extensive multi-disciplinary cooperation. WSN05 – Toulouse 5-9 September, 2005

Thank you: 

Thank you WSN05 – Toulouse 5-9 September, 2005