Impact of COVID lockdowns on spatio-temporal variability in land surface temperature and vegetation index
Authors: Apurba Tewari, Nishi Srivastava
Abstract
In urban areas, industrial and human activities are the prime cause that exacerbates the heating effects, also called the urban heat island (UHI) effect. The land surface temperature (LST), normalized difference vegetation index (NDVI), and the proportion of vegetation (Pv) are indicators of measurement of the heating/urbanization effects. In the present work, we investigated the impact of the COVID-19 lockdown, i.e., restricted human activities. We used Landsat-8 OLI/TIRS (level 1) data to investigate spatial and temporal heterogeneity changes in these urbanization indicators during full and partial lockdown periods in 2020 and 2021, with 2019 as the base year. We have selected three cities in India’s eastern coal mining belt, Bokaro, Dhanbad, and Ranchi, for the study. Results showed a significant decrease in LST values over all sites, with a maximum reduction over mining sites, i.e., Bokaro and Dhanbad. The LST value decreased by about 13–19% during the lockdown period. Vegetation indices (i.e., NDVI and Pv) showed a substantial increase of about 15% overall sites. With decreased LST values and increased NDVI values, these quantities’ correlations became more negative during the lockdown period. More positive changes are noticed over mining sites than non/less mining sites. This indirectly indicates the reduction in the heat-absorbing particles in the environment and surface of these sites, a possible cause for the reduction in LST values substantially. Reduction in coal particles at the land and vegetation surface likely contributed to decreased LST and enhanced vegetation indices. To check the statistical significance of changes in the UHI indicators in the lockdown period, statistical tests (ANOVA and Tukey’s test) are performed. Results indicate that most of the case changes have been significant. The study’s finding suggests the lockdown’s positive impact on the heating/UHI effects. It emphasizes the need for planned lockdowns as city mitigation strategies to overcome pollution and environmental issues.
Supplementary Information
The online version contains supplementary material available at 10.1007/s10661-023-11119-7.
Introduction
The COVID-19 (SARS-CoV-2) pandemic has plagued the world since January 2020, and the World Health Organization (WHO) declared it a global health emergency. This disease quickly captured the whole world after the first case was detected in Wuhan, China, and caused the death of millions of people globally (Shi et al., 2020). Many restriction policies, such as social distancing, quarantines, mass travel bans, and lockdowns, have been implemented worldwide to avoid the spread and risk of the COVID-19 pandemic. Its health impact has severely affected the world’s economy and social relations as several courtesies imposed total or partial lockdowns to restrict its spread (Bukhari & Jameel, 2020).
The world, including India, sees this pandemic as a crisis with all respect, i.e., social, economic, and health aspects. But this period became a boon for natural resources due to restricted human interaction, and notable improvements are noticed in various environmental pollution indicators (Gautam, 2020; Guha & Govil, 2021; Zambrano-Monserrate et al., 2020). Studies showed a significant decrease in air pollution and temperature during the lockdown period due to the lockdown (Yunus et al., 2020).
In India, as a preventive measure to restore the infrastructure to deal with this pandemic, the Indian government also announced a nationwide lockdown from 24 March to 14 April 2020. It was extended for a few more phases as the second (15 April to 3 May 2020), the third (4–17 May 2020), and the fourth (18–31 May 2020). All other activities were seized during this duration except for the emergency services. Thus, to observe the human-induced changes, the lockdown period provided was found suitable as the pollutants emission was reduced significantly during this. Urban heat island (UHI) effect, often known as the heating effect, is mainly caused by industrial and human activity in metropolitan areas. Indicators of the heating/urbanization effects include the land surface temperature (LST), normalized difference vegetation index (NDVI), and the percentage of vegetation (Pv).
Land surface temperature is a fundamental variable in climatological and environmental research that is extensively investigated in the study of climate change, environment, agriculture, hydrology, and urban planning (Hu et al., 2020; Shah et al., 2019; Wang et al., 2019; Weng et al., 2019). In addition, LST is important in various crop management strategies, including stress detection, yield forecasting, crop growth forecasting, and irrigation scheduling (Neinavaz et al., 2020). Along with LST, the normalized difference vegetation index is also an important parameter related to heat effects and measures the amount and vigor of vegetation at the surface. The proportion of vegetation is also a significant vegetation biophysical factor associated with Earth’s surface processes (Neinavaz et al., 2020). The ratio of the vegetation’s verticle projection area, which includes branches, stalks, and leaves on the ground, to the overall vegetation areas is known as the proportion of vegetation (Deardorff, 1978). The vegetation surface has a low LST; however, the concrete and bare land have a high LST.
According to Zhan et al. (2015), transportation, industrial, and residential activity are all closely tied to LST. Buyantuyev and Wu (2010) and Li et al. (2016) have observed a relationship between multiple factors, including anthropogenic concerns on LST. During the post-lockdown period, several studies noticed a drop in the LST values in various Indian cities (Maithani et al., 2020; Guha & Govil, 2021; Sahani et al., 2021). According to another study conducted in China (Wuhan City), the mean LST during COVID-19 was lower than the preceding 3 years’ mean LST on the exact dates (Hadibasyir et al., 2020). The NDVI is among the extensively used vegetation indicators derived from remote sensing data for land surface process assessment and ecosystem monitoring (Ke et al., 2015). Guha and Govil (2021) observed an increasing trend in the NDVI values.
