Archives of

1. Introduction
Dust has been one of the most important natural hazards and forms of pollution in recent years which has been the focus of many scholars in various dimensions [1]. In the field of dust, various studies have been carried out with different approaches. Source identification studies are one of the most important approaches to the environmental management of this phenomenon. Airborne dust events in the western and southwestern regions of Iran have severe adverse health effects, increasing cardiovascular and respiratory diseases and premature death in vulnerable groups [2,3]. Due to its location in the arid and semi-arid regions of the Earth, Iran is frequently exposed to numerous local and synoptic dust systems [4]. Various studies have indicated that atmospheric factors are directly related to dust phenomena. The nature of this relationship varies by region and is based on the climatic factors that influence weather phenomena. One of these phenomena is atmospheric instability. Humidity also has a significant effect on this phenomenon [5]. The advantage of such studies is to increase the ability to predict the occurrence of this phenomenon and the possibility of better management of dust storms.
Dust storms generally contain small soil particles that can be scattered up to several kilometers above the ground [6]. Strong winds and dust carry large amounts of dirt and sand from dry land, suspend it in the air, and darken it. This air, saturated with earthy matter, forms a cloud that covers the sun or makes it looks like a molten and pale pill [7]. Based on the travel path of particulate matter and its landing site, its origin and direction of travel can be determined [8]. Several studies have been conducted on the spatial origin of dust phenomena and their relationship with climatic factors. Darmany et al [9] used the dust particle source identification method based on the physical and chemical properties of the particles. Wang and Draxler [1] used the HYSPLIT model to identify the origin of particulate matter in North China and Mongolia, proposing management solutions to reduce the occurrence of the phenomenon. Rivandi et al [10] in the atmospheric studies of the dust phenomenon with the Lagrangian-Eulerian model of particle diffusion in the
HYSPLIT model, analyzed the simulation of dust pathways
entering Iran from the west and southwest, identifying
Iraq and Syria as the major sources of these events (more
from 80% of events). Likewise, Bouchlaghem et al [11]
used the HYSPLIT model algorithms to determine the
path of sand and dust storms in the Tunisian desert. In
general, the use of remote sensing data has proven to be
very effective in studying the particulate matter pathways
and identifying their origins [12].
Various studies have been conducted on the main
sources of dust imported into Iran, including the Arabian
Peninsula, the Persian Gulf countries, the north of the
Jordan Desert, the eastern regions of Syria, and parts of
Iraq that move dust and affect Iran as the storm moves
towards the east [13,14]. In particular, the west and
southwest of Iran, including Ilam, seem to be greatly
affected by the dust from the north wind, which is a
warm northwest wind, often blowing during the spring
and bringing large amounts of dust from Iraq [15]. Recent
observations have demonstrated that the occurrence
of this phenomenon is increasing in Iran, especially
in the western, southwestern, and central parts of the
country. Therefore, the dust phenomenon has become
an important challenge and concern in the western and
southwestern provinces [16]. Iran’s geographical situation
and longer border (420 km) with Iraq, as one of the main
sources of dust in the region, and its proximity to other
important dust centers (e.g., Eastern Syria, Saudi Arabia,
and Kuwait) have extremely exposed this province to the
dust phenomena [17]. By studying the origin of dust events
and examining their relationship with climatic factors, it
seems possible to estimate and manage this phenomenon
as accurately as possible. The existence of a link between
climatic indicators and the dust phenomenon has been
confirmed in various studies. The current study analyzed
the relationship between climatic parameters and dust
occurrence and identified the origin of dust events in
Ilam during 2011-2016 using HYSPLIT models.
1.1. Study Area
Ilam province, with 19 086 km2, occupies about 1.4%
of the area of Iran. It is located at 45° 24ʹ to 48°10ʹ East
longitude from the Greenwich meridian and 31° 58ʹ to 34°
15ʹ North latitude from the equator in the westernmost
point of Iran. The center of Ilam province is located at
an altitude of 1427 m above sea level, and to the west,
it shares a common border of 420 km with Iraq [18] as
shown in Figure 1.
According to the information of the General
Environment Department of Ilam, there were 11.4 days
of heavy dust (with an average of more than 250 μg/m3)
in Ilam in an 8-year period from 2004 to 2011 (the period
prior to this survey) on average. The dustiest days with a
frequency of 35% days and the lowest number of dusty days
were recorded in 2012 and 2006, respectively. Moreover,
the heaviest dust and the maximum concentration of
PM10 occurred in 2012, reaching 4970 μg/m3 [19].
