Investigating the Relationship between Air Pollutants and Meteorological Parameters in the Agricultural Sector of Mazandaran Province Using Logistic Regression

Paying attention to environmental effects on the production of agricultural yields can be very useful in the direction of sustainable agricultural management. Understanding biological behaviors in the production of pollutants can play an important role in reducing the adverse effects of air pollution. Logistic regression method is considered as a linear developed method to predict air pollution; Time series analysis of parameters affecting air pollutants and addressing how much data is needed in previous times to predict the amount of pollutants one step ahead is an issue that has been less studied. The current study aims to model the process of five important pollutants including carbon monoxide (CO), ozone (O3), particulate matter less than 10 μm in diameter (PM10), sulfur dioxide (SO2) and nitrogen dioxide (NO2) in Mazandaran province, using logistic regression method and time series analysis, to examine the efficiency and flexibility of the methods used in modeling and forecasting these pollutants. In this study, meteorological data from Ramsar, Amol, Babolsar and Nowshahr stations and air pollution data from Gulogah, Ghaemshahr, Sari and Kiasar stations were received daily in the second half of 2017 and 2018, the average of which has been used in data analysis. The findings reveal that NO2 and CO of Gulogah station and O3 of Kiasar station and SO2, NO2 and CO of Sari and Ghaemshahr pollution stations are completely related to the parameters of temperature, relative humidity and wind speed, which indicates the effect of these parameters on changing the concentration of these pollutants. Moreover, based on the patterns of univariate functions of regression equations, valid formulas for estimating logistic relationships between pollutants and meteorological parameters were extracted, according to which, having meteorological parameters in stations, it is easy to predict the pollution of the region.

Pressure on the environment for human activities is important not only environmentally but also economically. In Iran, due to the abundance of energy resources, there is waste and extravagance in their use for economic activities, which leads to an increase in environmental pollution. The current study aims to predict five important pollutants including carbon monoxide (CO), ozone (O3), particulate matter (PM10), sulfur dioxide (SO2) and nitrogen dioxide (NO2). In fact, this study is going to investigate the process of pollutants using logistic regression method and time series analysis to examine the efficiency and flexibility of these methods in modeling.

Studied air pollution measuring stations in this work include Gulogah, Ghaemshahr, Sari and Kiasar stations. Data from Ramsar, Amol, Babolsar and Nowshahr meteorological stations were also used. Meteorological data from synoptic stations and air pollution data from monitoring stations of the Environmental Protection Organization were received daily in the second half of 1396 and 1397, the average of which was used in data analysis. In statistical analysis, the correlation between the parameters was calculated and the correlation relationships were presented. In this study, logistic regression method was used.

With the obtained values ​​for special vectors suitable for each component, a suitable drawing of the relationship between meteorological parameters and air pollutants is created. The results showed that increasing each parameter has an increasing effect on the output, while decreasing each has a decreasing effect on it. A closer look reveals that the O3 pollutant is directly related to the temperature parameter and inversely related to humidity. Moreover, the findings from the test phase as well as the prediction error in the network test phase and the correlation between the actual data and the predicted data indicate that the coefficient of determination R2 between the actual data and the predicted data is equal to 0.62. The relationship between the actual values ​​of O3 and the error obtained from the network test reveal that there is no systematic relationship between the values ​​of O3 and the error and there are different errors for different values ​​of O3. Performing logistic regression and examining the accuracy obtained from them to predict the other four pollutants, it was concluded that there is no systematic relationship between the values ​​of these pollutants and the error and their different values ​​have different errors. Based on this, the error value for these pollutants in the range of 0.4-0 can be assumed to be approximately in a range from 0.07 to 0.1. Besides, studies show that there is a significant relationship between O3 and temperature in Amol meteorological station and Ghaemshahr pollution station. Furthermore, the correlation between meteorological parameters and pollutants in Ramsar meteorological stations and Kiasar pollution shows a significant relationship between SO2 and wind speed, while it indicates a significant relationship between O3 and temperature for Babolsar meteorological stations and Sari pollution. Moreover, correlation study for Nowshahr meteorological stations and throat pollution show a significant relationship between SO2 content and temperature.Based on the correlation results, there is a positive and significant relationship between O3 and temperature in Amol meteorological stations and Ghaemshahr pollution, as well as Babolsar meteorological stations and Sari pollution. Besides, the study of the correlation between meteorological parameters and pollutants in Ramsar meteorological stations and Kiasar pollution proves a negative and significant relationship between NO2 and temperature, and for Nowshahr meteorological stations and bottleneck pollution, shows a significant negative relationship between SO2 and temperature. Therefore, the results clearly indicate that temperature is the most effective factor in the process of creating pollutants in Mazandaran province. This result is consistent with the results of Khorshiddoost et al. (2015) which investigated the relationship between atmospheric parameters and air pollution in Tabriz. However, it contradicts the results of Mahneh (2015) Taste and Kakhki study, which examined the relationship between climate elements and air pollution fluctuations in Mashhad, in which relative humidity was identified as the most influential factor on CO and SO2 pollutants; On the other hand, it is noteworthy that at different stations, different elements have a significant relationship with temperature; This difference in the performance of spatial models for different stations has been confirmed in other studies.

According to the findings from the studied stations, it can be said that NO2 and CO of Gulogah station and O3 of Kiasar station and SO2, NO2 and CO of Sari and Ghaemshahr pollution stations completely indicate a significant relationship among the parameters of temperature, relative humidity and wind speed. The effect of these parameters is to change the concentration of these pollutants. Given the uniformity of changes in station data, it can be inferred that the resulting changes follow general patterns; thus, the stations that have a higher correlation coefficient have closer and more similar patterns and the stations that have a lower correlation coefficient will have unique and special patterns for the same station. Performing logistic regression and examining the accuracy obtained from them to predict pollutants, it was concluded that there is no systematic relationship between the values ​​of these pollutants and the error and their different values ​​have different errors. Finally, based on the purpose of this study, valid formulas to estimate logistic relationships between pollutants have been extracted in order to investigate the efficiency and flexibility of these methods in modeling the pollutant process using logistic regression and time series analysis, based on univariate models of regression equations of models. According to these equations, it is easy to predict the level of pollution in each region by having the meteorological parameters in the stations.