Considerable research has been conducted on the COVID-19’s influence on land surface temperature and air quality in India. However, the studies in eastern Indian cities, especially in the coal mining region, are less observed. Lockdown-induced changes in heat/urbanization indicator measures like LST, NDVI, and Pv are sparsely studied in eastern India, particularly in the Jharkhand region. This region is essential to be studied as it falls in the coal mining belt region. Thus, it will be scientifically exciting to observe changes in UHI indicators as a change in the human/industrial activity pattern.
Thus, this study has been performed to observe the changes in the spatial distribution of heat and vegetation indicators over the mining region of Eastern India. To achieve the objective, we used the remote sensing data (Landsat 8 OLI and TIRS) to derive and generate spatial distribution maps of LST, NDVI, and Pv for three cities in Jharkhand: (a) Bokaro, (b) Dhanbad, and (c) Ranchi; changes occurred in the spatial distribution of these parameters; and we observed the changes that occurred in the correlation between LST and NDVI as the effect of lockdown in our area of study.
We observed the changes induced in LST, NDVI, and Pv due to imposed lockdown on most anthropogenic activities. As the study sites are falling in the coal mining region, especially the Dhanbad site, it is necessary and interesting to see the exact effect of the lockdown on this region. We have compared the total lockdown period (i.e., 2020) and partial lockdown periods (i.e., 2021) with pre-lockdown (i.e., 2019). This analysis will help us to visualize the broader impact of lockdown on these indices.
Materials and methods
Study area
In this work, we explored the effect of the lockdown on the land heat, urbanization, and vegetation indicators over the three major cities in Jharkhand, India: Bokaro, Dhanbad, and Ranchi. Figure 1 shows the geographical locations of the sites of study. The geographical and climatic conditions of the study sites are as Fig. 1Location of the study areas
The Bokaro district has a land area of 2861 sq km and is located between latitudes 23.24.27 N and 23.57.24 N and longitudes 85.34.30 E and 86.29.10 E, as shown in Fig. 1a (https://bokaro.nic.in/). This location’s sea surface height is about 210 m. The summers are hot and dry, the monsoons are cool and humid, and the winters are chilly in this region. Around 1288.3 mm of rainfall on average falls in Bokaro each year. In the summer, Bokaro experiences hot weather with highs of 41 °C and lows of 24 °C. The wintertime temperature ranges from 7 to 26 °C, with May as the hottest month and December/January as the coldest (IMD, 2021).
The second site selected for the study is Dhanbad, situated in the mid-eastern part of Jharkhand state, which witnessed extensive mining activities which have led to massive population growth in the region (https://dhanbad.nic.in/). The district has a total area of 2074 sq km, varying between 23.37° N to 23.0° N longitudes and 86.06° E to 86.49° E latitude (Fig. 1b). This location is 222 m above sea level. The area experiences a tropical environment with sweltering summers and freezing winters. The hottest month is May, with temperatures ranging from 23.4 to 38.1 °C, and the coldest month is January, with temperatures ranging from 10.5 to 25.4 °C. At Dhanbad, rainfall occurs on an average of 1300.5 mm each year. Summer to monsoon relative humidity ranges from 35 to 80%. Light to moderate winds blows across this location (IMD, 2021).
The third site is Ranchi, the most populated district of Jharkhand, which geographically stretches from latitude 23.09.36° N to 23.34.01° N and longitude 85.08.40° E to 85.35.57° E, and the total geographical area is 5097 sq km (Fig. 1c) (https://ranchi.nic.in/). It features a hilly, valley-like topography with a humid subtropical climate. Ranchi’s elevation above sea level is 651 m. Among the three study sites, Ranchi receives the most rainfall (1316.1 mm). Ranchi enjoys more pleasant weather than the other two locations, with summer temperatures between 24 and 36 °C and winter temperatures between 10 and 24 °C. Over this location, the relative humidity is typically high (80%) (IMD, 2021). The Ranchi urban cluster is India’s 46th-largest city, according to the 2011 census, and urbanized rapidly after becoming the capital of Jharkhand state.
A drastic increase in vehicular concentration and associated emissions from vehicles, building construction, and industrial activities have increased significantly due to speedy urbanization. This destroyed natural space and the environmental state of all three cities progressively. Thus, the lockdown period provided the ideal ecological condition to quantify the impact of anthropogenic activities over these sites.
