Monitoring of Carbon Monoxide (CO) changes in the atmosphere and urban environmental indices extracted from remote sensing images for 932 Iran cities from 2019 to 2021

ABSTRACT Carbon Monoxide (CO) is an important urban pollutant with a relation to transition economies based on emission intensities. In this study, Sentinel-5, MODerate resolution Imaging Spectroradiometer (MODIS), and Landsat-8 images were used to investigate the variations of CO and urban environmental indices and the correlations between them. From the assessed correlations for 932 Iranian cities, it occurred that the assessed indices were all correlated. The highest CO levels were 0.031 in the spring of 2019 and 2020, whereas in 2021 it was equal to 0.030 in both the spring and winter, respectively. In 2019 and 2020 the maximum values of the Enhanced Vegetation Index (EVI) in the spring were 0.181 and 0.183. Exceptionally high Absorbing Aerosol Index (AAI) values of – 0.834 and – 1.0, along with Urban Index (UI) of 0.102 and 0.092, were correlated with recorded spikes in CO level, despite that these seasons’ EVI values were not so abnormal. It was forecasted that in 2030 rises in the CO level by 13.2% in the winter and by 17.5% in the fall are expected, with the simultaneous increase of AAI by 204.5% and 980.2%, and Aerosol Optical Depth (AOD) by 27% and 5% in the winter and spring, respectively.