2. Materials and Methods
This is a descriptive-applied study aiming to identify
the source of dust particles using the HYSPLIT model
and also the correlation of dust events with climatic
parameters in Ilam province in 2021. The study period
was from 2011 to 2016 (early March 2012 to late
February 2018). Considering that the data sent by satellite
correspond to the months of the Gregorian calendar, the
relevant calculations were made based on this calendar.
Therefore, the time intervals include 2012-2013, 2013-
2014, 2014-2015, 2015-2016, 2016-2017, and 2017-2018.
Four sources were used to collect the data. First, the
meteorological data include various parameters such as
wind speed, relative humidity, dew point, air temperature,
and standard pressure. The amount of PM10 was collected
with three-hour time steps from the Environmental
Figure 1. The Geographical location of Ilam province. Note. Studied area is marked with a star.
Alainejad et al
234 Arch Hyg Sci. Volume 11, Number 3, 2022
Protection Organization for Ilam province. The satellite
images were also prepared by the NOAA specialized site
that stores the data sent by the Landsat satellite. Further,
the required data were prepared based on the date of dust
events (e.g., the time and place of the event) using the
HYSPLIT model. The presented times are based on UTC
and Gregorian calendar, and the model was implemented
at an altitude of 1500 m above the level. The movement
path of particulate matter and its origin in dust events were
identified using HYSPLIT. The HYSPLIT model is a dual
model for calculating the trajectory of dust movement
as well as dispersion and simulation of its settling using
PUFF and particle approaches [20]. Moreover, data time
series and linear regression equations were analyzed
using Excel, SPSS 18, and OriginPro 2018 to determine
the relationship between particulate matter and climate
indicators.
3. Results
Time series of data is required for the analysis and
statistical modeling of dust events in this descriptiveapplied
study. In time series graphs, the general trend
of data values in an interval can be seen. According to
the variables used in the research, the time series graphs
of PM10 data, temperature, relative humidity, standard
pressure, wind speed, and dew point in the interval from
2012 to 2018 are presented in Figures 2, 3, and 4.
One of the goals of the current study was to analyze
the origin of dust events in Ilam province. HYSPLIT data
were used for this purpose. These data were related to
the altitude level below 1500 m. In the first step, the dust
events of Ilam province from March 1, 2012 to February
30, 2018 were determined based on the days when the
average amount of PM10 was more than 154 μg/m3 in each
Figure 2. Comparison of time series data from 2012 to 2018: particulate matter (left) and temperature (right).
Figure 3. Comparison of time series data from 2012 to 2018: dew point (left) and standard pressure (right).
Figure 4. Comparison of time series data from 2012 to 2018: wind speed (left) and relative humidity (right).
Arch Hyg Sci. Volume 11, Number 3, 2022 235
Dust and its relationship with climatic components in Ilam province
day (The basis of dust events is based on the standard
of the Environmental Protection Organization of Iran).
Other climatic factors and horizontal visibility were also
determined in the investigated times. Considering the
number of dust events in the research period, one event
(June 16, 2016), is analyzed, and an example of the output
of HYSPLIT for the origin of dust events in Ilam province
is presented in Figure 5. A total of 165 HYSPLIT outputs
were drawn and analyzed to investigate dust events in
Ilam province from March 2012 to February 2018.
The results of the dust event analysis on this date
showed that the horizontal visibility was below 500 m,
and the highest PM10 concentration was 1700 μg/m3.
The maximum concentration of Pm10 did not occur
concurrently with the maximum decrease in horizontal
visibility. In this event, the pressure trough of the Persian
Gulf spread to the northwest of Iraq. The summer pattern
was dominant, and a high-pressure atmosphere was
observed in the middle level (Figure 5A). The surface wind
speed field of 850 hPa exhibited a maximum speed of 8
m/s, which is much lower than that of the previous cases.
The wind direction is northwest and west (Figure 5B). The
wind rose indicated the direction of the prevailing wind in
the southwest with a speed of 6 m/s (Figure 5C). Further,
a 10-meter wind direction difference and a level of 850
hPa occurred due to the low-pressure summer shallow
structure. It seems that the mountainous structure of the
region played an effective role in changing the direction
of the wind. The satellite image showed the emission of
dust in northwest Iraq. Nevertheless, dust mass can also
be seen in the northwest regions (Figure 5D).
Figure 6 illustrates the summary of the results of the
dust events’ origin analysis in Ilam province in the period
between 2012-2018. These results indicated that out of
165 analyzed HYSPLIT outputs, a total of 69 dust events
(42%) originate from Iraq, 36 (22%) from Syria, 29 (17%)
from Saudi Arabia, 16 (10%) from the Persian Gulf, and
15 (9%) from Jordan.
Figure 6 presents the results of the analysis of dust events
in Ilam province during 2012-2018 based on the number
of dust events by time periods. A total of 666 dust events
(3 hours) were reported, and the highest number of dust
events was 163 in 2012. In addition, the lowest number of
dust events was 31 during 2014-2015. It should be noted
that the total number of daily dust events in this period
was 165. The results of dust event analysis in the period
from 2012 to 2018 revealed that the highest amount of
dust events in Ilam was during June (145 events) and May
(134 events), while the least dust event was reported in
December (6 events). Multiple regression equations and
Pearson’s correlation coefficient were used to investigate
the relationship between the variables of particulate
matter and other climatic components (temperature,
humidity, standard pressure, dew point, wind speed,
and cloudiness). Figure 7 depicts the output of multiple
linear regression between climatic variables, horizontal
visibility, and particulate matter.
Table 1 presents the summary of the results of the
Figure 5. (A) Open sea surface pressure (crosshatch) in hPa and Geopotential Height of 500 hPa (red line) in decameters, (B) Horizontal wind speed (black line)
in m/s, horizontal wind direction (arrow) vertical speed (red crosshatch) in hPa/s at 12 UTC, (C) Wind rose on june 16, 2016, and (D) Visible image of MODIS
Aqua sensor satellite, output of HYSPLIT model on June 16, 2016
B A
D C
Alainejad et al
236 Arch Hyg Sci. Volume 11, Number 3, 2022
multiple linear regression model between particulate
matter and climate indicators. These results showed
that, based on the multiple linear regression model, the
relationship between climate components and particulate
matter is estimated at a medium level (R = 0.589).
The results of the multiple linear regression model and
the beta coefficient between the particulate matter and
climate components are presented separately in Table 2.
These results showed that the temperature has a significant
relationship with the concentration of particulate matter
(P < 0.05). According to the results of the study, most
dust events occurred from May to August (67% of all
dust events), so the existence of such a relationship is
statistically justified.
Based on the results presented in Table 2, which is the
output of multiple linear regression calculations in SPSS,
the linear regression equation is presented as follows:
Y(PM10) = 0.370X1(temperature) + 0.08X2(standard
pressure) + 0.169X3(relative humidity) + 0.006X5 (wind
speed) + 0.22X6(dew point) + 0.023X7 (cloudiness)
In this equation, the values of PM10 (independent
variable-Y) are a function of climatic components
(standard pressure, wind speed, temperature, relative
humidity, dew point, and cloudiness) and horizontal
visibility. The effect coefficient of the absolute value
of the standardized coefficients is the beta coefficient.
Based on this, horizontal visibility with an effect factor
Figure 6. Frequency distribution of dust event origin in Ilam province
between 2012 and 2018.
Figure 7. Output from multiple linear regression between climatic variables, horizon, and particulate matter.
Table 1. Summary of the results of multiple linear regression model between
particulate matter and climate variables
Model R R-square Adjusted R-square Standard error of estimate
1 0.589a 0.347 0.337 467.48618
a. Variables: Wind speed, standard pressure, relative humidity, dew point,
cloudiness, and temperature.
b. Dependent variable: PM10.
Arch Hyg Sci. Volume 11, Number 3, 2022 237
Dust and its relationship with climatic components in Ilam province
of 0.534 and temperature with an effect factor of 0.370
have been identified as related and influencing factors
on particulate matter (P ≤ 0.05). Other variables did
not show a significant effect on particulate matter
(P ≥ 0.05). Pearson’s correlation coefficients were also
used to estimate the relationship between the variables of
particulate matter and climatic components, the results of
which are presented in Table 3. In the correlation analysis,
in addition to the variables of horizontal visibility
and temperature, humidity also showed a significant
correlation with the particulate matter at the level of 0.03
(P ≤ 0.05).