Dataset
In this work, we investigated the remote sensing data of LST, NDVI, and Pv assessed from Landsat 8 Level-1 OLI (Operational Land Imager) and TIRS (Thermal Infrared Sensor) through the US Geological Survey (USGS) (https://earthexplorer.usgs.gov/), and the details of the data are provided in Table 1. This study has been performed for 2019, 2020, and 2021 (for some Jan, Feb, Mar, Apr, May, Nov) divided into pre-lockdown, lockdown, and post-lockdown periods.Table 1Corresponding dates, image IDs, and other details about satellite imagesStudy areaDateImage IDCloud cover (%)Path/RowBokaro12-01-2019LC08_L1TP_140044_20190112_20190131_01_T10.08140/4413-02-2019LC08_L1TP_140044_20190213_20190222_01_T10.00140/4417-03-2019LC08_L1TP_140044_20190317_20190325_01_T10.17140/4402-04-2019LC08_L1TP_140044_20190402_20190421_01_T10.00140/4420-05-2019LC08_L1TP_140044_20190520_20190604_01_T10.51140/4428-11-2019LC08_L1TP_140044_20191128_20191203_01_T10.60140/4415-01-2020LC08_L1TP_140044_20200115_20200127_01_T10.01140/4416-02-2020LC08_L1TP_140044_20200216_20200225_01_T10.00140/4419-03-2020LC08_L1TP_140044_20200319_20200326_01_T122.34140/4404-04-2020LC08_L1TP_140044_20200404_20200410_01_T14.42140/4422-05-2020LC08_L1TP_140044_20200522_20200527_01_T10.12140/4414-11-2020LC08_L1TP_140044_20201114_20210317_01_T10.00140/4401-01-2021LC08_L1TP_140044_20210101_20210308_01_T10.02140/4402-02-2021LC08_L1TP_140044_20210202_20210306_01_T10.01140/4422-03-2021LC08_L1TP_140044_20210322_20210401_01_T10.83140/4407-04-2021LC08_L1TP_140044_20210407_20210416_01_T10.03140/44Dhanbad12-01-2019LC08_L1TP_140043_20190112_20190131_01_T10.67140/4313-02-2019LC08_L1TP_140043_20190213_20190222_01_T10.25140/4317-03-2019LC08_L1TP_140043_20190317_20190325_01_T10.81140/4302-04-2019LC08_L1TP_140043_20190402_20190421_01_T10.14140/4320-05-2019LC08_L1TP_140043_20190520_20190604_01_T10.03140/4312-11-2019LC08_L1TP_140043_20191112_20191115_01_T10.28140/4331-01-2020LC08_L1TP_140043_20200131_20200211_01_T10.49140/4316-02-2020LC08_L1TP_140043_20200216_20200225_01_T10.75140/4304-04-2020LC08_L1TP_140043_20200404_20200410_01_T12.37140/4322-05-2020LC08_L1TP_140043_20200522_20200527_01_T19.45140/4314-11-2020LC08_L1TP_140043_20201114_20210317_01_T10.11140/4301-01-2021LC08_L1TP_140043_20210101_20210308_01_T10.51140/4302-02-2021LC08_L1TP_140043_20210202_20210306_01_T12.07140/4322-03-2021LC08_L1TP_140043_20210322_20210401_01_T10.02140/4307-04-2021LC08_L1TP_140043_20210407_20210416_01_T10.53140/43Ranchi12-01-2019LC08_L1TP_140044_20190112_20190131_01_T10.08140/4413-02-2019LC08_L1TP_140044_20190213_20190222_01_T10.00140/4417-03-2019LC08_L1TP_140044_20190317_20190325_01_T10.17140/4402-04-2019LC08_L1TP_140044_20190402_20190421_01_T10.00140/4420-05-2019LC08_L1TP_140044_20190520_20190604_01_T10.51140/4412-11-2019LC08_L1TP_140044_20191112_20191115_01_T10.00140/4415-01-2020LC08_L1TP_140044_20200115_20200127_01_T10.01140/4416-02-2020LC08_L1TP_140044_20200216_20200225_01_T10.00140/4404-04-2020LC08_L1TP_140044_20200404_20200410_01_T14.42140/4422-05-2020LC08_L1TP_140044_20200522_20200527_01_T10.12140/4414-11-2020LC08_L1TP_140044_20201114_20210317_01_T10.00140/4401-01-2021LC08_L1TP_140044_20210101_20210308_01_T10.02140/4402-02-2021LC08_L1TP_140044_20210202_20210306_01_T10.01140/4406-03-2021LC08_L1TP_140044_20210306_20210312_01_T10.01140/4407-04-2021LC08_L1TP_140044_20210407_20210416_01_T10.03140/44
The Landsat 8’s TIRS and OLI sensors have two thermal and nine spectral bands. The TIRS bands have a spatial resolution of 100 m, albeit it is resampled to 30 m for the end user. The spatial resolution of the spectral bands is 30 m, except for band 8 (panchromatic band), with a spatial resolution of 15 m (Table 2). The tool used to study the imageries is ArcGIS 10.8 software (https://www.esri.com/en-us/home). The criteria for selecting the data for the study were data availability and low cloud cover.Table 2Band details of the Satellite imagery employed in this studyBandSpectral resolution (micrometer)Spatial resolution (meter)Aerosol/coastal—band 10.42–0.4530Visible blue (BLUE)—band 20.44–0.5130Visible green (GREEN)—band 30.51–0.6030Visible red (RED)—band 40.62–0.6730Near-infrared (NIR)—band 50.83–0.8730Short wavelength infrared 1 (SWIR1)—band 61.55–1.6530Short wavelength infrared 2 (SWIR2)—band 72.09–2.2930Panchromatic—band 80.49–0.6715Cirrus—band 91.35–1.3830Thermal infrared (TIRS) 1—band 1010.2–11.2100*(30)Thermal infrared (TIRS) 2—band 1111.4–12.4100*(30)*indicates the original spatial resolution of TIRS 1 and 2 bands 10 and 11
Determination of land surface temperature (LST)
The following steps are adopted to process and retrieve LST and NDVI data.
