Spatiotemporal variability/stability analysis of NO2, CO, and land surface temperature (LST) during COVID-19 lockdown in Amman city, Jordan

ABSTRACT The massive lockdown of human socioeconomic activities and vehicle movements due to the COVID-19 pandemic in 2020 has resulted in an unprecedented reduction in pollutant gases such as Nitrogen Dioxide (NO2) and Carbon Monoxide (CO) as well as Land Surface Temperature (LST) in Amman as well as all countries around the globe. In this study, the spatial and temporal variability/stability of NO2, CO, and LST throughout the lockdown period over Amman city have been analyzed. The NO2 and CO column density values were acquired from Sentinel-5p while the LST data were obtained from MODIS satellite during the lockdown period from 20 March to 24 April in 2019, 2020, and 2021. The statistical analysis showed an overall reduction in NO2 in 2020 by around 27% and 48% compared to 2019 and 2021, respectively. However, an increase of 7% in 2021 compared to 2019 was observed because almost all anthropogenic activities were allowed during the daytime. The temporal persistence showed almost constant NO2 values in 2020 over the study area throughout the lockdown period. In addition, a slight decrease in CO (around 1%) was recorded in 2020 and 2021 compared to the same period in 2019. Restrictions on human activities resulted in an evident drop in LST in 2020 by around 13% and 18% less than the 5-year average and 2021 respectively. The study concludes that due to the restrictions imposed on industrial activities and automobile movements in Amman city, an unprecedented reduction in NO2, CO, and LST was recorded.


Introduction
The first case of COVID-19 was reported in China in December 2019, and since then the disease has spread very quickly throughout the world (Li et al. 2020).Since March 2020, the COVID-19 has deeply hit the world and caused a dramatic increase in cases and mortality rates.This obliged the World Health Organization (WHO) to announce the novel COVID-19 disease as a pandemic on 11 March 2020 (WHO 2020b).The danger of the COVID-19 comes from the rapid human-to-human transmission and thus several governments around the world imposed lockdown operations on human activities to stop its rapid spread.
Like other countries, Jordan enacted an emergency response particularly when the first case of COVID-19 was detected early in March 2020.The Jordanian government took a set of quick and tough measures against COVID-19 and decided to lock down the human activities to the minimum capacity for controlling the spread of the pandemic across the country on 20 March 2020.The comprehensive lockdown of all human activities in all Jordanian cities includes closing the schools, universities, restaurants and cafés, public transportations, public and private service sectors, worship places (mosques and churches), and many other industrial activities.For almost two months, Jordanian people had to get their daily needs by foot because they were prohibited from using their private vehicles and moving between places.
As the human and industrial activities were substantially reduced by approximately 95% during the comprehensive lockdown period, Land Surface Temperature (LST) and the emitted atmospheric pollutants such as CO and NO 2 were expected to be significantly influenced during that period particularly in large populated and industrial cities (Li et al. 2020).To address the impact of the COVID-19 lockdown on air quality, several investigations (e.g.Chen et al. 2020;Zangari et al. 2020;Dutheil, Baker, and Navel 2020;Freitas et al. 2020;Isaifan 2020;Muhammad, Long, and Salman 2020;Nakada and Urban 2020;Sharma et al. 2020;Xu et al. 2020;Alqasemi et al. 2021) have been conducted in different cities around the world.These studies found a noticeable decrease in air pollutants during the lockdown period.More specifically, some studies (Abdullah et al. 2020;Dantas et al. 2020;Zambrano-Monserrate, Ruano, and Sanchez-Alcalde 2020;Tobías et al. 2020;;Vîrghileanu et al. 2020) linked the lower levels of traffic with air pollutants.Other studies such as Isaifan (2020), Wang and Su (2020), and Alqasemi et al. (2021) compared the air quality before and during the lockdown took place.For example, Wang and Su (2020) found that although some of the main pollutants were substantially reduced, there was a significant increase of PM2.5 during the lockdown period by around 20%.Furthermore, Briz-Redón and Serrano-Aroca (2020) analyzed the influences of climatic factors (e.g.temperature and humidity) on the spread of the COVID-19.However, no significant relationship was found between COVID-19 cases and climatic factors.
Apart from linking the spread of COVID-19 to temperature and humidity, Urban Areas Heat islands (UHI) have significantly been influenced by the reduction of anthropogenic activities.In high populated and industrial cities, LST was used to estimate the Surface Urban Heat Islands (SUHI) using remotely sensed satellite datasets such as Moderate Resolution Imaging Spectroradiometer (MODIS) and Landsat.The LST which is defined as the surface temperature of the Earth's skin is highly correlated to the changes in the urban environment (Moradi, Salahi, and Masoodian 2018).Previous studies have shown that there is a strong relationship between air pollutants and LST (Weng and Yang 2006;Feizizadeh and Blaschke 2013;Kahya et al. 2016).In this context, the changes of LST due to the low anthropogenic activities and its relation to the SUHI were investigated (Ghosh et al. 2020;Sahani, Goswami, and Saha 2020;Alqasemi et al. 2021).For example, Alqasemi et al. (2021) investigated the impact of the COVID-19 lockdown on SUHI in the Northern Emirates of the United Arab Emirates.The results revealed that the SUHI level declined by approximately 19.2% during the COVID-19 lockdown.A similar study was conducted in Pakistan by Ali et al. (2021) and the results showed a visible decrease in SUHI in megacities.However, these previous studies focused on the comparison between the concentration of air pollutants and LST before and during the lockdown period without investigating their spatiotemporal variability/stability.
Since NO 2 and CO are the major air pollutants as defined by various national environmental agencies across the world, which adversely impact human health (Mills et al. 2015), contribute to the tropospheric ozone and aerosol formation (Monks et al. 2015;Lane, Donahue, and Pandis 2008), and impact climatic cycles (Lin et al. 2015), this study was undertaken to investigate the potential impacts of COVID-19 lockdown operations on these variables as well as LST in Amman city.To our knowledge, no studies have examined the impact of COVID-19 lockdown on NO 2 , CO, and LST throughout Jordanian cities.The objectives of this study are to (i) demonstrate the spatiotemporal variability of NO 2 , CO, and LST during the lockdown period (20 March to 24 April) in 2020 and 2021, and the typical period in 2019, (ii) analyze the temporal persistence of these three variables during the lockdown period.This study provides an improved understanding of the spatiotemporal variability of major air pollutants for developing an action plan to improve air quality in Amman city.