Introduction
Sulfur Dioxide (SO 2 ), Carbon Monoxide (CO), Nitrogen Oxides (NO X ), Particulate Matter (PM), Ozone (O 3 ), and Volatile Organic Compounds (VOCs) are the principal criterion pollutants in the air (Zhang et al. 2016;Bhatti, Nizamani, and Mengxing 2022).Because human activity is one of the primary causes of releasing these major criterion pollutants into the environment, lockdown conditions are likely to reduce concentrations of these pollutants in the atmosphere (Ghahremanloo et al. 2021;Xiao et al. 2020).The burning of fossil fuels (Choi et al. 2009), biomass burning (van der Werf et al. 2006), soil microbial activity (Yienger and Levy 1995), and lightning (Choi et al. 2009) produce NO X .In the presence of daylight, NO X reacts with VOCs to create ground-level ozone (Wu et al. 2017).For instance, vehicle emissions, biofuel, biomass combustion, and industrial emissions are all sources of both VOCs and CO in China (Guo et al. 2004;Wu et al. 2017).SO 2 and CO are both primary gas-phase pollutants (Chen et al. 2001), and emission intensities have demonstrated a link between them and transition economies (Viguier 1999).While certain PM concentrations are caused by direct PM emissions, others are caused by emission precursors (e.g.SO 2 , VOCs, NO X , and NH 3 ) through secondary production within the atmosphere (Hodan and Barnard 2004) on substantially higher spatiotemporal scales (Köhler et al. 2018).
The origin, physical and chemical characteristics, and detrimental effects of gaseous pollutants on the atmosphere in an urban environment are all different.Black Carbon (BC) aerosols, as well as carbon, nitrogen, sulfur oxides, and tropospheric ozone, are atmospheric pollutants that affect atmospheric chemistry and radiation balance (Badarinath et al. 2007;Zhao et al. 2022).Carbon monoxide (CO) is a critical component of tropospheric chemistry (Badarinath et al. 2007;Safarianzengir et al. 2020).It consumes a significant amount of hydroxy (OH) in the troposphere and is frequently employed as a pollutant tracer.CO helps to create ozone in areas where there is enough NO X , and it impacts ozone concentrations by affecting OH and hydroperoxy radical (HO 2 ) concentration.According to observations, concentrations of CO in distant places of the southern hemisphere are around half those in the northern hemisphere (Hamilton and Mansfield 1991;Badarinath et al. 2007).The principal source of OH in the earth's troposphere is the photolysis of ozone and the subsequent interaction of atomic oxygen with water, which begins multiple photochemical chain reactions (Finlayson-Pitts and Pitts 1999).
When in situ measuring techniques are difficult or even impossible to perform, remote sensing methods give the possibility to estimate emissions and pollutant concentrations (Benaissa et al. 2019;Bhatti et al. 2021a;Mirzaei, Amanollahi, and Tzanis 2019;Badarinath et al. 2009).The development of remote sensing applications for assessing, forecasting, and managing air quality has progressed significantly over the last several decades (Mhawish et al. 2018;Zhao et al. 2021;Yin et al. 2021).Remote sensing instruments have provided useful information on the global distribution of pollutants, their temporal evolution within the atmosphere, and long-range transport (Gupta et al. 2006;Martin 2008;Lee et al. 2015;Zhang et al. 2021).In 2017, the Tropospheric Monitoring Instrument (TRO-POMI) was built and launched aboard the Copernicus Sentinel-5 Precursor satellite (Ghahremanloo et al. 2021;Tian et al. 2020).With a sun-synchronous orbit, TROPOMI analyzes worldwide major atmospheric pollutants (Veefkind et al. 2012).The integration of the disparate data coming from remote sensing methods, such as Landsat measuring the specific features of the surface over an unprecedented period (Mansourmoghaddam et al. 2021;Loveland and Dwyer 2012;Ghahremanloo, Mobasheri, and Amani 2019;Mansourmoghaddam et al. 2022b), or MODerate resolution Image Spectroradiometer (MODIS) encompassing a large number of data outputs for a variety of scientific uses (Masuoka et al. 1998;Ghahremanloo et al. 2021;Zhuo et al. 2022), with the data on atmospheric pollutants allows for their validation, application, and analysis (de Laat et al. 2020;Bhatti et al. 2021b).
Significant changes in emissions present a chance for researchers to study correlations between emissions and pollutant transportation (Ghahremanloo et al. 2021).For example, Ghahremanloo et al. (2021) studied nitrogen dioxide (NO 2 ), formaldehyde (HCHO), sulphur dioxide (SO 2 ), and CO concentrations, as well as the aerosol optical depth (AOD) over the Beijing-Tianjin-Hebei (BTH), Wuhan, Seoul, and Tokyo areas using data from the Sentinel-5P and Himawari-8 satellites, and compared results for February 2019 and February 2020.The authors found that the majority of pollutant concentrations were lower in 2020 than in 2019.In BTH, Wuhan, Seoul, and Tokyo, NO 2 levels were reduced by about 54, 83, 33, and 19%, respectively.From the results, it stemmed that the meteorological variables were not the primary cause of the remarkable decreases in air pollution.Also of interest are the relationships between natural land cover and atmospheric pollutants in cities. Pérez-Ruiz, Vivoni, and Templeton (2020) discovered that mesic landscaping with irrigated turf grass was predominantly regulated by plant photosynthetic activity, whereas a parking lot near a road showed a signature similar to that from automobile emissions.The CO 2 fluxes from the two other locations having a mix of irrigated vegetation and urban surfaces were intermediate (Pérez-Ruiz, Vivoni, and Templeton 2020).Mansourmoghaddam et al. (2022a) studied and predicted changes in the suspended dust in the atmosphere of Qazvin, Iran using remote sensing data.According to the findings of this study, the amount of suspended dust in the atmosphere in Qazvin province increased from 0.461 in 2015 to -0.603 in 2017.This rate subsequently decreased to 0.493 in 2018, before increasing to 0.575 in 2019.In 2020, the quantity of dust floating in the atmosphere fell significantly to 0.536.Rainfall, relative humidity, and vegetation were negatively correlated with AOD values, whereas wind speed, number of frost days, temperature, temperature island fluctuations, and sunlight hours were positively correlated with AOD (Mansourmoghaddam et al. 2022a).Badarinath et al. (2007) conducted simultaneous measurements of black carbon aerosols (BC), carbon monoxide (CO), surface ozone (O 3 ), and nitrogen oxides (NO X ) at Allahabad, northern India, to better understand the influence on pollutant concentrations during fog times.Their findings indicated that BC, NOx, and CO concentrations are greater in the morning and late at night in the study region.Ozone concentrations ranged from 14 ppb to 35 ppb over the day, increasing steadily after dawn, reaching a maximum value by evening time, and then gradually decreasing.BC, NO X , O 3 , and CO concentrations were found to be extremely low during fog periods.It occurred that BC was positively related to CO and negatively to ozone levels on the ground.From the slope of the linear relation between black carbon aerosols and ground-level ozone, it can be stated that for every 1 g/m 3 increase in black carbon aerosol mass concentration, there is a 0.7 g/m 3 reduction in surface ozone.The studies of the concentration of suspended particles in the atmosphere were also investigated in the thermal part of remote sensing by Amanollahi et al. (2013).This study showed the application of the Landsat thermal band to determine the concentration of particulate matter with an aerodynamic diameter smaller than 10 µm (PM10) in large areas (Amanollahi et al. 2013).Mittal et al. (2009) analyzed the influence of stubble-burning events on the air quality in and around Patiala city, India.Their results indicated that a significant increase in aerosol, SO 2, and NO 2 levels occurred during the crop stubble burning periods.
According to health risk assessment, short-term exposure to pollutants (e.g.O 3 and PM10) has been associated with respiratory and cardiovascular disorders, which result in higher hospital admissions and mortalities (Mulenga and Siziya 2019;Brunekreef and Holgate 2002).For example, it is reported that at least 50,000 Chinese individuals die each year as a result of chronic obstructive lung disease (COPD) caused by ozone concentrations that exceed permissible thresholds and exposure duration (Liu et al. 2018), with VOCs, NOx, and CO being the primary ozone precursors responsible for these fatalities (Placet et al. 2000).To lower the tropospheric ozone levels and prevent unnecessary deaths, lowering the levels of precursors, especially CO, is crucial (Ghahremanloo et al. 2021).However, atmospheric pollutants monitoring needs to be performed to understand if a pollution problem is present in the area of interest and what the temporal and spatial scales it concerns.Therefore, the present study aimed to monitor the changes in CO and quantities such as the Absorbing Aerosol Index (AAI), Aerosol Optical Depth (AOD) being an indicator of air particulate dust (Mansourmoghaddam et al. 2022a), Enhanced Vegetation Index (EVI), urban areas (UI), and Normalized Difference Bareness Index (NDBaI) calculated from remote sensing data for 932 cities of Iran during the period 2019-2021 and to assess the correlations between them.Additionally, the forecast of spatially averaged values of used indices for 2030 was performed to check if the problem of increasing atmospheric pollutant levels will become a higher environmental hazard and a much more significant threat to the health of the Iranian citizens.