Top of atmospheric spectral radiance
To generate spectral radiance imageries from the raw radiance data, spectral radiance (Lλ) values are required, and the following formula is used for the estimation of the same (Guha & Govil, 2021; Maithani et al., 2020; USGS, 2019):
1\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$
{L}{\lambda }={M}{L}\times {Q}{cal}+{A}{L}
#### Brightness temperature
The Earth’s emitted radiations are measured in terms of brightness temperature (BT) and extensively used parameters in the Earth’s remote sensing (Latif, 2014). Equation (2) is used to convert the reflectance into BT. The thermal constants provided with the Landsat image metadata file are used in this equation (Taoufik et al., 2021):2\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$
BT=\frac{{K}_{2}}{\mathit{ln}[(\frac{{K}_{1}}{{L}_{\lambda }})+1]}-273.15
$$\end{document}BT=K2ln[(K1Lλ)+1]-273.15
The thermal conversion constants for OLI/TIRS data are *K*~1~ = 774.89 and *K*~2~ = 1321.08 (Wm^−2^ sr^−1^ mm^−1^). The radiant temperature is corrected by adding absolute zero to convert the kelvin values into celcius.
#### Land surface emissivity
The land surface emissivity (LSE (*ε*)) is the efficiency with which thermal energy is transmitted from the surface into the atmosphere. It is a proportionality factor that scales blackbody radiance (Planck’s law) to estimate emitted radiance (Jiménez-Muñoz et al., 2006). To estimate LST, first, we need to assess LSE, and the following formula is used to calculate the LSE:3\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$
\mathrm{LSE}\left(\epsilon \right)=0.004.{\mathrm{P}}_{\mathrm{v}}+0.986
$$\end{document}LSEϵ=0.004.Pv+0.986where P~v~ is the proportion of vegetation.
#### Estimation of NDVI and P~v~
P~v~ is an essential vegetation parameter related to surface processes and is frequently used to monitor biodiversity, climate modeling, and various urban models(Gutman & Ignatov, 1998). It is the proportion of the vertical projection area of the vegetation on the surface to the total vegetated area (Deardorff, 1978). P~v~ is defined as follows (Arrofiqoh & Setyaningrum, 2021):4\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$
{\mathrm{P}}_{\mathrm{V}}={\left(\frac{\mathrm{NDVI}-{\mathrm{NDVI}}_{\mathrm{min}}}{{\mathrm{NDVI}}_{\mathrm{max}}-{\mathrm{NDVI}}_{\mathrm{min}}}\right)}^{2}
$$\end{document}PV=NDVI-NDVIminNDVImax-NDVImin2where NDVI~min~ is the NDVI of soil, and NDVI~max~ is the NDVI of vegetation.
NDVI estimates the density of green and vegetated regions over an area. NDVI is essential in surface classifications, biodiversity estimation, urban modeling, urban heat island estimations, and climate change-related studies. For the assessment of LST, analysis of NDVI and P~v~ are essential. As shown below, by combining the near-infrared (Band 5) and red (Band 4) bands, the NDVI is determined (Logan et al., 2020):5\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$
\mathrm{NDVI}=\left(\frac{\mathrm{Band }5-\mathrm{Band }4}{\mathrm{Band }5+\mathrm{Band }4}\right)
$$\end{document}NDVI=Band5-Band4Band5+Band4
The NIR and red bands represent bands 5 and 4 for OLI/TIRS data. The NDVI scale runs from − 1 to + 1, while the P~v~ values vary between 0 and 1. The NDVI values below 0.1 indicate the presence of barren land, rocks, sand, snow, and water. Moderate values within 0.2 to 0.3 indicate the abundance of shrubs and grassland, and very high values (i.e., 0.6–0.8) indicate dense and tropical rainforests. NDVI is the most frequently used parameter to study the vegetation cover of an area. Negative NDVI values are usually associated with water bodies, while high positive NDVI values are associated with lush, flourishing flora. NDVI is a critical remote sensing metric employed in this investigation to determine the impact of changing environments on LST.
