C. Spyrou, G. Kallos and N. Bartsotas
Modelling the regional characteristics of desert
dust sources
UNIVERSITY OF ATHENS FACULTY OF PHYSICS, DIVISION OF ENVIRONMENTAL PHYSICS - METEOROLOGY
ATMOSPHERIC MODELING AND WEATHER FORECASTING GROUP
http://forecast.uoa.gr
7th International Workshop on Sand/Duststorms and Associated Dustfall, Frascati (Rome), Italy, 2-4 December 2013
The presence of desert dust in the atmosphere has considerable
impacts
The impacts are ranging from modification of the radiative forcing
to cloud formation and precipitation
Therefore, perturbations in dust particle production can have
impacts on radiative properties, cloud formation and water budget
These feedbacks depend on the physical, chemical and optical
characteristics of mineral particles which in turn are defined by the
characteristics of the different source areas
Due to the importance of these phenomena the accurate
description of dust sources and production is essential
MOTIVATION
This presentation focuses on defining the areas acting as dust sources
More specifically:
A new surface characterization database has been developed for use in
regional and global atmospheric models
The dataset was implemented into SKIRON/Dust atmospheric model
Evaluation based on actual measurements
OBJECTIVES
INPUT DATASETS
Soil texture database: 16
category hybrid STATSGO/FAO
30″
Vegetation database: 24
category U.S. Geological
Survey 30″
INPUT DATASETS
Clay Content database : (%) clay with resolution 0.08°
SLOPE CUTOFF Slope gradients that exceeded a specific threshold are characterised as
areas with reduced dust uptake characteristics
Threshold was set at 5 degrees slope
Areas that met this rule were downgraded in the classification
SLOPE CUTOFF Slope gradients that exceeded a specific threshold are characterised as
areas with reduced dust uptake characteristics
Threshold was set at 5 degrees slope
Areas that met this rule were downgraded in the classification
• #1: Vegetation “desert” and soil is either “sand” or “loamy sand”
• #2: Vegetation “desert” and soil is “sandy loam”
• #3: Vegetation “desert” and soil is “loam”
• #4: Vegetation “semi-desert” and soil is either “sand” or “loamy sand”
• #5: Vegetation “semi-desert” and soil is “sandy loam”
• #6: Vegetation “semi-desert” and soil is “loam”
• #0: All other combinations.
• High clay amounts reduce the category by 1
• Topographic Slopes greater than 5 degrees produce no dust
NEW CLASSIFICATION CRITERIA
NEW CLASSIFICATION
North Africa
NEW CLASSIFICATION
Arabian Peninsula
NEW CLASSIFICATION
Some Features of SKIRON/Dust Model
Eight-size particle bin scheme (Zender et al. 2003; Spyrou et al., 2010)
Dust source identification and flux production by utilizing additional soil classes and properties (rocky soil, clay amounts) (Reynolds et al., 1999, Spyrou et al., 2010)
Updated dry and wet deposition schemes (Slinn and Slinn, 1980; Kumar et al., 1996; Pandis, 1998)
In-cloud scavenging (Pandis, 1998)
Utilization of high resolution SST (Spyrou et al., 2010)
Incorporation of the RRTMG radiative transfer scheme for SW and LW radiation (Iacono et al., 2008, Clough et al., 2005, Mlawer et al., 1997)
Aerosol – Radiation feedback (Spyrou et al., 2013)
New desert dust classification categories
Sensitivity Tests – Model Setup
• Simulation for 3 months March-April-May 2010 using original dust classification (control)
• Simulation for 3 months March-April-May 2010 using new dust classification (new)
• Evaluation using AERONET data
SKIRON/Dust
Input Data
Initial and boundary
conditions (ECMWF
analysis 0.5x0.5)
Topography (30΄΄x30΄΄) –
USGS. Vegetation
(30΄΄x30΄΄) – USGS. Soil
Texture (30΄΄x30΄΄)
FAO/UNESCO.