4. Discussion
Dust is one of the most important hazardous atmospheric
phenomena for various fields of application [21]. Today,
the adverse effects of this phenomenon are being
clearly observed in agriculture, public health and safety,
transportation (air and ground), and the like [22].
Therefore, monitoring and investigating dust occurrence
is one of the most useful indicators for decision-makers.
This phenomenon has become a natural process and
problem that can be seen throughout the globe [23]. Ilam
province is one of the semi-arid regions in western Iran,
which has recently faced a significant increase in dust
events. In several studies such as Ansari and Jamshidi [24]
and Su et al [25], the results of the HYSPLIT model for
the origin of the dust phenomenon have been found very
favorable. In a study conducted on the origin of particulate
matter, the results showed that the external origins of the
dust events in Ilam province were areas in Iraq, Syria,
Saudi Arabia, the coast of the Persian Gulf, and Jordan.
Analyzing the origin of dust events in southwestern Iran,
Najafi et al [26] and Javadian et al [27] declared Iraq and
Syria as the main source of most dust events in these
areas, which is consistent with the current study. In the
20-year statistical pattern of dust events in Ilam province,
Pilevaran et al [28] announced that, from June to July, the
dust occurs in western Iran due to the location of the lowpressure
core over the Mediterranean Sea and its landfall
over the deserts of Iraq, Syria, and Saudi Arabia. This can
justify the reason for the dust’s numerous occurrences
in these months in the current study. The results of
the current research also confirmed the existence of a
significant relationship between horizontal visibility and
climatic components.
In the correlation analysis, the variables of temperature,
humidity, dew point, and cloudiness showed a significant
correlation with the particulate matter at the level of 0.03.
Similarly, Zolfaghari et al [29] confirmed the existence
of a relationship between low-pressure systems and dust
events in southwestern Iran, which is consistent with the
results of this study. Jorba et al [30] also obtained similar
results regarding the effect of the high-pressure flow
pattern over Barcelona, Spain. In general, conducting
such studies can play a vital role in finding solutions
for dust management by obtaining information about
the origin and causes of this phenomenon as well as the
factors affecting it.
4.1. Limitations
The defect in the climatic data of the synoptic station of
Ilam is a limitation of the present research.
5. Conclusion
Ilam is one of the semi-arid regions in western Iran,
which has recently faced a significant increase in dust
phenomenon. The results showed that the source of dust
can be identified by examining various components,
including atmospheric pressure, horizontal and vertical
wind, synoptic analysis, and wind rose. In the analysis
of four dust events in Ilam province during 2012-2018,
the exact origin of each event was determined using the
HYSPLIT model. The results also confirmed the existence
Table 2. The results of multiple linear regression model and beta coefficient between particulate matter and climate components
Model
Non-standard coefficients Standard coefficients
t-statistics P value
B Standard error Beta coefficient
1
(Constant) 8496.899 5957.004 - 1.426 0.155
Temperature -25.659 7.948 -0.370 -3.229 0.001
Standard pressure -6.539 5.831 -0.080 -1.121 0.263
Relative humidity -4.909 2.725 -0.169 -1.802 0.072
Wind speed -1.813 13.108 -0.006 -.138 0.890
Dew point -12.529 6.671 -0.220 -1.878 0.061
Cloudiness -7.850 7.515 -0.023 -3.706 0.134
Table 3. Results of Pearson correlation test between particulate matter and climatic components
PM10 Temperature Pressure Humidity Wind speed Dew point Cloudiness
PM10
Pearson’s correlation 1 -0.702** 0.056 0.636* -0.395 -0.891* 0.810**
Significance level 0.000 0.300 0 .012 0 .080 0.019 0.000
Alainejad et al
238 Arch Hyg Sci. Volume 11, Number 3, 2022
of a significant relationship between climatic components
and particulate matter. Considering that the results
determined the external origin of the dust phenomenon
in Ilam province, it is suggested to use these results in
formulating plans to deal with haze.
Acknowledgment
Hereby, the authors would like to express their gratitude to the Ilam
General Environment Department and the Ilam Meteorological
Department for their cooperation in data collection.
Conflict of Interests
The authors declare that there is no conflict of interests.