#### Land surface temperature
The LST is finally calculated using Eq. (6) (Sahani et al., 2021):6\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$
\mathrm{LST}=\frac{\mathrm{BT}}{[1+\{(\lambda \times \frac{\mathrm{BT}}{\rho })\mathrm{ln}\epsilon \}]}
$$\end{document}LST=BT[1+{(λ×BTρ)lnϵ}]where BT is the brightness temperature at the sensor, *ln*
\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$
\epsilon
$$\end{document}ϵ is the emissivity, *λ* is the top of atmospheric reflectance, and \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$
\rho
$$\end{document}ρ is calculated using Planck’s equation (Arrofiqoh & Setyaningrum, 2021):7\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$
\rho =\frac{c}{\sigma }\times h=1.438\times {10}^{-2}\mathrm{mK}
$$\end{document}ρ=cσ×h=1.438×10-2mKwhere *c* is the speed of light (2.998 × 10^8^ m/s), \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$
\sigma
$$\end{document}σ is the Boltzmann constant (1.38 × 10^−23^ J/K), and *h* is Planck’s constant (6.626 × 10^−34^ J s).
The methodology used to estimate the LST and NDVI is depicted in Fig. 2 with the help of a flowchart.Fig. 2Flowchart for LST and NDVI retrieval
### Statistical test used in the study
In this work, we have estimated various statistical parameters such as the data’s mean, maximum, minimum, and standard deviation and the correlation coefficient between LST and NDVI. For the statistical validation of the data, we have also employed particular statistical tests like the ANOVA and Tukey tests. The theory of these tests is provided in the following paragraphs.
An ANOVA (analysis of variance) test is a sort of statistical analysis that checks for variance-based mean differences to see if there is a statistically significant difference between two or more category groups. A “one way” and “two way” variance test are the two most common ANOVA tests. A one-way ANOVA comprises one categorical independent variable and one normally distributed continuous dependent variable. A normally distributed continuous dependent variable and two or more categorical independent variables make up a two-way ANOVA. Datasets are analyzed using one-way analysis of variance (ANOVA) using the formula \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$
F = \frac{\mathrm{MST}}{\mathrm{MSE}}
$$\end{document}F=MSTMSE, where *F* is the ANOVA coefficient, MST is the mean sum of squares, and MSE is the mean sum of squares due to error. The *F* statistic, the outcome of the ANOVA formula, enables the study of numerous data sets to ascertain the variability within and across samples. The *F*-value in the ANOVA test also determines the *p*-value, which assesses whether or not the difference among the group means is statistically significant. In our study, *p* < 0.05, having a confidence interval (CI) of 95%, which further includes lower confidence interval (LCL), and upper confidence interval (UCL) values, was considered statistically significant.
We have also used the Tukey test along with the ANOVA test. The Tukey comparison test is one of the numerous tests that can identify which mean values among a group of mean values differ from the others. To do the analysis, first, we use the ANOVA test to evaluate that the means of populations differ. If this test concludes that the group means varies, then the Tukey comparison test is used. The LST, P~v,~ and NDVI-based results as tested based on these statistical tests.
## Results and discussion
We have examined the spatial variance of LST, NDVI, and P~v~ caused by the lockdown in the current work. To compare the results, the year 2019—the year before the lockdown—was used as the base year.
### Spatial variations in LST
LST is an important environmental and climatic parameter that is affected by a number of physical and atmospheric events. Cloudiness, solar irradiance, the study month, urban morphology, surface reflectance and type, and a number of other variables all affect how LST is distributed across a region. To compare the changes in the spatial variations of LST from pre-lockdown to full/partial lockdowns, we prepared LST spatial variation maps (Figs. 3 and 4). The surface temperature over our study sites before (April 2019), during (April 2020), and after (April 2021) the lockdown is shown in these figures. Spatial plots of various parameters are shown only for 1 month (April), though analysis is performed for different months of the year. Figure 3 gives the spatial variation of LST for the pre-lockdown period (i.e., 2019, Fig. 3 (1)), full lockdown period (i.e., 2020, Fig. 3 (2)), and their differences (Fig. 3 (3)) for Bokaro, Dhanbad, and Ranchi (A, B, and C) for April. The spatial variation indicates that the LST significantly decreased for all three sites. In general, this region experiences high incoming solar radiation.Fig. 3Spatial distribution of LST for **A** Bokaro, **B** Dhanbad, **C** Ranchi where (1) April 2019, (2) April 2020, (3) difference (2020–2019) (in **A**, **B**, **C**)Fig. 4Spatial distribution of LST for **A** Bokaro, **B** Dhanbad, **C** Ranchi where (1) April 2019, (2) April 2021, (3) difference (2021–2019) (in **A**, **B**, **C**)
The LST, which is directly linked to terrain morphology, surface characteristics, and radiation, has significantly decreased as anthropogenic activities have decreased. Though all three sites showed a decrease in LST as an aftereffect of the lockdown, the decrease was higher for Bokaro and Dhanbad than for Ranchi. The Jharkhand region is rich in minerals, and mining activities are happening in several places. Bokaro and Dhanbad are leading coal, quartzite, stone, and sand mining areas. And thus, Mining activities are much higher over these sites than in Ranchi. Coal mining with other minerals was also affected significantly in this period. The mining activities and transport processes associated with it cause a significant decrease in emissions of aerosols, especially the black carbon aerosols, in the environment, and results could be seen in terms of LST decrease. Due to mining site regions, a decrease in absorbing aerosol concentration at the surface will contribute towards reducing LST. For Bokaro and Dhanbad regions, spatial differences plots show that over these sites, a colossal decrease of about 20 °C was noticed (Fig. 3A and B). Ranchi also showed a substantial decrease in the LST with a maximum decrease of about 10 °C. The spatial difference plot for all sites indicates a decrease over most pixels.
We noticed a similar effect during other lockdown months though figures are not portrayed here. We have also studied the spatial changes in the year 2021. Figure 4 depicts the same as Fig. 3 but for 2021 instead of 2020. Again in 2021, a decrease was noticed in LST over Bokaro and Ranchi though a small portion showed positive values also (Fig. 4A and C). But the Dhanbad site showed a significant increase in LST values compared to the pre-covid year (Fig. 4B) over several pixels. The increase ranged between 3 and 6 °C at about 50% of spatial coverage. Again, we performed the same analysis for several months, but the figure is shown for April. These spatial variations provided qualitative information about the changes in the LST in terms of decrease and increase.
Still, we must perform statistical analysis for quantitative data for these parameters. To achieve quantitative statistical information, we estimate various statistical parameters such as the mean, maximum, minimum, and standard deviation of LST data (Table 3). Table 3 describes maximum, minimum, mean temperature, and standard deviation information. We have also performed the ANOVA and Tukey tests to statistically verify the mean/maximum/minimum values. The *p*-value and *F* values indicated in the table correspond to the mean values of the LST obtained from the ANOVA test. The *p*-value determines whether the difference between group means is statistically significant or not. Figure 5 shows a representative plot for the ANOVA test for Bokaro for April. The statistical values obtained from the ANOVA test confirm that the mean values were significantly different at a 0.05% significance level for all the months. We performed the Tukey comparison test to verify the findings from the ANOVA test, which confirms the monthly mean LST for respective months was significantly different except for February (2019–2020) at Dhanbad and April (2019–2021) at Bokaro. Detailed tables with results from the ANOVA and Tukey tests are supplied as Supplementary Material.Table 3Descriptive statistics of LST for Bokaro, Dhanbad, and Ranchi using a one-way ANOVA test
Study areasMonth201920202021Sig (*p*, *F* value)MaxMinMeanSDMaxMinMeanSDMaxMinMeanSDBokaroJan29.0413.6019.761.2928.1515.7620.861.1426.2514.4020.661.36(0.0, 68.6)Feb34.2116.3424.881.6033.3516.1825.272.0728.1614.6423.501.85(0.0, 158)Mar35.9819.2327.251.3730.5518.3620.186.3540.6123.4134.512.07(0.01, 9.94)Apr46.2225.0638.343.1639.682530.834.4042.8725.6637.142.16(0.0, 7.88)May42.3026.4336.502.4638.9926.0630.151.15––––Nov31.6117.1123.551.1132.7620.1126.121.43––––DhanbadJan32.1215.8020.731.4030.1315.8321.081.4228.8415.9921.321.57(0.0,18)Feb37.6518.1925.351.7233.6118.2225.482.0530.3517.0822.821.80(0.0, 1.64)Mar37.4821.3328.401.5335.9920.3611.908.0640.9924.0634.482.40(0.0, 16.82)Apr43.6926.0934.992.2941.2224.9531.443.6743.3925.1136.382.18(0.02, 2.49)May42.4527.8435.202.4636.1124.4929.561.25––––Nov34.4921.2124.931.2234.5222.1826.051.39––––RanchiJan26.3015.1919.481.3230.0917.5521.581.2229.5216.5021.261.48(0.0, 3.12)Feb32.5918.3525.331.7133.5318.6624.991.8932.2916.8423.691.87(0.0, 158)Mar29.9920.8125.750.9734.8221.5423.911.4139.6021.5533.062.00(0.00, 30.43)Apr44.1825.3737.782.1338.8312.5832.801.7239.9324.3036.001.73(0.04, 5.46)May43.5226.3638.342.1942.0925.3533.672.98––––Nov31.0420.4123.841.1632.8121.4726.371.54––––*p* values less than 0.05 (*p* < 0.05) are statistically significant, and the groups’ means are significantly different(-) dash denotes the unavailability of data*Max* Maximum, *Min* Minimum, *SD* Standard DeviationFig. 5Comparison of maximum and minimum LST values between three consecutive years (2019, 2020, 2021) in Bokaro (**a** and **b**), Dhanbad (**c** and **d**), and Ranchi (**e** and **f**)
Along with the descriptive information provided in Table 3, the lockdown-induced changes in the maximum/minimum values are pictorially also shown in Fig. 6. The mean values (± standard deviation) are pictorially shown in Fig. 10, along with the mean values of NDVI and P~v~. During the lockdown period, a notable decrease in LST was observed over all three sites. The same trend was seen in the maximum temperature in the lockdown months (April and May 2020), though values were only a few degrees smaller (Fig. 6. In November, with the relaxation in restrictions, the maximum of LST was equivalent to 2019 (Fig. 6). Minimum temperature showed a slight decrease in the values during the lockdown months of 2020, though, over Ranchi, a drastic decrease was seen in minimum values in April (Fig. 6). We have also performed the ANOVA and Tukey tests to statistically verify these maximum/minimum LST values. The information from these statistical tests (ANOVA/Tukey) state that maximum/minimum values for respective months during 2020/2021 were significantly different from the same month in 2019, except for a few months. These values were insignificant during February (19–20) and April (19–21) for Bokaro for minimum temperature. Dhanbad’s maximum LST difference between 2019 and 2020 during November was insignificant, while the minimum LST difference was insignificant during January and February. Except for these months, a significant difference between maximum/minimum LST was found from the ANOVA test, and the same was validated with the Tukey test. The statistical values from ANOVA/Tukey test for maximum/minimum temperature are not included in the table (supplied as Supplementary Material).Fig. 6Outcome from the ANOVA a representative plot for April month LST mean (Bokaro)
### Lockdown-induced changes in NDVI and P~v~
High-resolution remote sensing photos are used to create various bio-physical indices that can be used to measure changes in environmental factors related to urban morphology. The normalized difference vegetation index (NDVI) and the proportion of vegetation (P~v~), which are widely used in studies related to urban heat, have undergone alterations in this regard (Neinavaz et al., 2020). These factors substantially impact the LST and are linked to them.
