key: cord-0839731-gqw62i5l authors: Firozjaei, Mohammad Karimi; Fathololomi, Solmaz; Kiavarz, Majid; Arsanjani, Jamal Jokar; Homaee, Mehdi; Alavipanah, Seyed Kazem title: Modeling the impact of the COVID-19 lockdowns on urban surface ecological status: A Case Study of Milan and Wuhan cities date: 2021-02-23 journal: J Environ Manage DOI: 10.1016/j.jenvman.2021.112236 sha: 44fa288965f4a5953aaaf8324fa9ea821085a3f8 doc_id: 839731 cord_uid: gqw62i5l The COVID-19 pandemic has caused unprecedent negative impacts on our society, however, evidences show a reduction of anthropogenic pressures on the environment. Due to the high importance of environmental conditions on human life quality, it is crucial to model the impact of COVID-19 lockdown on environmental conditions. Consequently, the objective of this study was to model the impact of COVID-19 lockdown on the urban surface ecological status (USES). To this end, the Landsat-8 images of Milan for three pre-lockdown dates (Feb 13, 2018 (MD1), April 18, 2018 (MD2) and Feb 3, 2020 (MD3)) and one date over the lockdown (April 14, 2020 (MD4)), and Wuhan for three pre-lockdown dates (Dec 17, 2017 (WD1), March 23, 2018 (WD2) and Dec 7, 2019 (WD3)) and one lockdown date (Feb 9, 2020 (WD4)) were used. First, pressure-state-response (PSR) framework parameters including index-based built-up index (IBI), vegetation cover (VC), vegetation health index (VHI), land surface temperature (LST) and Wetness were calculated. Second, by combining the PSR framework parameters based on comprehensive ecological evaluation index (CEEI), the USES were modeled on different dates. Thirdly, the USES during the COVID-19 lockdown was compared with the USES for pre-lockdown. The mean (standard deviation) of CEEI for Milan on MD1, MD2, MD3 and MD4 were 0.52 (0.12), 0.60 (0.19), 0.57 (0.13) and 0.45 (0.16), respectively. Also, these values for Wuhan on WD1, WD2, WD3 and WD4 were 0.63 (0.14), 0.67 (0.15), 0.60 (0.13) and 0.57 (0.13), respectively. Due to the lockdowns, the mean CEEI of built-up, bare soil and green spaces for Milan and Wuhan decreased by [0.18, 0.02, 0.08], [0.13, 0.06, 0.05], respectively. During the lockdown period, the USES improved substantially due to the reduction of anthropogenic activities in the urban environment. Coronavirus disease, also known as COVID-19, is an infectious disease caused by the coronavirus Acute 28 Respiratory Syndrome (SARS-CoV-2) (Lai et al., 2020) . The disease was first reported in Wuhan, China 29 on December 31, 2019, and outbroke rapidly around the world due to its contagious nature in a way that 30 involved all countries of the world (Qin et al., 2020) . COVID-19 was recognized as an epidemic on 31 January 2020 (Velavan and Meyer, 2020) . Nearly 61 million people in the world are infected by the Google Earth images were used to extract training-testing datasets for land cover classification. Daily 108 water vapor (MOD07) and surface temperature products (MOD11A1) were also used to calculate the LST 109 obtained from the Landsat-8 reflective and thermal bands. 110 The schematic flowchart shown in Fig. 2 was used to model and evaluate the impact of COVID-19 112 lockdown on the USES. In the first step, the parameters related to the PSR framework including index-113 based built-up index (IBI), vegetation cover (VC), vegetation health index (VHI), LST and Wetness were 114 calculated based on the combination of Landsat-8 reflective and thermal bands at different dates. In the 115 second step, by combining the information of PSR framework parameters based on CEEI, the USES were 116 modeled at different dates. In the third step, the USES during the COVID-19 lockdown period were 117 compared against the USES in the pre-lockdown period. 118 The used spectral indices and methods to extract the PSR framework parameters are presented in Table 1 . 122 Normalized difference vegetation index (NDVI) is one of the most widely used index in assessment and 123 modeling of vegetation (Robinson et al., 2017) . VC can show vegetation fraction in one pixel in the range 124 of 0-1 values indicating pixels with least vegetation cover i.e., totally impervious or soil surface to pixels 125 with most vegetation cover i.e., totally vegetation, respectively (Sobrino et al., 2008) . VC can easily 126 demonstrate the vegetation fraction in one pixel; however, not ideal for indicating vegetation health. 