The combined Fog Monitoring System over the

FOG PILOT
The combined Fog Monitoring System over the Veneto Region – Italy
F. Zardini, F. Domenichini, A. Rossa, A. Della Valle, M. Calza, V. Gaspari, ARPAV – DRST, Meteorological Centre of Teolo - [email protected]
The FP7 project ROADIDEA studies the innovation potential of the European ITS sector by analysing available data sources, revealing existing
problems for data utilisation and service build-up. The Fog Pilot, one of ROADIDEA’s pilots, constitutes a proof of concept, as well as a
practical application of how road safety can be improved by accounting for adverse meteorological conditions, in this case reduced visibility.
Fog in the Po Valley
The presence of fog is a frequent phenomenon in
the Po Valley (up to 80-120 days per year), with
consequent strong impact on road, air, ship and
railway traffic, by dangerous reduction of visibility.
The network of visibility sensors
For the ROADIDEA Fog Pilot a network of 10
visibilimeters was installed. The siting is limited to
the Veneto plain most prone to intense and
persistent fog. Visibilimeters provide direct
visibility measures which then are converted in a
probability of fog (reduced visibility conditions) for
merging with other data sources.
The Fog Pilot data merging
The merging methodology for the different data
sources was chosen to be probabilistic and can
be divided into the following steps:
• estimate a fog probability map for each data
source;
• attribute to each data source a relative quality,
or weight, which accounts for the reliability for the
current conditions;
• statistically combine the individual probability
fields, accounting for the relative data quality.
The final product consists in a probability map of
reduced visibility, which can yield fog alerts
based on selected thresholds (e.g. 40%).
Social Economical Impact of Fog
The total social costs produced by fog in the
region Veneto is of the order of 35M€ (total costs
for accidents in Italy in 2003 is about 35G€, of
which 1% attributed to fog, of which 10%
attributed to Veneto).
CART
[Italian Central Institute of Statistics, Automobile Club Italia]
Present warning systems
Deployment of fog monitoring warning systems is
generally limited to selected stretches of the main
motorway networks, yielding only information on
the local fog conditions, not however on its areal
extension. Most of the regional to national road
networks are not covered by such a service.
Monitoring of fog
Fog monitoring is well known to be a significant
challenge, not least due to the high spatial and
temporal variability of the phenomenon. The
development and refinement of fog forecast
methods requires first a good monitoring system.
Visibilimeters
Visibility sensors (red boxes) used in the visibilimeter network for the Veneto Fog Pilot.
An emitted infrared beam is scattered by the fog droplets. The instrument measures the
scattered beam component, which is larger in presence of fog, and then estimates the
visibility in meters.
Satellite
Ground data statistical elaboration
The data obtained from the automatic surface
meteorological stations network yields another,
indirect probability estimate for fog occurrence
based on current meteorological conditions by
applying a statistical decision tree methodology
(CART).
Tree 7 graph for Vis categ
1
2
3
4
5
Num. of non-terminal nodes: 11, Num. of terminal nodes : 12
ID=1
N=28855
1
dT500PD
<= 0.004990
ID=2
> 0.004990
N=24334
1
ID=3
RH
1
> 94.500000
N=11180
ID=5
N=4521
1
<= 95.500000
N=13154
ID=106
T 2m (°C)
1
1
> 95.500000
N=911
ID=107
> 3.850000
N=5262
ID=35
1
N=7892
<= 0.010707
ID=108
1
N=825
> 0.010707
ID=109
4
ID=112
1
4
> 1008.125000
N=3272
ID=113
P Teolo
, ...
= Other(s)
N=1637
N=3610
<= 1008.125000
N=86
WD categ
=N
4
P Teolo
dT500PD
<= 3.850000
ID=34
4
RH
<= 94.500000
ID=4
ID=36
ID=37
1
<= 987.725000
N=3625
ID=114
1
> 987.725000
N=284
ID=115
4
N=2988
dTMtGr_Lgn
<= 0.005579
ID=46
1
> 0.005579
N=2511
ID=47
4
N=1114
P Teolo
<= 1008.100000
ID=58
4
> 1008.100000
N=998
ID=59
1
N=116
dT 12h
<= -6.500000
ID=60
5
N=150
4
1
N=338
The decision tree obtained
with the CART analysis
allows us to derive a fog
probability depending on the
meteorological conditions
(temperature, wind, humidity
etc.) recorded by the ground
stations. The fact that this
method very rarely gives high
fog probabilities limits its
value in comparison to the
other methods used in the
Fog Pilot.
Fog
prob
> -6.500000
ID=61
N=848
Fog
prob
• to set up a cost effective, areal fog monitoring
system for diagnosing reduced visibility conditions
over an extended geographical region as
alternative to the expensive in situ monitoring;
• set up a relatively low resolution network of
direct visibility measurements (visibilimeter);
=67%
Satellite information: elaboration by SAFNWC
The METEOSAT satellite images have the
potential to provide areal fog information and
were used to aid the spatial interpolation of the
direct visibility measurements. The satellite
information was processed by cloud classification
algorithm of the Satellite Application Facility
Nowcasting (SAFNWC), a software system
developed over the last 13 years by a consortium
of European national weather services.
CT product
SAFNWC
• set up of an information model to merge direct
and indirect visibility information derived from
satellite and meteorological surface stations;
• maintain open architecture to allow integration of
complementary and innovative data sources for
visibility (reports, intelligent cars, webcams, etc.).
METEOSAT Satellite image.
The output product of the Fog Pilot: on the left side the three products derived from single
monitoring methods, while the large frame represents the merged final product. In each
map the fog information is represented by the probability of having visibility under 500
meters.
Conclusions
The main results of the ROADIDEA Fog Pilot can
be summarized as follows:
• a probabilistic areal fog monitoring system has
been built using a limited number of visibility
sensors, a satellite cloud classification system and
a meteorological surface station network;
• probabilistic verification of the Fog Pilot shows
good performance with a probability of detection of
about 75%, probability of false detection of about
10%;
=0.5%
Objectives of the ROADIDEA Fog Pilot
The main objectives of the ARPAV Fog Pilot in
ROADIDEA are:
Merged Product
• most value to the final product is brought by the
direct visibility measurements, whereas the
satellite information brings added value for the
spatial interpolation; the impact of the
meteorological surface observations appears to be
limited, but has yet to be established more
rigorously;
• to the knowledge of the authors, the Fog Pilot
constitutes the first attempt of an areal combined
fog monitoring system for road transport purposes,
and has the potential to give a significant support
for accident prevention and traffic management.
ARPAV (Regional Agency for Environmental Prevention and Protection of
Veneto) is the regional meteorological service of the north-eastern Italian
region Veneto and, as such, is responsible for meteorological support to
institutional and private users. In the framework of ROADIDEA ARPAV
developed pilot system for the fog monitoring.
Cloud Type product derived by the SAFNWC
software package.