Ethical Considerations
This paper is extracted from the doctoral dissertation by Mozhgan
Alainejad, a Ph. D student in the field of Environmental Sciences,
Islamic Azad University, Ahvaz branch, with the ethics code
1064823910064381397183196.
References
1. Wang Y, Stein AF, Draxler RR, de la Rosa JD, Zhang X. Global
sand and dust storms in 2008: observation and HYSPLIT
model verification. Atmos Environ. 2011;45(35):6368-81.
doi: 10.1016/j.atmosenv.2011.08.035.
2. Rashki A, Middleton NJ, Goudie AS. Dust storms in Iran–
Distribution, causes, frequencies and impacts. Aeolian Res.
2021;48:100655. doi: 10.1016/j.aeolia.2020.100655.
3. Naimabadi A, Ghadiri A, Idani E, Babaei AA, Alavi N,
Shirmardi M, et al. Chemical composition of PM10 and its
in vitro toxicological impacts on lung cells during the Middle
Eastern Dust (MED) storms in Ahvaz, Iran. Environ Pollut.
2016;211:316-24. doi: 10.1016/j.envpol.2016.01.006.
4. Maleki H, Sorooshian A, Goudarzi G, Nikfal A, Baneshi
MM. Temporal profile of PM10 and associated health effects
in one of the most polluted cities of the world (Ahvaz, Iran)
between 2009 and 2014. Aeolian Res. 2016;22:135-40. doi:
10.1016/j.aeolia.2016.08.006.
5. Cao H, Liu J, Wang G, Yang G, Luo L. Identification of sand and
dust storm source areas in Iran. J Arid Land. 2015;7(5):567-
78. doi: 10.1007/s40333-015-0127-8.
6. Goudie AS. Dust storms: recent developments. J
Environ Manage. 2009;90(1):89-94. doi: 10.1016/j.
jenvman.2008.07.007.
7. Baddock MC, Bullard JE, Bryant RG. Dust source identification
using MODIS: a comparison of techniques applied to the Lake
Eyre Basin, Australia. Remote Sens Environ. 2009;113(7):1511-
28. doi: 10.1016/j.rse.2009.03.002.
8. Rostami D, Hosseini SA. Analysis and tracking dust
phenomenon in south and southeast of Iran by using HYSPLIT
model and the principles of remote sensing. Journal of Spatial
Analysis Environmental Hazards. 2018;5(3):103-19. doi:
10.29252/jsaeh.5.3.103. [Persian].
9. Darmany M, Ara H, Rashki A, Mafi A. Source identifying and
characterizing physical and chemical fine dust in Sarakhs
city. Journal of Geography and Environmental Hazards.
2020;9(3):21-37. doi: 10.22067/geo.v9i3.87970. [Persian].
10. Rivandi A, Mirrokni M, Mohammadiha A. Investigation of
formation and propagation of dust storms entering to the
west and southwest of Iran using Lagrangian particle diffusion
model, HYSPLIT. J Clim Res. 2013;4(13-14):1-16. [Persian].
11. Bouchlaghem K, Chtioui H, Gazzah MH. Analyzing the
impact of Saharan sand and dust storms based on HYSPLIT
algorithm in Tunisian regions. Arab J Geosci. 2021;14(10):834.
doi: 10.1007/s12517-021-07174-4.
12. Sun K, Su Q, Ming Y. Dust storm remote sensing monitoring
supported by MODIS land surface reflectance database.
Remote Sens (Basel). 2019;11(15):1772. doi: 10.3390/
rs11151772.
13. Broomandi P, Dabir B, Bonakdarpour B, Rashidi Y.
Identification of dust storm origin in South-West of Iran.
Environ Health Sci Eng. 2017;15(1):16. doi: 10.1186/s40201-
017-0280-4.
14. Jafari M, Mesbahzadeh T, Masoudi R, Zehtabian G, Amouei
Torkmahalleh M. Dust storm surveying and detection
using remote sensing data, wind tracing, and atmospheric
thermodynamic conditions (case study: Isfahan province,
Iran). Air Qual Atmos Health. 2021;14(9):1301-11. doi:
10.1007/s11869-021-01021-x.
15. Namdari S, Karimi N, Mohammadi G. Pattern analysis of dust
storms in the west and southwest provinces of Iran and its
relation to the Middle East dust storms. J Clim Res. 2017;8(31-
32):103-22. [Persian].