Study area
Amman is the capital of Jordan and has the largest economic and trading activities with a population of approximately 4 million (i.e.40% of the total population in Jordan).Geographically, the city covers approximately (1680 km 2 ) and is located between latitude (31° 34ʹ 35ʺN and 32° 5ʹ 5ʺN) and longitude (35° 43ʹ 11ʺE and 36° 11ʹ 39ʺE) (Figure 1).
The elevation of the city ranges from 700 m to 1150 m above sea level.Thus, the city was built on seven hills.The climate is categorized as a Mediterranean climate with hot dry summer and modest-wet winter.The annual averages of rainfall and temperature are 300 mm and 17°C, respectively.In addition, the city may exhibit dust storms coming from the Red Sea Trough during spring and autumn followed by unstable weather conditions.On the other hand, heat waves may occur during the summer season and lead the temperature to hit 45°C.Since the city has different reliefs, it consists of different microclimates and almost every district exhibits its weather condition.The existence of microclimate is not only because of the wide range of topographic factors but also due to anthropogenic activities such as trading, industrial, social, sports, and tourism.In the last decades, emission sources have been accelerated due to the high increase of economic, political, industrial, and vehicular leading to a surge in urbanization rate and energy consumption in all metropolitan cities including Amman.For example, according to the ministry of transportation, the number of registered vehicles in Amman has tripled between 2000 and 2020.This inevitably adds to a net increase in emission sources of air pollutants, particularly in the last decade.
Like other cities in the world, Amman city was comprehensively lockdown on 20 March 2020 to prohibit transmission of the COVID-19 and to protect people from infection.During the lockdown period (20 March-24 April 2020) people have been forced to stop all outdoor activities and use their feet for shortdistance movements.In 2021, Amman city was also partially lockdown during the night (6 pm to 6 am) from 1 March to 13 May.The partial lockdown includes closing all socioeconomic activities, stopping outdoor activities, and vehicle movements.Therefore, it is expected that the air pollutants have dropped during the comprehensive and partial lockdown.Also, as the LST is highly correlated with anthropogenic activities and air pollutants (Weng and Yang 2006;Alseroury 2015;Kahya et al. 2016), it is expected to decrease to lower than the normal average.

Data and its pre-processing
To link the variability/stability of NO 2 , CO, and LST with reduced human and industrial activities due to the COVID-19 lockdown (20 March to 24 April), remote sensing data from MODIS and TROPOsferic Monitoring Instrument (TROPOMI) were acquired.The Sentinel-5P satellites which were launched on 13 October 2017 by the European Space Agency were used to observe the NO 2 and CO datasets.These satellites offer a daily worldwide coverage using their nadir-viewing spectrometer that measures reflected sunlight with a range of multiple spectral domains from ultraviolet (UV) to near-infrared (NIR) through the visible bands (Alqasemi et al. 2021;Vîrghileanu et al. 2020).The daily tropospheric NO 2 and CO column density were derived from the fourth band with a spectral range of 405-500 nm and the 2.3 µm of the shortwave infrared (SWIR), respectively.Since July 2018, the Sentinel-5P began to offer information about atmospheric trace gases, aerosols, and climate, and thus it is now a suitable data source for air quality monitoring (Ialongo et al. 2020).Furthermore, the collected data are available to users at two processing levels namely L1B and L2.While the L1B is geolocated and radiometrically corrected Top-of-Atmosphere (TOA) (top-ofatmosphere) radiances, the L2 provides radiance, solar irradiance products, aerosols, clouds, and different pollutants products such as total columns of O 3 , SO 2 , NO 2 , CO, and CH 4 (Vîrghileanu et al. 2020).All of the Sentinel-5P datasets appear in Near Real-Time (NRTI) and Offline (OFFL) (except CH 4 , which is only available as an OFFL product).Although the NRTI information is provided in a smaller area than the OFFL, it appears within 3 h after acquisition.However, the OFFL products are available within 12 h for L1B and 3-5 days for L2 products.
The MODIS Aqua satellite was used to acquire the LST data on daily basis.The MODIS aboard the Aqua and Terra satellites was launched by NASA on 18 December 1999 and 4 May 2002, respectively.While Terra passes from north to south across the equator at approximately 10:30 a.m.local time, the Aqua passes south to north over the equator at about 1:30 p.m. Since the viewing swath is so large (2330 km), both of which offer nearly daily coverage of the Earth using 36 spectral bands ranging from visible to thermal infrared.
In this study, the NO 2 (COPERNICUS/S5P/NRTI/ L3_NO 2 ) and CO (COPERNICUS/S5P/OFFL/ L3_CO) data collections were acquired on daily basis from 20 March and 24 April for the years 2019, 2020, and 2021 at a 3.5-5 km resolution.The LST data at 1 km spatial resolution were acquired from 2014 to 2021 for the same period due to the missing 2019 LST data.These data were obtained and processed using Google Earth Engine (GEE) platform because it provides high computational capabilities as well as highquality satellite images (Gorelick et al. 2017).Table 1 summarizes the satellite data sources, data type, and spatial and temporal resolution.
To use the abovementioned dataset in the analysis, preprocessing steps were conducted.First, the cloud contaminated pixels were checked and excluded from the analysis.Therefore, pixels with quality assurance values of less than 75% were removed (Gorelick et al. 2017;Mesas-Carrascosa et al. 2020).To overcome the gaps in daily data, an 8-day composite of LST was collected from MODIS Aqua satellites (MYD11A2-v6) using the Google Earth Engine platform.To make the NO 2 and CO dataset compatible with the LST data, an average of 8-day was also computed.Second, the LST data unit of measurements was converted from Kalvin to Celsius by subtracting 273.15 from the original data.Finally, the datasets were clipped to the geographic extent of a boundary polygon that defined the study area.

Methods
To understand the spatial and temporal variability/ stability of NO 2 , CO, and LST, 66 random sample locations were created within the study area to calculate the mean, standard deviation, and variation coefficient of the three variables.The number of samplesas shown in Figure 1-was proportionally distributed within each administrative district in the city based on their areas, such as the larger administrative districts exhibited more samples than the smaller ones.The statistical approaches used in this study can be written as (Brocca et al. 2010;Almagbile et al. 2019): ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi where ([ j , [ i Þ, (σ j , σ i Þ and ðC:V j , C:V i Þ are spatial and temporal of mean, standard deviation, and variation coefficient.x ij denotes NO 2 in micromole/m 2 , CO in millimole/m 2 , and the LST values in °C, M is the number of 8-d composites from the whole period, which is in this case (4), N is the 66 sample locations, i and j present, respectively, the spatial and temporal dimensions.
To show the impact of the COVID-19 lockdown on the NO 2 , CO, and LST, the percentage change of the air pollutants and LST during the lockdown period was computed as follows: where X and Y denote the NO 2 , CO, and LST in the current and previous year, respectively.Note that since the LST values in 2019 were not available, the average of 5 years (2014-2018) was used instead.
Temporal stability of NO 2 , CO, and LST relative to location i and time j is given by the relative difference (δ ij ) and its mean ( � δ i ) (MRD) as follows (Cosh et al. 2008;Brocca et al. 2010

The spatiotemporal variability
Figure 2 shows the spatiotemporal variability of NO 2 concentration throughout the comprehensive lockdown period in 2020 and partial lockdown period in 2021 compared to the typical condition period in 2019.The results revealed that the NO 2 concentration level was significantly dropped during the comprehensive lockdown period in 2020 in all sample locations compared with 2019 and 2021.This referred to the reduction in the anthropogenic activities (e.g.industrial activities, public, and private transportation) which were restricted across a massive national lockdown.During the partial lockdown period in 2021, the level of NO 2 concentration was not considerably dropped but even was closer to that in the 2019 period.This might be related to the partial reopening of the industrial and automobile activities during the day hours from 6 am until 7 pm.
To understand the variability of NO 2 concentration during the period of interest in 2019, 2020, and 2021, the Variation Coefficient (CV) of NO 2 concentration was calculated among the 66 sample locations (see Figure 1).Figure 2(a) shows that the temporal averages during the period of interest in 2019, 2020, and 2021 were in the ranges 80-130, 65-85, and 80-145 micromole/m 2 , respectively.The temporal variation coefficient (Figure 2(b)) in 2020 was within a range of 4%-16% which was found to be less than the values in 2019 (i.e.6%-23%) and 2021 (i.e.6%-26%).Meanwhile, the spatial average of NO 2  Figure 3 shows the spatial distribution of NO 2 over the study area during the typical condition in 2019 and the comprehensive and partial-lockdown periods in 2020 and 2021, respectively.In general, the NO 2 concentration gradually increased toward the northern part of the city, particularly toward the Central Business District (CBD) which represents the heaviest commercial and business activities in the city.The concentration of NO 2 during 2019 ranged from approximately 65 micromoles/m 2 in the southern part to around 160 micromoles/m 2 in the northern part.During 2020 and 2021 however, the range of the NO 2 concentration level from the south to the north part was from approximately 65 to 100 micromole/m 2 and 65 to 160 micromole/m 2 , respectively.Note that, although the NO 2 concentration in 2021 was approximately like that in 2019, its spatial distribution was considerably shrunk in 2021 due to the partial lockdown status.This emphasizes the effect of lockdown on the NO 2 concentration levels in Amman city.
Furthermore, Figure 4 shows the percentage change of NO 2 concentration between the three years during the period of interest.In Figure 4(a), it can be noticed a significant decrease in NO 2 between 2019 and 2020 Additionally, it was noticed that within 20-28 March 2019, the percentage change of NO 2 was 10% less than that in the same period in 2021.This might be attributed to the high amount of precipitation during that period in 2019 which caused a reduction of the NO 2 concentration level.Nevertheless, positive percentage change values of 18%, 2%, and 22% were found for the periods 29 March-6 April, 7-15 April, and 16-24 April, respectively.

The spatiotemporal variability
Figure 6 shows the results of the CO concentration variations in both spatial and temporal dimensions during the three periods of study.From Figure 6(a), the temporal variation of CO in almost all the sample locations during the three periods of the study were in the range of 32 to 35 millimole/m 2 .During the comprehensive lockdown period in 2020, the CO  concentration level was slightly less than that during the partial lockdown and the typical study periods in 2021 and 2019, respectively.This situation can be notably seen from the temporal variation coefficient (Figure 6(b)) results which showed that the variation of CO during 2020 in all sample locations ranged from 2% to 4%.In addition, the variation coefficient during 2021 and 2019 was higher than those observed in 2020 with values ranging from approximately 2% to 7%.This revealed that the CO concentration level during the comprehensive lockdown period was slightly reduced due to the suspension of the anthropogenic activities, transportations, and industrial closure.
Besides, the spatial variability analysis (Figure 6(c)) revealed that -with exception of the period 20-28 March in 2020 -the CO concentration level was less than the CO concentration values in 2019 which was in the range 32 to 33 millimole/m 2 .In 2021, the results showed that the period 20 March to 6 April had higher CO values than the same period in both 2020 and 2019, but since then, the values became less than those in 2020 and 2019.The variation coefficient (Figure 6(d)) results showed that the values were 2% and 1.3% during 20-28 March and 7-15 April 2020, respectively.These were slightly less than that during 2021 and 2019.However, for 16-24 April, the value of CO concentration level (i.e.3%) was the highest among the whole period.The variation coefficient in 2021 recorded the highest values (2.2% to 3%) from 20 March to 15 April.The maps in Figure 7 highlighted the spatial distribution and dynamic of CO over the study area during the three periods of study.For instance, the highest concentration level (around 35.5 millimole/ m 2 ) occurred in the Central Business District(CBD) and tends toward the northeastern part of the city during 2019.In 2020 however, a slight shift of the highest CO concentration level toward the eastern and western parts occurred.This might be due to the massive lockdown of industrial activity and traffic across the CBD.Hence anthropogenic activities might be redistributed toward the city-hinterlands.Additionally, the highest CO concentration level (around 35.5 millimole/m 2 ) appeared in small patches in the western part.The situation in 2021 was similar to 2019 because the CBD and the northeastern part had the highest CO concentration level with the typical and partial opening of industrial, traffic, and anthropogenic activities.
Figure 8 shows the percentage change of CO concentration in the three periods of the study.A substantial drop in the CO concentration between 2019 and 2020 (Figure 8

The temporal stability
Analysis of the temporal stability in terms of the MRD and STRD measures of CO concentration is given in Figure 9.It shows that the MRD values (Figure 9(a)) of the 66 sample locations ranged between -0.03 and 0.05 millimole/m 2 .In general, MRD in 2020 was slightly less than that in both 2019 and 2021.Notably, samples 26, 27, 41, and 42 had the highest CO concentration among the total sample locations.The STRD (Figure 9

The spatiotemporal variability
Figure 10 shows the LST averages of the 8-d intervals and the coefficient of variation in both spatial and temporal scales for the 66 sample locations.The analysis of the temporal average (Figure 10(a)) of LST values emphasized that the LST during the comprehensive lockdown period in 2020 was less than those recorded in both the 5-year averages and 2021 for the equivalent time frames.The maximum LST value in 2020 was about 34°C within the period 7-15 April, and it was even less than the minimum value (i.e.35°C) in 5-year averages.Overall, the temporal average of LST in 2021 was higher than that of 2020 and 5-year averages for the period 20 March-7 April.However, the LST in 2021 in the second half of the lockdown period (i.e.7-24 April) was the highest (i.e.36°C to 44°C).The temporal variation coefficient (Figure 10(b)) emphasized that except for the period 20-28 March, the values in the 5-year averages (i.e.7%-8.2%) was less than 2020 (i.e.7.9%-8.5%)and 2021(i.e.6.5%-12.1%).Figure 10(c) shows lower LST spatial average values in 2020 than those in the 5-year average of LST, and 2021.During the periods of study, the LST spatial average values ranged from approximately 26°C to 34°C, 31°C to 38°C, and 30°C to 39°C, in 2020, the 5 years averages, and 2021, respectively.Also, it can be noticed that the LST values in 2021 were slightly higher than those recorded in the 5-year averages.In the case of the spatial variation coefficient (Figure 10(d)), the results revealed that the LST in 2020 recorded the highest variation coefficient at approximately 10%-25% compared with 2021 (i.e.15%-24%) and the 5-year averages (i.e.6%-12%).Note that, though the lockdown of human and industrial activities in 2020, other factors such as geographical location, topography, urban density, weather condition, and urban vegetation distribution, among others, could cause variation in the LST from one place to another (Feng et al. 2019;Khandelwal et al. 2018).
The spatial pattern of LST over the study area is illustrated in Figure 11.As displayed, the study area could be divided into two distinguished zones such as the eastern and western parts.The eastern part experienced higher LST values than the western part in all three periods.For instance, it was in the range 36-40°C in the eastern part and the 31-35°C in the western part during the 5-year average.Similar findings were observed during the partial lockdown period in the year 2021.However, during the comprehensive lockdown period in 2020, the LST values dropped down to the range of 16-30°C.In 2020, the eastern area exhibited LST values in the range 31-35°C whereas the western part experienced 26-30°C.This might be related to several reasons including the high urban and population densities, the low vegetation cover, industrial areas, and local climate conditions in the eastern part compared to the western part of the study area.Figure 12 shows the percentage change of the LST between the 5-year averages, 2020, and 2021 during the period of interest.The analysis indicated that all 8-day intervals encountered a reduction in LST, and this was emphasized when comparing the LST in the 5-year average and 2020 (Figure 12(a)).Notably, a negative percentage change was observed, and the maximum difference was around -25% in the period 20-28 March.On the other hand, a positive percentage change was found between 2020 and 2021 at all 8-day intervals (Figure 12(b)).This means that the LST in 2021 was higher than that in 2020.Finally, the percentage change between the 5-year average and 2021 (Figure 12(c)) was negative between 20 March and 6 April and positive in the second half of the lockdown period.Overall, the reduction in LST during the whole period in 2020 was approximately 13% and 18% less than the 5-year average and 2021, respectively.However, a tiny increase of 1% between the 5-year average and 2021 can be noticed.

