SENSITIVITY OF MODEL RAMS TO

1
http//www.lamma.rete.toscana.it
SENSITIVITY OF MODEL RAMS TO THE REPRESENTATION
OF SURFACE FEATURES
F. Meneguzzo, G. Maracchi, D. Grifoni, G.
Messeri, M. Pasqui, M. Rossi IATA - CNR LaMMA
- Laboratory for Meteorology, climatology and
Environmental Modelling
2
Few literature results - 1

• Land Cover - LAI
• Complex sensitivity experiments demonstrated that
  atmospheric processes, including mesoscale
  circulations and the formation of clouds and
  precipitating systems, can be highly dependent on
  surface heat and moisture fluxes that are largely
  determined by live and dead vegetation and soil
  moisture storage. The general conclusion is that
  land surface processes play a significant role in
  defining local, regional, and global weather and
  climate.

Lu et al., 2001 Seasonal RAMS simulations over
US, changing LAI. 2-3 effect on maximum
temperature, strong unpredictable effects on
precipitation
Max. temp.
Daily precip.
3
Few literature results - 2

• Land Cover - Vegetation
• Correctly representing the vegetation cover is
  crucial to the quality of mesoscale atmospheric
  simulations.

Model output cloud and water vapour mixing ratio
fields on the third nested grid (grid 4) at 2100
UT on May 15, 1991, for the simulation with USGS
current vegetation (a) and the simulation with
natural landscape (b). The clouds are depicted by
white surfaces with qc 0.01g/kg, with the Sun
illuminating the clouds from the west. The vapour
mixing ratio in the planetary boundary layer is
depicted by the shaded surface with qv8g/kg .
The flat surface is the ground. Areas formed by
the intersection of clouds or the vapour field
with lateral boundaries are flat surfaces, and
visible ground implies qvlt8g/kg. The vertical
axis is height, and the back planes are the north
and east sides of the grid domain. From Pielke et
al. (1997).
Pielke et al., 1997 Simulation of a thunderstorm
event with current (USGS) and natural (short
grass) landscape in US. Dramatic differences!
4
Few literature results - 3
An extensive review of recent findings about the
links between surface moisture and heat fluxes
and cumulus convective rainfall from the global
to the regional scale is documented by Pielke
(2001), which reveals that the quality and
accuracy of the representation of the vegetation
classes and parameters, such as the leaf area
index, in atmospheric models at all scales is
decisive to the weather and climate forecasts and
of the same order of, or sometimes even greater
than, other well known forcings such as
atmospheric chemistry changes and sea surface
temperature anomalies (e.g. el Nino). A relevant
study was that of Avissar et al. (1998) on the
impacts of different scales of land surface
heterogeneity (landscape patchiness) on the
convective boundary layer (CBL), by means of
large-eddy simulations (LES) performed with model
RAMS. They pointed out that as long as the
patchiness of the landscape has a
characteristic landscape length scale smaller
than about 5-10 km, the impacts of the landscape
heterogeneity on the CBL are insignificant, while
they become very relevant at larger scales. In
other words, when mesoscale models are used at
5-10 km resolution, it would be sufficient to
compute the mean (gridscale) surface heat fluxes
from the distribution of fluxes obtained from the
different patches that constitute the landscape,
with no need for instance to implement separate
equations for each patch. Anyway, this is still a
matter of debate and recent results seem to
partially contrast.
5
RAMS used at LaMMA for ...

• Weather forecasts
• Tuscany Regional Government (Civil Protection -
  Internet, Radio, TV broadcast (RAI), wind energy
  planning, air quality, sea oil spill monitoring)
• River flood modelling
• Project DECIDE - Project STORM (ESA - Arno River
  Basin Authority)
• Sea state and hydrodynamics - Oil Spill
  Modelling
• Atmospheric Remote Sensing Research
• Project EURAINSAT (EU) - Precipitation
  assimilation into RAMS
• Research at ESA-ESRIN - e.g. RAMS verification
  with ERS-SAR images
• Atmospheric Science Research Applications
• National Program for Antarctic Research (PNRA)
• Extreme events predictability (GNDCI-CNR)
• Partnership with several medium to large
  Companies and public Institutes

6
COMPUTATION FACILITY FOR RAMS
7
RAMS Model Parallelism Architecture
Domain Decomposition (Master 9 Nodes)

• Single processor code divided into master process
  (initialization and input/output functions) and
  node/compute processes (all computation)
• Domain decomposition to distribute load over
  compute nodes
• Nodes exchange overlap rows with adjacent nodes
  at appropriate times
• Designed for distributed memory architectures but
  works very well on shared memory machines

Subdomain
MORE PCs MORE PHYSICS in the operational setting
8
RAMS features
9
SENSITIVITY ANALYSES - 1
INCREASINGLY AVAILABLE FROM E.O.
EO - GIS LAND COVER VEGETATION STATUS SOIL
MOISTURE SEA SURFACE TEMPERATURE ATMOSPHERIC
PROFILES CLOUDS AND PRECIPITATION ...
REQUIREMENTES FROM SENSITIVITY ANALYSES
10
SENSITIVITY ANALYSES - 2

• RAMS EXECUTED A SERIES OF TEN-DAYS SIMULATIONS
• STUDY PERIOD WAS JULY 6th TO 15th, 2000
• A CONTROL RUN AND SEVERAL SENSITIVITY RUNS
• SENSITIVITY VERIFIED TO
• LAND COVER TYPE (homogeneous vs heterogeneous
  cover)
• LAND COVER PATCHINESS (area average vegetation
  cover vs. full resolution 1 km patchiness)
• VEGETATION STATUS (leaf area index)

11
CONFIGURATION OF RAMS
Two grids (two-ways nested) Outer coarser grid
(100x97 grid points) with a horizontal resolution
of 20 km Inner grid (90x90 grid points) with
4 km horizontal resolution Vertical resolution
80 m near the ground, up to 1.2 km near the top
with 24 levels In the control simulation, the
generalised Kuo cumulus parameterisation scheme
is activated only over the coarser grid, while
explicit convection is allowed over the inner
grid. Full microphysics package of RAMS is
activated, with CCN0.8109 Sea surface
temperatures from NOAA-AVHRR satellite
observations Topography and land use data from
the U.S.G.S. at 1 km resolution
Punctual verification at 3 WMO sites
12
WEATHER IN THE STUDY PERIOD - 1
The study period, 6 to 16 July 2000, was
characterised by variable weather conditions over
the study area, in particular the Italian
Peninsula, due to the occurrence of several
fronts associated with the development of low
centres over northern Italy and sometimes the
central Mediterranean sea, especially from 10
July onwards. Stronger than normal winds affected
mainly central Italy, the alpine region, southern
France and the seas around Italy and southern
France. Several thunderstorm outbreaks affected
central and northern Italy and surrounding
European areas. As a consequence of
thunderstorms, cloudiness and prevailing westerly
to north-westerly winds, central and northern
Italy experienced below average air
temperatures. Severe events occurred on 10th
July, when a tornado developed in the area close
to the city of Parma in Northern part of Italy
and on 11th-12th July when severe thunderstorms,
hail, and snowfall at relatively low altitudes
(2000 m) affected mainly the north-east part of
Italy. Apart from the 6th,7 th,8th and13th June,
on all the other days of the study period,
thunderstorms occurred over the central and
northern part of Italy.
13
WEATHER IN THE STUDY PERIOD - 2
NOAA - AVHHR visible left to right 6, 10, 15
July 2000 (AVHRR at 12 UTC) LIGHTNING STRIKES
14
The surface scheme in RAMS LEAF-2

• Submodel of RAMS
• Evaluates fluxes of energy, water vapor, and
  momentum between atmosphere and surface
• Solves heat and water balance equations for
  multiple soil layers, multiple snowcover layers,
  vegetation, and canopy air
• Uses mosaic approach to subdivide each surface
  grid cell into multiple landuse types or
  patches
• Each patch contains independent set of soil and
  snowcover layers, vegetation, and canopy air, or
  permanent water (oceans, lakes, etc.)
• Energy and moisture fluxes evaluated between
  each patch and the overlying atmospheric grid cell

SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
15
LEAF-2 vertical levels and patches
SINGLE ATMOSPHERIC COLUMN
NZS 3 NZG 7
Canopy air Vegetation Snowcover Water Soil
NPATCH5 P1 P2 P3 P4 P5

SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
16
RAMS Soil Moisture TransportRichards Equation
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
17
TOPMODEL Wetness IntervalsDarcy Law
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
18
LEAF 2Combined Richards Equation and Darcy Law
W1
W2
W3
W4
W5
Ground surface
Soil layer 5
Soil layer 4
Soil layer 3
Soil layer 2
Soil layer 1
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
19
LEAF - 2Representation of Land Cover Subgrid
Mosaic
20
SENSITIVITY TO VEGETATION TYPE - 1
SUMMER-TIME SEA BREEZE THE HOMOGENEOUS COVER
PRODUCES MUCH LIGHTER BREEZE, DELAYED DEVELOPMENT
AND SHORTER PENETRATION INLAND.
RED HETEREOGENOUS COVER CYAN HOMOGENEOUS COVER
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
21
SENSITIVITY TO VEGETATION TYPE - 2
WINTER-TIME SYNOPTIC FLOW THE VECTOR DIFFERENCES
ARE LARGE (SOMETIMES MORE THAN 60). HETEROGENEOU
S REPRESENTATION PRODUCES HIGHER SPEEDS.
PISTOIA
PISA
RED HETEREOGENOUS COVER CYAN HOMOGENEOUS
COVER BLACK CIRCLES METEO STATIONS
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
22
SENSITIVITY TO VEGETATION TYPE - 3
SUMMER-TIME SEA BREEZE PUNCTUAL COMPARISONS
(EXAMPLE OBSERVED AND SIMULATED WIND DIRECTIONS
AT PISA) SHOW HIGHER AGREEMENT WITH HETEROGENEOUS
COVER
Staz observations Veg heterogeneous
cover NoVeg homogeneous cover
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
23
SENSITIVITY TO VEGETATION TYPE - 4
WINTER-TIME SYNOPTIC FLOW PUNCTUAL COMPARISONS
(EXAMPLE ERRORS OF SIMULATED WIND DIRECTIONS AT
PISTOIA) SHOW HIGHER AGREEMENT WITH HETEROGENEOUS
COVER
Veg heterogeneous cover NoVeg homogeneous
cover
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
24
SENSITIVITY TO VEGETATION TYPE - 5
SUMMER-TIME SEA BREEZE THE INVERSION OF AIR-MASS
TRAJECTORIES IS STRONGLY SENSITIVE TO SURFACE
COVER REPRESENTATION
HETEROGENEOUS
HOMOGENEOUS
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
25
SENSITIVITY TO LAND COVER PATCHINESS - 1
Sensitivity of model simulated screen height
temperature to 1 km patchiness over the finer
grid, at 9UTC on 7 July 2000.
Locally, 2- 3 differences
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
26
SENSITIVITY TO LAND COVER PATCHINESS - 2
Sensitivity of model simulated precipitation to 1
km patchiness over the finer grid, at 9UTC on 7
July 2000.
Several mm differences
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
27
SENSITIVITY TO VEGETATION STATUSleaf area index
(LAI) - 0
Some vegetation specific parameters for LEAF-2
classes
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
28
SENSITIVITY TO VEGETATION STATUSleaf area index
(LAI) - A
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
29
SENSITIVITY TO VEGETATION STATUSleaf area index
(LAI) - B
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
30
SENSITIVITY TO VEGETATION STATUSleaf area index
(LAI) - 1
Sensitivity of model simulated air surface
temperature to LAI values over the coarser grid,
at 9UTC on 7 July 2000.
Locally, 3- 4 differences
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
31
SENSITIVITY TO VEGETATION STATUSleaf area index
(LAI) - 2
Sensitivity of model simulated air surface
temperature to LAI values over the finer grid, at
9UTC on 7 July 2000.
Locally, 3 - 4 differences
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
32
SENSITIVITY TO VEGETATION STATUSleaf area index
(LAI) - 3
SURFACE TEMPERATURE Screen height Temperature and
Humidity are simulated quite well Little
differences (?1.5C) arise until day 7 Later
differences grow, together larger changes in
overall simulated atmospheric states Differences
are larger in other areas of the domain (next
slides) SURFACE DEW POINT DEPRESSION Errors of
simulated rainfall are larger, and differences
reach about 10 mm / day Again, differences are
larger in other areas of the domain (next
slides) DAILY RAINFALL
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
33
SENSITIVITY TO VEGETATION STATUSleaf area index
(LAI) - 4
Sensitivity of model simulated clouds to LAI
values over the finer grid, at 0UTC on 16 July
2000.
Very high sensitivity
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX
34
SENSITIVITY TO VEGETATION STATUSleaf area index
(LAI) - 5
Sensitivity of model simulated precipitation to
LAI values over the finer grid, on 15 July 2000.
Several mm differences
SENSITIVITY TO LAND COVER AND LEAF AREA INDEX