Figure 7 gives the spatial variation of NDVI during 2019 and 2020 and the differences between these years for April for Bokaro, Dhanbad, and Ranchi. Bokaro and Dhanbad sites showed a very substantial improvement in the value of NDVI, which reflects the increase in the vegetation cover over these sites as an effect of anthropogenic activity restrictions. The rise in NDVI over Bokaro and Dhanbad sites was as high as 0.27 and 0.2 (Fig. 7A and B). Over the Ranchi, also increment in NDVI values was observed, though slightly lesser than the other two sites. A rise in positive NDVI values indicates an increase in the vegetated region. According to these figures, the number of pixels with NDVI values in the 0.2–0.5 range increased considerably in 2020. The NDVI values in this range indicate sparse to dense vegetations type. Thus, the lockdown contributed positively to the growth of greenery. The reduction in LST values and fewer coal particle concentrations are possible causes for the increase in NDVI values during this time. An increase in cultivation activities owing to the lockdown may also be a probable cause of the rise in NDVI values.Fig. 7Spatial distribution of NDVI for **A** Bokaro, **B** Dhanbad, **C** Ranchi where (1) April 2019, (2) April 2020, (3) difference (2020–2019) (in **A**, **B**, **C**)
Next, we studied the changes in the NDVI during the same month in 2021 (Fig. 8). In the year 2021, when several anthropogenic activities had resumed, these effects were reflected in NDVI values. These plots show that over most of the pixels, NDVI values decreased in general during the year 2021, but some nominal increase could also be noticed over a few pixels. Similar to LST, this analysis is also performed for several months, but only a spatial plot is shown for April.Fig. 8Spatial distribution of NDVI for **A** Bokaro, **B** Dhanbad, **C** Ranchi where (1) April 2019, (2) April 2021, (3) difference (2021–2019) (in **A**, **B**, **C**)
Next, we studied the heterogeneity that occurred in the P~v~ distribution causes as an aftereffect of the lockdown over the study regions. Figure 9 shows the spatial variation of P~v~ for Bokaro (Fig. 9A), Dhanbad (Fig. 9B), and Ranchi (Fig. 9C). Similar to other surface parameters, a significant improvement was noticed in this indicator also. In 2020, P~v~ values significantly improved, and impact of these improvements, we have already observed NDVI plots. Spatial variation of P~v~ showed that Ranchi has the highest greenery among all the sites. As Bokaro and Dhanbad are prime mining sites, thus, the restriction on anthropogenic activities gave favorable conditions for the natural resources. As an effect, P~v~ also showed improvement in the year 2020 during the lockdown months. But once anthropogenic activities resumed, P~v~ values decreased in 2021. The pixels with lower P~v~ as compared to 2020 increased in 2021.Fig. 9Spatial distribution of P~v~ for **A** Bokaro, **B** Dhanbad, **C** Ranchi where (1) April 2019, (2) April 2020, (3) April 2021(in **A**, **B**, **C**)
The spatial value plot was shown for April, but a decrease/increase in LST/NDVI and P~v~ was also observed in other months of lockdowns. The spatial plot of NDVI/P~v~ also provides qualitative information regarding the changes in spatial variation. To assess the quantitative information, we have estimated the differences in these parameters’ mean, maximum, and minimum values. The ANOVA and Tukey tests have also checked the significance of the change in these values. The statistical information regarding the LST is shown in Table 3 and discussed in the previous section. The statistical information regarding the changes in maximum/minimum, mean values, and standard deviation of NDVI and P~v~, along with details from the statistical test (*p*- and *F* values), which check their