The temporal stability
Figure 13 illustrates the analysis of the temporal stability of LST in the 66 sample locations for the 5-year averages, 2020, and 2021.In the case of the 5-year averages (Figure 13

Discussion
Since the COVID-19 lockdown situation has been implemented in many countries, several recent studies have been done to analyze the impact of the lockdown on different environmental aspects including the atmospheric gases, air quality, water quality, land surface temperature, and many other physical and environmental variables (Vîrghileanu et al. 2020;Alqasemi et al. 2021;Ali et al. 2021).However, most of these studies were directed to highlight the variability of such variables using ground-based and/or satellite data (e.g.MODIS and Sentinel-5p) in urban areas at different scales.This study analyzed the spatiotemporal variability/stability of NO 2 , CO, and LST during a comprehensive lockdown period in 2020, a partial lockdown period in 2021, and a typical period in 2019 in Amman city.
The results of the spatiotemporal variability of NO 2 highlighted a significant reduction during the comprehensive lockdown period in 2020 compared with 2019.The spatial distribution maps confirmed that the CBD area had lower NO 2 values in 2020 than that in 2019.The overall decreasing trend of NO 2 between 2020 and 2019 was around 27%.These results agreed with other previous studies conducted in various regions and spatial scales  2020) recorded an average reduction in NO 2 by 49% and more than 76% in China and Mumbai (India), respectively.In similar environments to Jordan, a decline in NO 2 by 23.7% and 96% was observed over the northern part of the United Arab Emirates (Alqasemi et al. 2021), and in Sale city in Morocco (Otmani et al. 2020), respectively.These results emphasized the impact of the lockdown operations of human activities and vehicular movement upon reducing NO 2 in Amman city as well as in other cities around the world.However, since vehicles and socioeconomic activities were only allowed between 6 am and 6 pm in 2021, the overall NO 2 concentration increased by 7% and 48% compared to 2019 and 2020, respectively.The temporal persistence in NO 2 provided almost constant values in 2020 over the study area which confirmed the fact that traffic-related activities and lockdown of socioeconomic activities are the main emission source of NO 2 .
The results, furthermore, showed a slight reduction in CO (i.e.around 1%) in 2020 and 2021 compared to 2019.These results were compatible with previous studies which observed a reduction in CO values (Ali et al. 2021;Ju, Oh, and Choi 2021;Zhou et al. 2021;Yuan et al. 2021).Zhou et al. (2021) reported a reduction in CO with a range of 2%-13% in most Chinese cities with no change in Wuhan city.Nevertheless, Yuan et al. (2021) found an average of 30% reduction in CO in the Yangtze River Delta during the COVID-19 lockdown period.The variation in CO reduction values could be attributed to variations in urban densities, climate conditions, industrial activities, and the length of the lockdown period.The CO temporal stability results confirm that the CO values in 2020 over Amman city were more stable than those observed in 2021 and 2019.
The influence of lockdown status on LST in the temporal and spatial scales was visible throughout the lockdown period over the study area.The overall drop in LST was approximately 13% and 18% less than the 5-year average (2014-2018) and 2021, respectively.In general, the reduction of LST in Amman city was comparable with previous studies.Ali et al. (2021) reported a decline in the surface urban heat island in megacities in Pakistan by around 20%, while the north UAE experienced a reduction in UHI by around 19.2% (Alqasemi et al. 2021).The temporal persistence in this study showed a lower STRD in 2020 and 2021 than that of the 5-year average.