127 J o u r n a l P r e -p r o o f Hence, VHI was used to tackle this challenge, since it considers the levels of chlorophyll, nitrogen and 128 xanthophyll for quantifying vegetation health . Consequently, in order to consider each 129 of these characteristics, NDVI, nitrogen reflectance index (NRI), and normalized difference senescent 130 vegetative index (NDSVI), which represent chlorophyll, nitrogen, and xanthophyll were combined using 131 Principal Component Analysis (PCA) and consequently VHI was calculated . Studies 132 show that IBI is highly effective in showing the percentage of impervious surface cover in a pixel. A pixel 133 with a higher IBI value contains a higher fraction of built-up (Hu and Xu, 2018; Xu, 2008) . Information from at least two or more reflective and thermal bands of satellite images was used to 142 calculate these parameters. The quick atmospheric correction (QUAC) model was used for atmospheric 143 correction of Landsat-8 reflective bands. This model is straightforward and requires minimal inputs for 144 atmospheric correction. The NDVI threshold method (Sobrino et al., 2008) The support vector machines (SVM) method was used to generate land cover maps (Foody and Mathur, 149 2004; Otukei and Blaschke, 2010) . This method is one of the non-parametric and supervised classification 150 methods. The main advantage of this method is its high ability to use less training samples and its 151 reported higher accuracy compared to other methods. The main purpose of this algorithm is to find the 152 maximum distance between two classes and thus increase the classification accuracy. In this study, the 153 RBF kernel (With set 300 for C and 3 for γ) were used in the SVM model to classify land cover. (2) In equation 2 and 3, B C was the values of the standardized parameters, B C was the initial values of the 170 parameters, and B DC: and B DEF were the lowest and highest parameter values in the area, respectively. 171 In this study, to quantify the USES, the standardized parameters related to the PSR framework have been 172 used as Eq. 4. 173 The CEEI value obtained from Eq. 4 is standardized into 0 and 1 using Eq. 2. Values of 1 and 0 for CEEI 174 indicate the worst and best USES, respectively. Areas with maximum LST and IBI and minimum VC, These areas have poor USES. Ecosystem performance is unfavorable. These areas lack vegetation and human activities in these areas are very high and the effect of urban heat island and drought in these urban areas is obvious. To investigate the impact of COVID-19 lockdown on the USES, the following steps were followed: WD1, WD2 and WD3 for Wuhan) and during the lockdown (MD4 for Milan and WD4 for Wuhan) dates. 214 The mean and SD of the surface characteristics of these cities in different dates were different due to 215 seasonal changes, land cover and human activities changes (Table 4) (Table 5) . For the pre-lockdown dates, the 250 worst USES was related to built-up lands in these cities, but in the lockdown date, the mean CEEI 251 J o u r n a l P r e -p r o o f decreased and was lower than the mean CEEI of bare soil lands. The lowest and highest differences 252 between the mean CEEI of built-up lands and green space lands were in the date of lockdown (MD4 for 253 Milan and WD4 for Wuhan) and the similar dates in last year (MD2 for Milan and WD2 for Wuhan). In 254 general, the results show that among the different dates, the best USES in Milan and Wuhan cities were 255 on the lockdown dates. 256 Milan and Wuhan cities were related to the COVID-19 lockdown date (Table 6 ). The results show that 261 among the different dates, the best USES in Milan and Wuhan cities were related to the COVID-19 262 lockdown dates. However, the USES of these cities were bad on the same dates in previous years. The 263 highest area of Poor class for Milan and Wuhan cities was related to MD2 and WD2 dates, respectively, 264 due to the various factors such as climatic and seasonal conditions and human activities. 