Horizontal
Resolution 0.15o x 0.15o
Vertical Levels 38 (up to 22Km)
Timestep 60 sec
Radiation
Algorithm RRTMG
SKIRON computational domain
Sensitivity Tests – AERONET data
Data Selection criteria based on Astitha et al., 2012
• AOD 550 > 0.2
• Angstrom exponent (AE 500 − 870 nm) < 1.2
Test Case 16-18 April 2010
Test Case 16-18 April 2010
Dust particles recorded at Lampedusa
Strong winds at 850 hPa transport
dust particles towards Central and
Eastern Mediterranean on the 15th
of April 2010
Test Case 16-18 April 2010
Dust Flux – Control Dust Flux – New
Test Case 16-18 April 2010
Lampedusa
0
0.2
0.4
0.6
0.8
1
1.2
4/15/2010
12:00
4/16/2010
0:00
4/16/2010
12:00
4/17/2010
0:00
4/17/2010
12:00
4/18/2010
0:00
4/18/2010
12:00
4/19/2010
0:00
4/19/2010
12:00
Date
AO
D 5
50
Data Ctrl New
Test Case 16-18 April 2010
Lampedusa
0
0.2
0.4
0.6
0.8
1
1.2
4/15/2010
12:00
4/16/2010
0:00
4/16/2010
12:00
4/17/2010
0:00
4/17/2010
12:00
4/18/2010
0:00
4/18/2010
12:00
4/19/2010
0:00
4/19/2010
12:00
Date
AO
D 5
50
Data Ctrl New
Better agreement between
actual AOD data and the
updated SKIRON model
Further testing with all 4
stations the entire
simulation period is
needed
Simulations March – April – May 2010
April 2010
Athens - April 2010 - AOD 550
y = 0.6539x + 0.0037
R2 = 0.7857
y = 0.8986x + 0.0032
R2 = 0.8183
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Model
AE
RO
NE
T
New Control Linear (Control) Linear (New)
- Control New
BIAS 0.085 -0.029
RMSE 0.228 0.166
Corr 0.886 0.904
FORTH_CRETE - April 2010 - AOD 550
y = 0.9347x + 0.0038
R2 = 0.8078
y = 0.7062x + 0.0044
R2 = 0.7186
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Model
AE
RO
NE
T
New Control Linear (New) Linear (Control)
- Control New
BIAS 0.061 -0.040
RMSE 0.219 0.196
Corr 0.886 0.847
March 2010 – May 2010
- Control New
BIAS 0.138 -0.015
RMSE 0.425 0.383
Corr 0.844 0.881
- Control New
BIAS -0.134 -0.064
RMSE 0.215 0.205
Corr 0.820 0.828
Lampedusa - March - May 2010 - AOD 550
y = 0.931x + 0.0036
R2 = 0.7762
y = 0.6854x + 0.0038
R2 = 0.7135
0.1
0.3
0.5
0.7
0.9
1.1
1.3
1.5
1.7
1.9
0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9
Model
AE
RO
NE
T
New Control Linear (New) Linear (Control)
Rome - March - May 2010 - AOD 550
y = 1.0724x + 0.0015
R2 = 0.6859
y = 0.8842x + 0.0014
R2 = 0.6738
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Model
AE
RO
NE
T
New Control Linear (New) Linear (Control)
Some Concluding Remarks
• Desert uptake is defined by the source characteristics
• Vegetation, soil texture and clay content are the most important aspects of
dust productive areas
• Steep sloped areas reduce dust production
• A new dataset was created to address these factors on a global scale
• By implementing all the above in a new dataset the model was able to
accurately describe dust production
• The physical, chemical and optical characteristics of dust originate from the
source. Therefore it is important to address these issues in the future as well
The presence of dust particles create significant perturbations in the weather and
climate. Therefore it is essential to accurately describe the desert dust cycle: