Predicting the Annual Dusty Days in Khorasan Razavi Province Using Spatial-Temporal Analysis

One of the most important environmental challenges in recent years in Khorasan Razavi province is the dust storm phenomenon. A survey on the average number of annual dust storm days in the Khorasan Razavi province shows that it has risen from an average of 6 days to 21 days from 1951 to 2016. And since 1993, it has grown strongly, which shows that reviewing and forecasting the coming years is of high importance. For the purpose of the study Gstat, Space time, SP, Raster, Spdep, and R Google Maps packages of R software are used alongside with Kriging method of spatial-temporal changes to investigate the number of annual dusty days and to predict their occurrence in the coming years.

In this study, the wind speed of 15 meters per second and more and a horizontal view below 1000 meters were considered by the World Meteorological Organization as a dusty day. Then, the SP Data array (Pebesma, 2013) was constructed as a combination of the matrix and vector in STFDF and STF classes (Hengel et al., 2015) according to the following equation:Where the index i refers to the location number (position of view) and the j index to the time number. After that, the empirical Spatial –Temporal variogram were calculated using the Kriging method (Mohammadzadeh, 2012) and were calculated from the following equation:where in:    Represents the set of all couples of observations whose distance is in the neighborhood of the vector  and its time interval near   . Then all the separable and non-separable models were fitted to experimental data model, where the metric variogram with the least average square error was selected as the best model for predicting the number of dust storm days in the year.

The output of the model showed that data up to 5 years have a spatial-temporal dependence, and it is possible to estimate the number of dust storm days until 2022. Therefore, the most important spatial points are the annual number of dust storm days from higher values. They include: Northeast, Northwest and South East of Razavi Khorasan province. The predicted values of annual dust storm days in the central areas of Khorasan Razavi province in 2018 and 2019 shows that these areas will have the smallest number of dust storm days, but will gradually face with an increase in the number of dust storm days.Accordingly, in 2022 most points except the limited points in the east, west and south east will have a significant increase in the number of dust storm days in the central areas.  The results show in 2019, Ghochan, Mashhad, Sarakhs, Sabzevar, Khaf, Torbat Jam and Fariman stations, with a maximum reliability of 95%, will have 40.48.41.46,56,47.41,41 dusty days, respectively. The highest number of dust storm days is related to Sabzevar station with 56 days, and after that Sarakhs with 47 days. Spatial-temporal forecasting trend of the number of dusty days in 2022 indicates that Gonabad Station has the best air quality index of 7 days. Furthermore, the maximum probability of occurrence of dusty days in Ghouchan, Mashhad, Sarakhs, Sabzevar, Khaf, Torbat Jam, and Fariman stations reaches up to 51, 46, 49, 61, 51, 44, and 45 days respectively, and will increase from 39 days to 43 days throughout the province in 2022.

According to the increasing number of dusty days in Khorasan Razavi province, investigating the prediction models is very important. Spatial-temporal Kriging method can use the intrinsic stability of the data to predict the number of dusty days in three dimensions of x, y, and t. This means that we can predict the number of dusty days in different places and times. Indeed, the intrinsic structure of the data plays a significant role in this regard. If the data algorithm is annual, monthly or daily, the output of the model will be proportional to the algorithm. Annual algorithm was able to deliver acceptable results in this research.