Study area
In the present study, the changes in the values of various indices related to the levels of atmospheric pollutants for 932 cities in Iran were studied.Iran's country territory spans between 25 and 39°N and from 44 to 63°E (Figure 1).Along the Caspian coast and in the northern woods, Iran's climate ranges from dry, semiarid to subtropical.Temperatures seldom drop below freezing on the country's northern border (the Caspian coastal plain), and the area is humid for most of the year.Annual precipitation for the eastern half of the Caspian coastal plain is 680 mm, whereas the western part, in turn, receives more than 1700mm of rain.Settlements in the Zagros basin (in the western part of the country) are subject to colder weather, harsh winters with average daily temperatures below zero, and considerable snowfall.The eastern and central basins are dry, receiving less than 200 mm of rainfall per year, with predominantly desert terrain.The Persian Gulf and Gulf of Oman coastal lowlands in southern Iran feature warm winters and humid, scorching summers.The yearly rainfall ranges from 135 to 355 mm (Jalili et al. 2013;Palmer 1965).

Data sources
To investigate the relationship between carbon monoxide and climatic parameters remote sensing data recorded by several satellites was used.Obtained information, initially processed by the Google Earth Engine (GEE) system, was ready to be processed and analyzed.The main spatiotemporal properties of the downloaded data with the mentioned aim of their use are presented in Table 1.

Data preprocessing
Before performing any image processing, the necessary radiometric, atmospheric, and geometric corrections were applied and their accuracy and validity were confirmed.Then, each type of data was converted into monthly averages.Extraction of the information from the image multiset requires a step of concatenating the images of the two platforms at the lowest spatial resolution.To do so, the images with the highest spatial resolution were downsampled to make the images' pixels sizes equal (Piqueras et al. 2018).After the obtained data were matched in terms of spatial resolution, the information could be used for further processing and analysis.After that, all the data were analyzed and defective images, with errors or cloudiness, were removed.Finally, the points under study included only the areas where all the data was present, and the information for other spatial points was neglected.

Carbon Monoxide (CO) measurement
To extract carbon monoxide data for Iranian cities, Sentinel-5 data obtained from the TROPOMI sensor was used.The CO worldwide abundance is measured by TROPOMI on the Sentinel 5 Precursor (S5P) satellite using clear-sky and cloudy-sky Earth radiance measurements in the 2.3 m spectral regions of the shortwave infrared (SWIR) section of the solar spectrum.TROPOMI clear sky measurements provide total columns of CO with sensitivity to the tropospheric boundary layer.The column sensitivity varies depending on the light path in foggy atmospheres (TROPOMI 2018).Among the bands in the L3_CO product available in the GEE, the band co_column_num-ber_density, which contains carbon monoxide data based on mol/m 2 , was used.Downloaded data were then averaged on a quarterly scale and evaluated.

