Official HYGEOS website


Prevision de la Visibilite

dans le cycle de vie du Brouillard

a partir d'Observations Sol et Satellite


Visibility Forecast

in the Fog life cycle

from Satellite and ground-based Observation



I) Radiative fog formation nowcasting

Data acquired during two fog seasons were analysed to identify predictors of the fog formation and to define most favourable conditions for fog formation. Data were acquired at the SIRTA platform, 20 km South-West of Paris, during the 2011-2012 and 2012-2013 seasons, and by the SEVIRI instrument onboard the METEOSAT Second Generation satellite. The cloud cover classification by the EUMETSAT/NWCSAF program was exploited. Six predictors of the formation of developed and thin fogs by radiative cooling were identified. The predictors concern the atmospheric visibility, the vertical thermal gradient in the first 30 m and the cloud cover both above the SIRTA and in a larger region. 23 pre-fog and 78 no-fog events consistent with the conditions of radiative cooling fog formation were studied (Fig. 1).


Figure 1. Top part of the decision tree showing the fog formation probability for different scenarii. The 2nd line shows that 23 pre-fog and 78 no-fog moderate visibility (mv) events occur in conditions of radiative cooling fog formation, i.e. clear-sky according to the ceilometer (CS).


Seven favourable scenarii (Fig. 2a and 2b) were defined using the six predictors and appropriate criteria on these predictors, where the developed and thin fog formation probability was 33%. Six unfavourable scenarii (Fig. 2a and 2b) were defined, where the fog formation probability was 8%. Almost 50% of the no-fog events were correctly identified, with a false alarm ratio of 60%, and 87% of the pre-fog events were identified. The anticipation time is varying in average between 2 and 4 hours, being slightly longer for thin than for developed fogs. Pre-fog conditions were not missed when the vertical thermal gradient was either large (DTv > 0.060°C/m) either close to zero but three pre-fog events were missed for intermediate values of the vertical thermal gradient (0.035 < DTv < 0.060°C/m).


Figure 2a. The cloud-free (CF) bottom part of the decision tree. The box of the 1st line is equal to the left hand side box of the 3rd line of Fig. 1. The blue lines show the scenario giving the developed fogs, the red lines show the scenarii giving the thin fogs, the green lines show the scenarii giving the intermediate fogs. The black lines are used when the type of fog is not yet defined. The boxes with white background show the unfavourable scenarii. The probability is computed as Npre-fog / ( Npre-fog + Nno-fog ).

CF = cloud-free bottom to top

LCI = regional increase of low cloud cover

CSI = regional increase of clear-sky

MHI = regional increase of mid and high level cloud cover

MHD = regional decrease of mid and high level cloud cover

NS = not stratified over 30 m

MS = moderately stratified over 30 m

Str = strongly stratified over 30 m


Figure 2b. As Fig. 2a but for the cirrus (CIR) bottom part of the decision tree. The box of the first line is equal to the centre box of the third line of Fig. 1.


Also, a correlation was observed between the regional cloud cover change and the thermal vertical gradient, as the developed fog sequence deploys only under cloud-free sky in conditions of regional increase of low cloud cover (LCI for low cloud increase). Consequently, only one scenario was defined for the six developed fogs (Fig. 2a), with 26% probability of fog formation: visibility between 5 and 10 km (the mv event), negligible temperature change from the surface level up to 30 m height (NS for not stratified), and cloud-free sky over the SIRTA (CFb-t for cloud-free), but with low clouds replacing clear-sky in the 9x9-pixel zone around the SIRTA (LCI). If visibility decreased below 5 km after this scenario, the developed fog formation probability increased from 26 to 40%. Also the no-fog event identification rate increased up to 61% and the false alarm rate decreased from 67 to 60%. However the anticipation time decreased to around 2 hours or less. In the same cloud cover configuration but with large temperature change, the formation probability is 33% for 6 of the thin fogs (Fig. 2a). Cirrus detected by MSG/SEVIRI were never observed before formation of developed fog. However the fog formation probability was 44% when visibility was included between 5 and 10 km below cirrus for other four of the thin fogs (Fig. 2b). With the further criteria on regional tendency of cloud cover and atmospheric stratification, probability reached 57%.

To decrease the false alarm rate, it is necessary to find further predictors in the conditions of cloud-free cover over the SIRTA but with increasing low cloud cover in the region around the SIRTA, where 39 no-fog events could not be distinguished from the pre-fog events (Fig. 2a).

The PreViBOSS results were used to provide the fog formation probability in real time, following the evolution of observed parameters, in a decision assistance tool developed by HYGEOS. For example on 15 November 2011, the radiative fog formation probability started to increase 2 hours before the fog formation (Fig. 3), and reached almost 100% 30 minutes before. The PreViBOSS decision tool showed the evolution above the SIRTA but also in a larger region, tens of km around the SIRTA.


Figure 3. Decision tool developed by HYGEOS for the PreViBOSS project. The radiative fog of 15 November 2011 at SIRTA is taken as an example. Arrows and text are added to briefly describe the different graphs.

cloud type index CT = 1: clear-sky (deep blue)

CT = 2: very low clouds

CT=3: low clouds (green)

CT=4-5: mid and high thick clouds

CT=6: high thin clouds (cirrus) (red-orange)

CT=7: scattered


Animations are proposed for several situations:

Succeeded nowcasting of a developed fog (scenarion F1): 15 November 2011

False alarm of a developed fog (scenario F1): 13 November 2011

Missed fog, starting as a thin fog (scenario U5), and then developing in th vertical direction: 01 November 2011

Succeeded nowcasting of a thin fog and false alarm, under cloud-free sky (scenario F2): 19 November 2011

Succeeded nowcasting of a thin and short fog under cirrus (scenarii F3 and F7): 20 November 2011 and 21 November 2011

False alarm of a thin fog under cirrus (scenario F3): 28 November 2011



II) Enhanced extinction of visible radiation due to hydrated aerosols in mist and fog

The PreVIBOSS project also  assesses the contribution of aerosols to the extinction of visible radiation in the mist-fog-mist cycle. Relative humidity is large in the mist-fog-mist cycle, and aerosols most efficient in interacting with visible radiation are hydrated and compose the accumulation mode. Measurements of the microphysical and optical properties of these hydrated aerosols with diameters larger than 400 nm were carried out under ambient conditions. Eleven mist-fog-mist cycles were observed in November 2011 at SIRTA, with a cumulated fog duration of 95 h, and a cumulated mist-fog-mist cycle duration of 240 h.

In mist, aerosols grew by taking up water at relative humidities larger than 93%, causing a visibility decrease below 5 km. While visibility decreased down to few km, the mean size of the hydrated aerosols increased, and their number concentration (Nha) increased from approximately 160 to approximately 600 cm-3. When fog formed, droplets became the strongest contributors to visible radiation extinction, and liquid water content (LWC) increased beyond 7 mg/m3. Hydrated aerosols of the accumulation mode co-existed with droplets, as interstitial non-activated aerosols. Their size continued to increase, and some aerosols achieved diameters larger than 2.5 micrometers. The mean transition diameter between the aerosol accumulation mode and the small droplet mode was 4.0±1.1 micrometers. Nha also increased on average by 60% after fog formation. Consequently the mean contribution to extinction in fog was 20±15% from hydrated aerosols smaller than 2.5 micrometers and 6±7% from larger aerosols. The standard deviation was large because of the large variability of Nha in fog, which could be smaller than in mist or three times larger.  Fig. 4 shows such variability for three mist-fog-mist cycles, as well as the variability in fog droplet contributions and in consecutive total particle extinction coefficient.



Figure 4. contribution of particles in different size range to the total particle extinction coefficient, during 3 mist-fog-mist cycles of November 2011 at SIRTA..


The particle extinction coefficient in fog can be computed as the sum of a droplet component and an aerosol component, which can be approximated by 3.5 Nha (Nha in cm-3 and particle extinction coefficient in Mm-1). We observed an influence of the main formation process on Nha, but not on the contribution to fog extinction by aerosols. Indeed, in fogs formed by stratus lowering (STL), the mean Nha was 360±140 cm-3, close to the value observed in mist, while in fogs formed by nocturnal radiative cooling under cloud-free sky (RAD), the mean Nha was 600±350 cm-3. But because visibility (extinction) in fog was also lower (larger) in RAD than in STL fogs, the contribution by aerosols to extinction depended little on the fog formation process. Similarly, the proportion of hydrated aerosols over all aerosols (dry and hydrated) did not depend on the fog formation process.

Measurements showed that visibility in RAD fogs was smaller than in STL fogs due to three factors: 1) LWC was larger in RAD than in STL fogs; 2) droplets were smaller; 3) hydrated aerosols composing the accumulation mode were more numerous.  A paper with peer-review was submitted.















Project manager: Thierry Elias / HYGEOS

  Voir le profil de Thierry Elias sur LinkedIn ResearchGate