عنوان مقاله [English]
Mineral dust is an aerosol, mostly affecting radiation budget, temperature change, cloud formation, convection, and precipitation, both directly and indirectly. During the two recent decades, new sensors and models have become available, allowing new research activities on dust. Important studies considered Atmospheric Optical Depth (AOD) as the key parameter for remote sensing and modeling of dust. The available model with the help of satellite and ground-station datasets have been used to detect and characterize mineral dust phenomenon in affected regions and dust sources. Nonetheless, regional classification over entire Iran, using remote sensing parameters, is still lacking.
Materials and Methods
The present study aims at modelling and detecting homogeneous areas of high dust concentration in Iran, using dust AOD at 550 nm from the MODIS satellite Aqua and Terra sensors (2003-2012) with a spatial resolution of 0.125°×0.125° or about 14 km2.
Among vector techniques, S-mode application, as a Principal Component Analysis (PCA) or an example of Empirical Orthogonal Functions (EOFs), is the most applicable and controversial method of classification for doing so. The S-mode analysis was applied on a matrix, made of satellite observations at regularly spaced grid points of daily AOD values for ten years (2003-2012). The S-mode analysis was applied to identify the geographycal distribution of high dust concentrations. PCA of the n x m matrix was utilized and the scree test and North's rule were used to cut-off the statistically relevant components to be kept. Finally, in order to determine the best theoretical representation of the data, physical relations got embedded within the input matrix. Also to localize the territory to simpler structures, specific modes of the residual components got rotated by varimax. Varimax rotation means that each component had a few large loadings and many small loadings. This helps in the process of interpretation in case the results are prone to high values of the explained variance. The rotated patterns, however, illustrate simpler, more interpretable, and rational structures of mineral dust as principal modes. Identification of sub-regions and extreme dust loading was performed, using dust AOD values, assuming arbitrary thresholds of 87% and 95%, respectively. Therefore, the first threshold was used to determine sub-regions. Consequently, the regions would have zero overlapping. The second threshold helped extracting the days with extreme AODs of each region. Herein, the Kolmogorov-Smirnov (K-S) test was used to infer whether the regional mean time series PCs of each different sub-region were statistically different or not.
Results and Discussion
The spatial map-patterns of dust, amounting to 91% of AOD variability, had been divided into six subregions on Iran that were the major centers, affected by the dust. All of the sub-regions coincided with regional map-patterns, depending on the distance and proximity to dust sources around the territory. Therefore, overlapping of identified dust areas related to dust extremities in each of Iran’s regions showed that the dominant dust patterns of Iran were under the infuence of expansion and growth of dust extremities. The geographical location of source areas and the special dynamic conditions over mid-eastern atmosphere of Iran have been influenced by severe storms originating from the Karakum Desert. The northeast region is affected by the dust plume, from the Karakum Desert to Tabas Desert in the southeast of Kavir Desert. These results showed that ground-based station studies, albeit long-term, had not been able to detect the northeast region as a distinctive region under infuence by southward dust plume. The same was true for the Central plateau, East and Southeast regions. In return, more focus was directed at the role of 120-day winds as a main cause of dust transport. Considering the mentioned reasons, previous studies had not divided the borderline regions across Iran. Meanwhile, weakness and intensity of dust-affected areas showed that the multiplicity and adjacency of dust flow to southeastern and eastern parts of the country were different, playing a decisive role in the formation of east and southeast subregions. The shortcoming were observed for west-northwest and southwest regions, too. In a case study (on horizontal visibility), not only were the researchers capable of distinguishing the dusty subregions because of limited observations in the interested area, but also could not analyze the identified subregions, based on corresponding seasolality and extremities, identified by 95% and 87% in each region, respectively. The detected extremes showed that the identified sub-regions were a function of volume, growth, and expansion of dust particles, originated from the dust source regions across the Middle East and southwest Asia. Finally, the classification techniques showed that technical conversion of a dynamic phenomenon, like dust, into simpler and more meaningful physical structures geographically revealed a simple and interpretable understanding of dust distribution inside the territory of Iran. Morever, the use of remotely-sensed data utilized in the present study highligted the sub-regional distribution of dust over Iran, neglected by previous studies that provided a description of a dynamic process that was complementary to the ground-based observation analisys. In some cases, a day event only based on ground-based observations may have had a high dust AOD with very horizontal visibility, capable of being ignored due to the height of the dust layer. Therefore, the used technique integrated the knowlegde of dust based on grounded-measurement, providing a large scale view of dust advection and diffusion.
The study results showed that extraordinary dry conditions inside Iran, combined with outside dusty sources, had made the country to be influenced by high mineral dust aerosols. In addition to domestic sources of dust, the study highlighted that the mineral dust conditions in Iran were influenced by several arid and semi-arid sources beyond its boundaries acting as dust sources. The subregions that form the spatial patterns of dust distribution in a six-distinct region of northeast, west-northwest, southeast, southwest, central, and east Iran were affected by high dust aerosol optical depth (AOD). They were major centers of activity and high gradient areas (regions affected by dust) that followed a trend-distinctive seasonality. This managed to illustrate identified sub-regions’s seasonalities and regional extremes by remotely-sensed data of atmospheric optical depth. The study results demonstrated that dominant spatial dust patterns of Iran were functions of growth and expansion of dust extremes from source regions in the Middle East and southwest Asia. As a result, the present study showed that technical conversion of a dynamic phenomenon, such as dust, to simpler structures paved the way towards a geographical interpretation of dust distribution.