Absorbing Aerosol Index (AAI)
To investigate the aerosol in the air for Iranian cities, the Sentinel-5 monthly product was used and analyzed.The UV Aerosol Index (UVAI), also known as the Absorbing Aerosol Index (AAI), is depicted in near real-time high-resolution photography in the L3_AER_AI product available in the GEE.The AAI is built on wavelength-dependent variations in Rayleigh scattering for a pair of wavelengths in the UV spectral region.In the AAI, there is a disparity between observed and simulated reflectance findings.When the AAI is positive, UV-absorbing aerosols, such as dust and smoke, are present.This index may be used to follow the evolution of episodic aerosol plumes caused by dust storms, volcanic ash, or biomass burning (Google Developers 2022).Obtained data was already a monthly average from GEE and, similar to other downloaded data for Iranian cities, it was analyzed for seasons coming from three years.

Aerosol Optical Depth (AOD) measurement
AOD is an important measurement in the study of dust and refers to the distribution of dust aerosols in the atmosphere (Ackerman et al. 2004) and the mass size of suspended particles in an atmospheric column (Van Donkelaar, Martin, and Park 2006;Chu et al. 2002), which is dominated by near-surface emission sources (Hu 2009).Therefore, the values of this product indicate suspended particles in the atmosphere (regardless of their origin and path of motion).This is the reason why the term "suspended dust in the atmosphere" was used in the present study (Mansourmoghaddam et al. 2022a).AOD is a dimensionless index with values in the range between zero and one.The values from 0.1-0.5 are a sign of clear air, whereas higher values are a sign of the air being dusty (Ackerman et al. 2004).To calculate AOD for Qazvin province, the ready-made product MCD19A2 version 6 based on the data from MODIS satellite sensors was used, which determines the geometric properties and characteristics of the atmosphere (Lyapustin 2022).From the data for the whole of Qazvin province from the Google Earth engine system, the monthly and annual averages were calculated.

Enhanced vegetation index (EVI) index
To evaluate the vegetation conditions in Iranian cities, the product of EVI MODIS was used for the years from 2019 to 2021.This index, while properly showing the temporal and spatial changes of vegetation in the region, also reduces the effects of soil and dust on vegetation (Pettorelli et al. 2005).It was extracted from MOD13Q1 MODIS images of the Terra satellite with a 16-day temporal resolution.The criteria for capturing these images are low cloud, low viewing angle, and the highest value of this index in pixels (Didan 2022).MOD13Q1 images were downloaded from the USGS NASA website, and after applying a scale factor of 0.0001, the EVI value for each pixel was obtained, and then monthly and quarterly averages were calculated.

Urban index (UI)
To obtain the UI index the Landsat-8 T1_SR product, which makes use of the urban area's brightness relationship (Kawamura 1996;Ali, Hafid Hasim, and Raiz Abidin 2019;Mansourmoghaddam 2021), was downloaded from GGE.Then the 5th band, Near Infrared (NIR), and 7th band, Shortwave Infrared-2 (SWIR2) were extracted and the UI index was calculated as (Kawamura 1996): The changes in the UI index values were then obtained for each period by calculating the difference between the calculated period and the previous period.

Normalized difference bareness index (NDBaI)
The Landsat 8 Thermal Infrared Sensor (TIRS) thermal band (10) was downloaded from GEE to calculate the NDBaI index to identify the barren lands as (Tucker 1979):

Prediction
The Time Series Smoothing Algorithm (TES) is a simple approach for smoothing the time series data.Its usefulness for predicting climatic parameters has already been proven in various studies (Mansourmoghaddam et al. 2022a;Indriani et al. 2020).In this method, the weight reduction is attributed to the data exponentially over time (Gardner 1985).The present study has used the TES algorithm to predict the change in the spatially averaged values of used parameters for Iran territory.To smooth the time series and remove the high-frequency information, the TES algorithm was applied to the data three times (Kalekar 2004).For this purpose, a sequence of research timeseries data was considered as the KT algorithm.Then, the seasonal change cycle L was selected according to the predicted period of the research, which was 2 years, and the prediction process started from T = 0. Thus, the TES algorithm determined the most optimal estimate of future time data at time T + 1 (i.e.smoothed value) as output, using the time series data available at time T, which was used as KT input, and the data trend line calculation.The final output was the predicted estimate for the value of K at time T + M, where M had to be greater than zero, based on the data up to time T (with T included) (Dev et al. 2018).To evaluate the accuracy of the algorithm, the area of defined land cover classes (the classes for which the area was known) in 2019 and 2020 was predicted by the time series data for previous years, and the results were evaluated by the statistical criterion of Root Mean Square Error (RMSE).