265 With the significant reduction of human activities in the urban environment due to the COVID-19 278 lockdown in 2020, the USES of these cities in the same climatic and seasonal conditions significantly 279 improved. The average CEEI differences in different lands vary between different dates (Table 8) . For 280 Milan, the mean CEEI difference between MD4 and MD2 in built-up, bare soil, and green spaces were -281 0.18, -0.02 and -0.08, respectively. Also, for Wuhan, the mean difference of CEEI between WD4 and 282 WD2 in these lands were -0.13, -0.06 and -0.05, respectively. This indicates the greatest changes in USES 283 between the COVID-19 lockdown date and the pre-lockdown dates were related to the built-up lands. The Environmentally unfriendly human activities in some urban environments have increased substantially in 293 The results of this study showed that COVID-19 lockdown significantly improved USES (Table 5-7) . 308 Recent studies have shown that the COVID-19 lockdowns significantly reduced air and water pollution 309 have the highest CEEI values due to the low value of greenness and wetness and high value of impervious 317 surface cover, dryness and heat ( Fig. 3 and 4 and Table 5 ). During the lockdown period, due to the 318 decrease in human activities, the heat and dryness of the built-up land area decreased significantly, so that 319 the most improvement in USES was related to urban areas with a high percentage of impervious surface 320 cover. Consequently, the results of this study confirm that reducing human activities in the urban 321 environment can be one of the most effective and useful solutions to improve environmental conditions 322 One of the important challenges in this study is modeling the temporal changes of USES due to changes 334 in human activities . Modeling the daily USES changes with high spatial resolution in 335 urban environments can be important, versatile and useful. To solve this challenge, the use of satellite 336 imagery with a better temporal resolution than Landsat, such as MODIS images, can be used. However, 337 the spatial resolution of these images is low, which due to the heterogeneity of various characteristics and 338 parameters related to PSR framework in the urban environment, the use of these images can be less The prevalence of COVID-19 had significant negative effects on various aspects of human life in urban 349 and non-urban environments. However, due to reduction of human activities during the lockdown period, 350 high LST values on MD2 (Milan) and WD2 (Wuhan) were changed to areas with medium and low 208 LST values on MD4 (Milan) and WD4 (Wuhan). However, changes in other surface characteristics at Improving the disaggregation of MODIS land surface temperatures 378 in an urban environment: a statistical downscaling approach using high-resolution emissivity Derivation of a tasselled cap transformation based on 381 Landsat 8 at-satellite reflectance Changes in US air pollution during the COVID-19 pandemic. Science of the 383 Total Environment COVID-19 outbreak: Migration, effects on society, global environment 385 and prevention Assessment of land surface temperature and heat fluxes 387 over Delhi using remote sensing data Characterizing spatiotemporal 389 dynamics of anthropogenic heat fluxes: A 20-year case study in Beijing-Tianjin-Hebei region in China Modelling rural residential settlement patterns with 392 cellular automata Downscaling MODIS land surface temperature over a heterogeneous 394 area: An investigation of machine learning techniques, feature selection, and impacts of mixed pixels Improved DisTrad for downscaling thermal 397 MODIS imagery over urban areas A new 399 approach for modeling near surface temperature lapse rate based on normalized land surface 400 temperature data Modelling surface 402 heat island intensity according to differences of biophysical characteristics: A case study of Amol city Surface Ecological Index (RSUSEI): An Analytical Framework for Assessing the Surface Ecological Status in 406 Urban Environments Monitoring and forecasting 408 heat island intensity through multi-temporal image analysis and cellular automata-Markov chain 409 modelling: A case of Babol city Surface anthropogenic 411 heat islands in six megacities: An assessment based on a triple-source surface energy balance model Toward intelligent training of supervised image classifications: directing 414 training data acquisition for SVM classification Mobile phone location data reveal the effect and geographic variation of social distancing on the 417 spread of the COVID-19 epidemic Analysis and modelling of surface Urban Heat Island 419 in 20 Canadian cities under climate and land-cover change A new remote sensing index for assessing the spatial heterogeneity in urban 422 ecological quality: A case from Fuzhou City Application of the pressure-state-response 424 framework to perceptions reporting of the state of the New Zealand environment Land surface temperature 427 retrieval methods from Landsat-8 thermal infrared sensor data Severe acute respiratory syndrome 429 coronavirus 2 (SARS-CoV-2) and corona virus disease-2019 (COVID-19): the epidemic and the challenges 432 The positive impact of lockdown in Wuhan on containing the COVID-19 outbreak in China A new method to quantify surface 435 urban heat island intensity Effect of lockdown amid COVID-19 pandemic on air quality of the 437 megacity Delhi COVID-19 pandemic persuaded lockdown effects on environment over stone 439 quarrying and crushing areas Modeling 441 outdoor thermal comfort using satellite imagery: A principle component analysis-based approach The urban heat island effect, its causes, and 444 mitigation, with reference to the thermal properties of asphalt concrete COVID-19 pandemic and environmental pollution: a blessing 447 in disguise? Science of The Total Environment A comparison of different regression models for downscaling 449 Landsat and MODIS land surface temperature images over heterogeneous landscape COVID-19 pandemic: Impacts on the air quality during the partial 452 lockdown in São Paulo state, Brazil. Science of The Total Environment The socio-454 economic implications of the coronavirus and COVID-19 pandemic: a review Land cover change assessment using decision trees, support vector 457 machines and maximum likelihood classification algorithms 459 Dysregulation of immune response in patients with COVID-19 in Wuhan, China A dynamic Landsat derived normalized difference vegetation index (NDVI) 463 product for the conterminous United States Land surface emissivity retrieval from different VNIR and TIR sensors Urban sprawl and air quality in large US cities Impact of urbanisation on urban heat island intensity during 470 summer and winter over Indian metropolitan cities Changes in air quality during the lockdown in Barcelona (Spain) one month into the SARS-CoV-2 473 epidemic The COVID-19 epidemic. Tropical medicine & international health 25 A Partition Modeling for Anthropogenic Heat Flux Mapping in 477 China Thermal infrared remote sensing for urban climate and environmental studies: 479 Methods, applications, and trends Normalizing land surface 481 temperature for environmental parameters in mountainous and urban areas of a cold semi-arid climate The impact of land use and land cover changes on land surface temperature in 485 a karst area of China A new index for delineating built-up land features in satellite imagery Prediction of ecological effects of potential 489 population and impervious surface increases using a remote sensing based ecological index Detecting Ecological Changes with a Remote Sensing 492 Based Ecological Index (RSEI) Produced Time Series and Change Vector Analysis. Remote Sensing 11 Rapid urbanization and policy 495 variation greatly drive ecological quality evolution in Guangdong-Hong Kong-Macau Greater Bay Area of 496 China: A remote sensing perspective Indirect effects of COVID-19 on 498 the environment conditions on it needs to be evaluated in more details. Also, the use of techniques to improve spatial and 366 temporal resolution to produce daily USES maps and the impact of human activities on it with high 367 spatial and temporal resolution can be considered in future studies. Despite the fact that COVID-19 368 pandemic has caused tremendous damages to our society in terms of health and economy, it has brought 369 up thoughts about transferring some of the lessons learned from the pandemic and its lockdowns for 370 climate change and its emerging consequences. 371 ☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: