78
A. Bisignano et al.
Here, preliminary simulations with the new SMART (Spray—Moloch Atmospheric Regional Tool) suite are compared to the results obtained with the RMS
(RAMS-MIRS-SPRAY) modelling system for a case of one-day emissions in the
complex terrain of the Agri valley in southern Italy. The investigation focuses on
assessing the effect of the different meteorology on the pollutant dispersion. Comparisons between the meteorological variables, the plume dynamics and the concentration fields are presented and discussed.
13.2 The Case Study and the Simulations
In order to proceed with preliminary tests of the SMART suite, we considered a
one-day case, November 30 in 2013, referring to a recent impact assessment study
[2], where the RMS modelling system was applied to simulate the dispersion of
the pollutant emitted by the COVA oil refinery plant in the Agri valley. This last
is located in the south-western sector of Basilicata Region in southern Italy, it is
orientated NW-SE and is bordered on both sides by the Apennine Mountains, thus
it is characterized by complex and heterogeneous terrain.
Four nested 3D grids were used in RAMS atmospheric model, respectively with
48, 12, 4 and 1 km horizontal grid size, all with 35 levels on a stretched vertical grid
(first level at 24 m, top of domain at 22 km). A one-year simulation was run for 2013
and to reduce the time needed to perform the yearly simulation, RAMS analysis fields
were acquired from previous runs over Italy for the two coarse domains of 48 and
12 km resolution. They were then used as input and nudging on hourly basis for the
two nested domains at 4 and 1 km resolution. This implies that the two-way nesting
was not active from the two finest to the two coarsest grids. From RAMS fields
on the finest 1-km resolution domain (45 × 30 km
2 ), MIRS boundary-layer code
calculated the surface layer and turbulent variables needed by SPRAY Lagrangian
particle dispersion model.
A selection of the MOLOCH outputs at the grid resolution of 1.25 km, stored for
year 2013, was gathered for a domain 152 × 176 km
2 with 47 stretched vertical levels
(first level at 33 m, top of domain at 11 km). In addition, a devoted MOLOCH run
was performed using a finer resolution of 0.5 km on a domain 172 × 180 km
2 , with
the same vertical levels. In both cases, the new interfacing code ARAMIS was run to
elaborate MOLOCH fields in the format useful for SPRAY model and to calculate
the turbulent variables needed by it.
In SPRAY dispersion simulation, the Lagrangian particles were emitted every 30 s
considering the emission from the different stacks as point sources. The number of
the released particles is established to generate a minimum concentration associated
at the single particle of 0.005 µg m
−3 for NO x , CO and SO 2 . Such minimum value
is appropriate for reproducing the concentrations with a good detail. The heights
of the stacks vary between 12 and 33 m and high temperatures and exit velocities
characterize the emissions, leading to strong plume rises in SPRAY model.
A. Bisignano et al.
Here, preliminary simulations with the new SMART (Spray—Moloch Atmospheric Regional Tool) suite are compared to the results obtained with the RMS
(RAMS-MIRS-SPRAY) modelling system for a case of one-day emissions in the
complex terrain of the Agri valley in southern Italy. The investigation focuses on
assessing the effect of the different meteorology on the pollutant dispersion. Comparisons between the meteorological variables, the plume dynamics and the concentration fields are presented and discussed.
13.2 The Case Study and the Simulations
In order to proceed with preliminary tests of the SMART suite, we considered a
one-day case, November 30 in 2013, referring to a recent impact assessment study
[2], where the RMS modelling system was applied to simulate the dispersion of
the pollutant emitted by the COVA oil refinery plant in the Agri valley. This last
is located in the south-western sector of Basilicata Region in southern Italy, it is
orientated NW-SE and is bordered on both sides by the Apennine Mountains, thus
it is characterized by complex and heterogeneous terrain.
Four nested 3D grids were used in RAMS atmospheric model, respectively with
48, 12, 4 and 1 km horizontal grid size, all with 35 levels on a stretched vertical grid
(first level at 24 m, top of domain at 22 km). A one-year simulation was run for 2013
and to reduce the time needed to perform the yearly simulation, RAMS analysis fields
were acquired from previous runs over Italy for the two coarse domains of 48 and
12 km resolution. They were then used as input and nudging on hourly basis for the
two nested domains at 4 and 1 km resolution. This implies that the two-way nesting
was not active from the two finest to the two coarsest grids. From RAMS fields
on the finest 1-km resolution domain (45 × 30 km
2 ), MIRS boundary-layer code
calculated the surface layer and turbulent variables needed by SPRAY Lagrangian
particle dispersion model.
A selection of the MOLOCH outputs at the grid resolution of 1.25 km, stored for
year 2013, was gathered for a domain 152 × 176 km
2 with 47 stretched vertical levels
(first level at 33 m, top of domain at 11 km). In addition, a devoted MOLOCH run
was performed using a finer resolution of 0.5 km on a domain 172 × 180 km
2 , with
the same vertical levels. In both cases, the new interfacing code ARAMIS was run to
elaborate MOLOCH fields in the format useful for SPRAY model and to calculate
the turbulent variables needed by it.
In SPRAY dispersion simulation, the Lagrangian particles were emitted every 30 s
considering the emission from the different stacks as point sources. The number of
the released particles is established to generate a minimum concentration associated
at the single particle of 0.005 µg m
−3 for NO x , CO and SO 2 . Such minimum value
is appropriate for reproducing the concentrations with a good detail. The heights
of the stacks vary between 12 and 33 m and high temperatures and exit velocities
characterize the emissions, leading to strong plume rises in SPRAY model.