Flowchart of the data processing
The schematic data processing diagram is shown in Figure 2.

Results
To evaluate the spatiotemporal changes in CO, AAI, AOD, EVI, UI, and NDBaI, the maps of the average seasonal values from the period 2019-2021 of these indices were plotted and presented in Figure 3.The trends fitted to the time series of the spatially averaged values of indices, presented in Figure 4, indicate an overall decrease in all calculated indices.During the study period from 2019 to 2021, the AAI index decreased by 37.2%, AOD by 15.6%, UI by 48%, NDBaI by 103.8%,EVI by 24.4%, and CO by 23.4%.
According to the results presented in Figure 5, where the spatially averaged seasonal values of calculated indices are presented for the separate years from the period 2019-2021, the smallest value of average AAI in Iran cities in 2019 was -0.738 during the fall (Hammer et al. 2016) and the highest was -0.287 in the summer.In the winter it was also quite low, at -0.659.In 2020 it was the highest in the summer (−0.586) and lowest in the winter (−0.923), whereas in the fall of 2021, the AAI was positive and high (0.576).The lowest seasonal AAI value in 2021 was observed in the winter (−1).The average amount of suspended dust in the atmosphere (AOD) indicated that the highest rate could be observed in the summer (0.188, 0.191, and 0.191 in 2019, 2020, and 2021, respectively), whereas the lowest in the winter (0.092, 0.09, and 0.098 in 2019, 2020, and 2021, respectively).The same was for NDBaI, as the highest rate could be observed in the summer (0.04, 0.049, and 0.049 in 2019, 2020, and 2021, respectively), whereas the lowest in the winter (−0.048, -0.057, and -0.044 in 2019, 2020, and 2021, respectively).In 2019 the UI value was the highest in the summer (0.15) and the lowest in the fall (0.1), in 2020 the highest was in the winter (0.151) and the lowest in the summer (0.086), whereas in 2021 the highest was in the spring (0.102) and the lowest in the fall (0.085).The EVI index in 2019 showed its highest and lowest values in the spring (0.181) and winter (0.117), respectively, in 2020 the lowest value (0.109) was in the winter, and the highest in the spring (0.183), whereas in 2021 the highest and lowest values were observed in the summer and winter (0.169 and 0.114, respectively).In 2019 CO values showed their highest level (0.031) in the spring and the lowest level (0.027) in the fall, in 2020 the highest CO value was in the spring (0.031) and the lowest in the winter (0.027), whereas in 2021 the highest level was in the winter and the spring (0.030) and lowest (0.028) in the fall.The average annual values were estimated, and for 2019 they were equal to CO: 0.029, AAI: -0.575, AOD: 0.136, UI: -0.134, NDBaI: -0.011, and EVI: 0.146, for 2020 -CO: 0.028, AAI: -0.780, AOD: 0.028, UI: 0.118, NDBaI: -0.005, and EVI: 0.139, whereas for 2021 -CO: 0.022, AAI: -0.361, AOD: 0.115, UI: -0.07, NDBaI: 0.01 and EVI: 0.111.
In Figure 6 the results of the correlation analysis between CO values and other climatic parameters for Iranian cities are presented for separate years from the period 2019-2021.In 2019 a positive and relatively high correlation between CO and UI and EVI can be observed, with R = 0.7 and also with AOD (R = 0.6).CO also showed a high negative correlation with NDBaI (R = −0.6).Although CO levels in 2019 were not directly related to AAI (R = 0.0), the AAI was positively correlated with UI and NDBaI, and negatively correlated with EVI.In 2020 similar trend can be observed, as CO had a positive correlation with AOD (R = 0.7), and with the UI and EVI (R = 0.6).Contrary to 2019, in 2020 CO was positively correlated with AAI (R = 0.5).The AAI values were correlated with UI, NDBaI, and EVI, which were positive correlations in the first two cases (R = 0.6) and negative in the last case (R = −0.6).In 2021 a positive correlation of CO with AOD and UI (R = 0.7) and with EVI of (R = 0.5) was observed.In this year the CO measurement had the highest correlation with AAI from the whole study period (R = 0.6).The UI and NDBaI were positively correlated with AAI (R = 0.6), whereas EVI was negatively correlated with AAI (R = −0.6).