Conclusions
Nitrogen dioxide (NO 2 ), carbon monoxide (CO), and land surface temperature are among the most pollutants and environmental variables that are mainly related to human activities especially burning fossil fuels in industrial and transport sectors.This study evaluates the impact of COVID-19 lockdown on the trend of the spatiotemporal variability/stability of these three variables in Amman city in Jordan.The analysis was performed over three time frames such as a comprehensive lockdown period in 2020, a partial lockdown period in 2021, and a typical period in 2019.The analysis was based on the assessment of 66 random sample locations within the study area as collected from various remote sensing data i.e.Sentinel-5p and MODIS.Results showed a significant reduction in the spatiotemporal variation of NO 2 during the comprehensive lockdown 2020 compared to the typical period in 2019.The temporal persistence in NO 2 provided almost constant values in 2020 over the study area which confirmed the fact that traffic-related activities and lockdown of socioeconomic activities are the main emission source of NO 2 pollution in the city.The results of CO showed a slight reduction in CO (i.e.around 1%) in 2020 and 2021 compared to 2019.The CO temporal stability results showed that the CO values in 2020 were more stable than in 2021 or 2019.
The temporal and spatial analysis of LST showed a noticeable drop of LST during the comprehensive lockdown period over the study area.These results present an opportunity for country-wide policies to mitigate the impact of air pollution on the health of  people.However, it should be noted that various environmental factors could influence LST and the air pollutants including wend speed and direction, relative humidity, cloud cover, and vegetation cover.We recommend characterizing these factors for a holistic understanding of their effect on the spatiotemporal distribution of NO 2 , CO, and LST and its impact on the livelihoods of the city in future research.

Figure 1 .
Figure 1.(a) Amman city with random sample locations used for analyzing the spatiotemporal stability/variability of air pollutants and LST from remote sensing data, and (b) Jordan borders.
concentration within the 8-d composite over the study area showed a significant decrease in NO 2 levels during the comprehensive lockdown period in 2020.That were 68 to 80 micromole/m 2 in 2020, 98 to 120 micromole/m 2 in 2019, and 90 to 120 micromole/m 2 in 2021 (Figure2(c)).This can be declared when comparing the spatial variation coefficient (Figure2(d)) for the same period in 2019, 2020, and 2021 with their corresponding values of 15-20%, 8-10%, and 18-27%, respectively.

Figure 2 .
Figure 2. (a) Spatial variation among the 66 sample locations using the temporally averaged data, (b) temporal CV, (c) temporal variation using the spatially averaged data of 66 sites, and (d) spatial CV of NO 2 during the comprehensive, partial lockdown, and the typical period from 20 March to 24 April in 2020, 2021, and 2019, respectively.
Figure 5 illustrates the results of the temporal stability analysis including MRD and STRD of NO 2 concentration for the 66 sample locations during the study period.Overall, all sample locations experienced relatively low MRD values (Figure 5(a)) with a range from approximately -0.3 to 0.4 micromole/m 2 .Additionally, the MRD values could be divided into three groups such as (i) positive MRD (i.e.samples 1-24), (ii) nearly zero MRD values (i.e.samples 25-42), and (iii) negative MRD values (i.e.samples 43-66).It can also be seen that the MRD during the comprehensive lockdown period in 2020 was considerably less than the partial lockdown period in 2021, and the typical year of 2019.The results of STRD (Figure 5(b)) revealed that the lowest values occurred during the comprehensive lockdown period in 2020 among the other two years (2021 and 2019).Meanwhile, it was noticed the NO 2 concentration values were relatively stable at less than 0.1 micromole/m 2 during 2020, while it fluctuated in 2019 and 2021.