16. Heidari Farsani M, Shirmardi M, Alavi N, Maleki H, Sorooshian
A, Babaei A, et al. Evaluation of the relationship between
PM10 concentrations and heavy metals during normal and
dusty days in Ahvaz, Iran. Aeolian Res. 2018;33:12-22. doi:
10.1016/j.aeolia.2018.04.001.
17. Nikfal A, Ranjbar Saadatabadi A, Rahnama M, Tajbakhsh S,
Moradi M. Evaluation of the WRF/Chem aerosol models-a
dust episode case study. Iran J Geophys. 2021;14(4):1-18. doi:
10.30499/ijg.2020.104569.
18. Alaghehbandan R, Lari AR, Joghataei MT, Islami A, Motavalian
A. A prospective population-based study of suicidal behavior
by burns in the province of Ilam, Iran. Burns. 2011;37(1):164-
9. doi: 10.1016/j.burns.2010.04.010.
19. Amarloei A, Fazlzadeh M, Jonidi Jafari A, Zarei A, Mazloomi
S. Particulate matters and bioaerosols during Middle East dust
storms events in Ilam, Iran. Microchem J. 2020;152:104280.
doi: 10.1016/j.microc.2019.104280.
20. Draxler RR. HYSPLIT4 User’s Guide. Silver Spring, MD:
National Oceanic and Atmospheric Administration; 1999.
21. Vandaele AC, Korablev O, Daerden F, Aoki S, Thomas IR,
Altieri F, et al. Martian dust storm impact on atmospheric H2O
and D/H observed by ExoMars Trace Gas Orbiter. Nature.
2019;568(7753):521-5. doi: 10.1038/s41586-019-1097-3.
22. Viúdez-Moreiras D, Newman CE, Forget F, Lemmon
M, Banfield D, Spiga A, et al. Effects of a large dust
storm in the near-surface atmosphere as measured by
InSight in Elysium Planitia, Mars. Comparison with
contemporaneous measurements by Mars Science Laboratory.
J Geophys Res Planets. 2020;125(9):e2020JE006493. doi:
10.1029/2020je006493.
23. Elsheikh EAA, Islam MR, Habaebi MH, Ismail AF, Zyoud A.
Dust storm attenuation modeling based on measurements in
Sudan. IEEE Trans Antennas Propag. 2017;65(8):4200-8. doi:
10.1109/tap.2017.2715369.
24. Ansari A, Jamshidi R. Identification of sources and tracking
dust storm routes entering from domestic sources to
Arak metropolitan using HYSPLIT model. Environ Sci.
2018;16(1):101-10. [Persian].
25. Su L, Yuan Z, Fung JCH, Lau AKH. A comparison of HYSPLIT
backward trajectories generated from two GDAS datasets.
Sci Total Environ. 2015;506-507:527-37. doi: 10.1016/j.
scitotenv.2014.11.072.
26. Najafi MS, Khoshakhllagh F, Zamanzadeh SM, Shirazi MH,
Samadi M, Hajikhani S. Characteristics of TSP loads during
the Middle East Springtime Dust Storm (MESDS) in Western
Iran. Arab J Geosci. 2014;7(12):5367-81. doi: 10.1007/
s12517-013-1086-z.
27. Javadian M, Behrangi A, Sorooshian A. Impact of drought on
dust storms: case study over Southwest Iran. Environ Res Lett.
2019;14(12):124029. doi: 10.1088/1748-9326/ab574e.
28. Pilevaran R, Rastgoo Z, Nikfal A, Moradpour B. Combining
remote sensing and numerical methods in the study of dust
storms, J Environ Sci. 2021;19(2). doi: 10.52547/envs.34186.
[Persian].
29. Zolfaghari H, Masoumpour Samakosh J, Shaygan Mehr S,
Ahmdi M. A synoptic investigation of dust storms in western
regions of Iran during 2005-2010 (a case study of widespread
wave in July 2009). Geography and Environmental Planning.
2011;22(3):17-34. [Persian].
30. Jorba O, Pérez C, Rocadenbosch F, Baldasano J. Cluster analysis
of 4-day back trajectories arriving in the Barcelona area, Spain,
from 1997 to 2002. J Appl Meteorol. 2004;43(6):887-901.
doi: 10.1175/1520-0450(2004)043<0887:caodbt>2.0.co;2.