Validation
To evaluate the accuracy of the algorithm proposed to predict the values of studied indices for Iranian cities in 2030, firstly the validation of the results had to be performed.The results of the validation of the TES model proposed for the prediction of the studied indices values for Iranian cities are presented in Table 2, where the actual (ACT) values from remote sensing data and predicted (PRE) values for 2021 for Carbon Monoxide (CO), Absorbing Aerosol Index (AAI), Aerosol Optical Depth (AOD), Enhanced Vegetation Index (EVI), Urban Index (UI), and Normalized Difference Bareness Index (NDBaI) are presented.From the results, it occurred that the proposed model had an average RMSE of 0.05 for the winter, spring, and summer and 0.24 for the fall (Table 2).This model also had the lowest error in predicting CO with an average RMSE of 0.001 and the highest error in predicting AAI, with an average RMSE of 0.5.
In Figure 7   According to the results of the forecast, CO values in the winter and fall will increase by 13.2% and 17.5%, respectively, and in the spring and summer will decrease by 15 and 17.6%, respectively.AAI values for the winter, spring, and summer seasons are projected to be lower by 204.5%, 150.3%, and 238%, respectively, and in the fall it tends to be 980% larger.In 2030, AOD values are predicted to rise in all seasons, with the highest growth of 70.7% in spring, and 27% in the winter.The future values of the UI index were predicted to be larger, with the highest rise of 349% for the summer season, and then 210.7% for the winter season.NDBaI values are also projected to be larger in all seasons of 2030, with the highest rise of 436% in the spring.Vegetation (EVI) values, however, are forecast to decrease in all 2030 seasons in Iranian cities.This decrease was the highest in the

Discussion
All of the computed parameters revealed a decline in the studied indices.Between 2019 and 2021, the AAI index fell by 37%, the AOD by 15.6%, the UI by 48%, the NDBaI by 1.4%, the EVI by 24%, and the CO by 23%.From the obtained results, the reduction of CO in Iranian cities had been confirmed (Raispour and Khosravi 2020).The results of the present study indicated that higher amounts of CO in 2019 and 2020 were observed in the spring (0.031).In these two years, the highest values of vegetation expressed through EVI were 0.181 and 0.183, respectively, in the spring, which is connected to the rise of background CO by vegetation (WHO Regional Office for Europe 2000).Previous studies also showed a positive relationship between vegetation indices and CO 2 (Lim, Kafatos, and Megonigal 2004).These two years also had the second-highest spring UI (0.148 and 0.135) and NDBaI (0.001 and 0.017) values.The positive correlation of UI with CO in 2019 (R = 0.7) and 2020 (R = 0.6), as well as the negative correlation with NDBaI (R = −0.6) in both these years, confirmed that decreasing barren lands area by the expansion of the cities was a reason for the observed increase in CO values.While the present study showed the correlation between urban expansion and CO, previous studies have confirmed the positive relationship between urban development and CO 2 rise (Wang et al. 2015).In 2021, the highest amount of CO was jointly observed in the spring and winter (0.030), which is consistent with the results of previous studies (Safarianzengir et al. 2020).Although EVI values in these seasons were not so high, a very high AAI (−0.834 and -1), along with high UI values (0.102 and 0.092) were probably affected by the same mechanisms that were responsible for such abnormal CO increases.The correlation between CO and AAI in 2021 (R = 0.6) was the highest one between these quantities during the whole study period, which also supports the above statement.Since the values obtained from AAI were negative, which indicates non-absorbable aerosols (Hammer et al. 2016), which in turn are largely related to Secondary Organic Carbon (SOC) (Hammer et al. 2016;Flores et al. 2014;Liu et al. 2015) or clouds (Kooreman et al. 2020), the increase in SOA values could lead to an increase in CO in 2021 and vice versa.In general, the negative correlations between EVI and NDBaI (not significant), EVI and UI, CO and NDBaI (not significant only in 2021), AAI and EVI indices were found, whereas the correlations between UI and NDBaI indices, CO and UI, CO and EVI, AOD and CO, AOD and UI (not significant), AOD and NDBaI (not significant), AOD and EVI (not significant), AAI and AOD (significant only in 2019), AAI and CO (not significant only in 2019), AAI and UI, AAI and NDBaI were always positive.The CO forecast for 2030 indicates an increase in the winter and fall by 13.2% and 17.5%, respectively, compared to 2021.In the winter and fall, the forecast indicated that a 204.5% decrease and 980.2% increase in AAI should be expected, whereas AOD will rise by 27% and 5%, respectively.The UI, meanwhile, is projected to increase by 10.7% in the winter.Although UI growth had been predicted for the summer of 2030, likely, the predicted increase in AOD and a relative decrease in EVIs will also have an impact on the increase of the winter CO in 2030.Obtained findings indicated that a direct impact of urban growth and an increase of CO in Iranian cities should be expected in the future.