Figure 3 .
Figure 3. Spatial distribution maps of the spatial average of NO 2 during the period from 20 March to 24 April 2020 in the study area in (a) 2019, (b) 2020, and (c) 2021.

Figure 4 .
Figure 4. Percentage change of NO concentration during the period from March to April in the study area between (a) 2019-2020, (b) 2020-2021, and (c) 2019-2021.
(a)) reflected a negative percentage change in all the 8-d intervals during the lockdown period except for 20-28 March.This percentage change was +2, -2, -3, and -0.3 during 20-28 March 2029 March-6 April, 7-15 April, and 16-24 April, respectively.The overall reduction of the CO concentration level between 2019 and 2020 was approximately 1%.For a comparison between 2020

Figure 5 .
Figure 5. Spatial variation of (a) mean relative difference (MRD), and (b) standard deviation (STRD) which represents temporal stability of NO 2 concentration in the 66 sample locations during the period from 20 March to 24 April in 2021, 2020, and 2019.
(b)) determined the temporal stability of the CO concentration over the study area.Its values in 2020 were relatively less than those in 2019 and 2021.Additionally, the high MRD values in samples 26, 27, 41, and 42 reflected the highest STRD values with ranges between 0.05 and 0.06 millimole/m 2 .However, the STRD values in the other sample locations fluctuated around 0.02 millimole/m 2 .

Figure 6 .
Figure 6.(a) Spatial variation among the 66 sample locations using the temporally averaged data, (b) temporal CV, (c) temporal variation using the spatially averaged data of 66 sites, and (d) spatial CV of CO during the comprehensive, partial lockdown, and the typical period from 20 March to 24 April in 2020, 2021, and 2019, respectively.

Figure 7 .
Figure 7. Spatial distribution maps of the spatial average of CO during the period from 20 March to 24 April 2020 in the study area in (a) 2019, (b) 2020, and (c) 2021.
(a)), the MRD values ranged from 10°C to 13°C, this means that all sample locations experienced positive LST values.The STRD results also reflected negligible positive values of around 1°C.The situation during the comprehensive lockdown period in 2020 and the partial lockdown period in 2021 (Figure 13(b)) was entirely different.In both 2020 and 2021, the MRD fluctuated around zero with ranges from -0.1 to 0 and the STRD values for 2021 (i.e.0.03°C to 0.13°C) were slightly lower than those obtained in 2020 (i.e.0.01°C to 0.06°C) and this emphasized the variability of the LST in 2020 due to the different factors as mentioned earlier.

Figure 9 .
Figure 9. Spatial variation of (a) mean relative difference (MRD), and (b) standard deviation (STRD) which represents temporal stability of CO concentration in the 66 sample locations during the period from 20 March to 24 April in 2021, 2020, and 2019.

Figure 10 .
Figure 10.(a) Spatial variation among the 66 sample locations using the temporally averaged data, (b) temporal CV, (c) temporal variation using the spatially averaged data of 66 sites, and (d) spatial CV of LST during the comprehensive, partial lockdown, and the typical period from 20 March to 24 April in 2020, 2021, and 2019, respectively.

Figure 11 .
Figure 11.Spatial distribution maps of the spatial average of LST during the period from 20 March to 24 April in the study area in (a) the 5-year average 2014-2018, (b) 2020, and (c) 2021.The white area in (A) represents missing data.

Figure 13 .
Figure 13.Spatial variation of the mean relative difference (MRD) and standard deviation (STRD) which represents temporal stability of LST in the 66 sample locations during (a) 5-year average 2014-2018, (b) 2020, and 2021 during the period of interest from 20 March to 24 April.
; Zhang and Shao 2013; Almagbile et al. 2019); And the Standard Deviation (STRD) of � δ i is:As a rule of thumb, the values of the standard deviation of σ i ðδ i Þ represent temporal stability.If a location has stable values of NO 2 , CO, and LST, this location is considered temporally stable and vice versa.

Table 1 .
Description of the remote sensing data used in this study.