Conclusions
The findings of the performed analyses are that CO levels were highest in the spring of 2019 and 2020 (0.031), while in 2021 the highest level of CO was 0.030, which occurred in both the spring and winter.The maximum values of vegetation in 2019 and 2020 expressed through EVI were 0.181 and 0.183, respectively.Although these seasons' EVI values were not the highest, other factors such as exceptionally high AAI values (−0.834 and -1 for 2019 and 2020, respectively) combined with high UI values (0.102 and 0.092 for 2019 and 2020, respectively) were affected by the same mechanisms that were responsible for the rise of the CO levels in these years.From the correlation analysis, it resulted that the EVI and UI, CO and NDBaI, and AAI and EVI indices were always negatively correlated, whereas the UI and NDBaI, CO and UI, CO and EVI, AOD and CO, AAI and AOD, AAI and CO, AAI and UI, AAI and NDBaI indices were always positively correlated.
When compared to the values from 2021, the CO estimated for 2030 shows a 13.2% increase in the winter and a 17.5% increase in the fall.AAI index is forecasted to be 204.5% lower in the winter and 980.2% higher in the fall, AOD will rise by 27% and 5% in these seasons, respectively, whereas the UI, is expected to rise by 10.7% in winter.The increase in CO in the winter of 2030 is probably related to the projected increase in AOD and a proportional drop in EVI for this season.Such a large increase in atmospheric pollutant levels forecasted for 2030 for sure will become a higher environmental hazard and a much more significant threat to the health of the Iranian citizens, and therefore mitigation actions should be implemented.The results of this study can serve as an indicator for the Iranian policymakers to closely monitor the atmospheric pollutants levels and properly plan their actions.
It is foreseen that in future research the investigation of the changes in the CO values and their relation to other environmental pollutants will be performed for Iranian cities to discover the possible relationships between them.The assessment of the impact of confinement related to the lockdown linked to the COVID-19 pandemic on air quality in Iran is of special importance and it is planned that it will be analyzed in the subsequent article.

Figure 1 .
Figure 1.The map of Iran with the marked location of the cities (top) and land cover types from MODIS (MCD12Q1) image from 2019 using the International Geosphere-Biosphere Programme (IGBP) classification (bottom)

Figure 2 .
Figure 2. The flowchart of the data processing the comparison of the values of spatially averaged values of studied indices for 2021 and 2030, and the average seasonal values of predicted indices for Iran in 2030 are presented.

Figure 3 .
Figure 3. Maps of average seasonal values of Carbon Monoxide (CO), Absorbing Aerosol Index (AAI), Aerosol Optical Depth (AOD), Enhanced Vegetation Index (EVI), Urban Index (UI), and Normalized Difference Bareness Index (NDBaI) in Iran from 2019 to 2021 (white areas represents data that was deleted because of the errors or cloudiness images)

Figure 4 .
Figure 4.The trend of the time series of the spatially averaged values of Carbon Monoxide (CO), Absorbing Aerosol Index (AAI), Aerosol Optical Depth (AOD), Enhanced Vegetation Index (EVI), Urban Index (UI), and Normalized Difference Bareness Index (NDBaI) for Iranian cities for the period 2019-2021

Table 1 .
Properties of remote sensing data used in the study

Table 2 .
The actual (ACT) values from remote sensing data and the predicted (PRE) values from the TES model proposed for the prediction of the studied indices values for Iranian cities for 2021 for Carbon Monoxide (CO), Absorbing Aerosol Index (AAI), Aerosol Optical Depth (AOD), Enhanced Vegetation Index (EVI), Urban Index (UI), and Normalized Difference Bareness Index (NDBaI)