significance, are provided in Tables 4 and 5 respectively. Figure 10 depicts the changes that occurred in the mean values of these parameters for various months in the year 2019, 2020, and 2021. The figures in the first row, i.e., Fig. 10a, b, and c, are for the Bokaro, the second for Dhanbad, and the third for the Ranchi. This plot reveals that the LST values decreased during the lockdown period and increased after normal activities were resumed. The values have risen again in the year 2021. The NDVI/P~v~ values increased in the lockdown period, which again dropped after the ease of restrictions from anthropogenic activities. The *p*-values from the ANOVA test indicate the significance of the difference of maximum/minimum/mean values of 2020/2021 from 2019 for respective months regarding the NDVI and P~v~.Table 4Descriptive statistics of NDVI for Bokaro, Dhanbad, and Ranchi using a one-way ANOVA testStudy areasMonths201920202021Sig (*p*, *F* value)MaxMinMeanSDMaxMinMeanSDMaxMinMeanSDBokaroJan0.30 − 0.080.120.050.36 − 0.060.170.060.30 − 0.060.120.05(0.0, 14)Feb0.34 − 0.070.150.050.36 − 0.110.180.060.32 − 0.060.130.05(0.004, 8.66)Mar0.31 − 0.050.160.040.35 − 0.170.190.050.31 − 0.050.150.04(0.001, 12.66)Apr0.47 − 0.080.180.050.36 − 0.050.200.060.34 − 0.050.140.04(0.0, 18.66)May0.45 − 0.050.230.060.51 − 0.060.310.08––––Nov0.38 − 0.110.180.050.39 − 0.100.220.07––––DhanbadJan0.30 − 0.080.110.050.37 − 0.060.170.070.32 − 0.070.110.05(0.0, 20.66)Feb0.37 − 0.100.140.050.39 − 0.070.160.060.29 − 0.070.100.05(0.0, 20.66)Mar0.32 − 0.070.150.040.34 − 0.080.160.050.31 − 0.070.140.05(0.177,2)Apr0.43 − 0.100.170.050.40 − 0.080.200.070.44 − 0.040.140.040.0, 18)May0.44 − 0.050.250.070.47 − 0.050.300.08––––Nov0.41 − 0.170.230.080.44 − 0.070.270.08––––RanchiJan0.34 − 0.050.140.050.38 − 0.050.190.060.32 − 0.090.150.04(0.004, 8.66)Feb0.47 − 0.060.170.050.43 − 0.070.200.060.41 − 0.150.170.05(0.004, 8.66)Mar0.44 − 0.040.170.040.43 − 0.060.180.050.42 − 0.100.170.05(0.53, 0.66)Apr0.48 − 0.100.210.050.41 − 0.070.240.050.35 − 0.040.150.03(0.0, 32.66)May0.46 − 0.050.220.060.51 − 0.050.290.07––––Nov0.43 − 0.070.260.060.47 − 0.170.240.07––––*p* values less than 0.05 (*p* < 0.05) are statistically significant, and the groups’ means are significantly different(-) dash denotes the unavailability of data*Max* Maximum, *Min* Minimum, *SD* Standard DeviationTable 5Descriptive statistics of Pv for Bokaro, Dhanbad, and Ranchi using a one-way ANOVA testStudy areasMonths201920202021Sig (*p*, *F* value)MaxMinMeanSDMaxMinMeanSDMaxMinMeanSDBokaroJan0.470.180.310.030.510.170.340.030.470.190.320.03(0.03, 4.66)Feb0.510.150.330.030.530.140.350.030.500.160.320.03(0.18, 2)Mar0.530.170.340.020.570.150.360.030.520.160.330.02(0.03, 4.66)Apr0.600.140.350.030.580.170.360.040.510.190.330.02(0.00, 8.66)May0.580.170.380.040.620.160.430.05––––Nov0.510.180.350.030.570.160.380.04––––DhanbadJan0.450.190.310.020.560.140.340.040.480.170.310.03(0.00, 8.66)Feb0.530.140.320.030.540.130.330.030.460.170.300.03(0.03, 4.66)Mar0.510.180.330.020.530.170.340.020.530.170.320.02(0.18, 2)Apr0.600.140.340.030.610.140.360.040.520.190.320.02(0.00, 8)May0.570.200.390.040.620.160.420.05––––Nov0.530.170.380.050.560.150.370.04––––RanchiJan0.490.170.330.030.560.130.360.030.510.170.330.03(0.00, 8.66)Feb0.560.130.340.030.610.120.360.030.560.120.340.03(0.11, 2.66)Mar0.560.170.350.020.600.140.360.030.600.110.340.03(0.18, 2)Apr0.660.110.370.030.610.140.380.030.540.180.330.02(0.0, 14)May0.600.150.370.040.620.140.420.05––––Nov0.560.160.400.030.580.120.380.04––––*p* values less than 0.05 (*p* < 0.05) are statistically significant, and the groups’ means are significantly different(-) dash denotes the unavailability of data*Max* Maximum, *Min* Minimum, *SD* Standard DeviationFig. 10Lockdown induced changes in mean (± standard deviation) of LST, NDVI, and P~v~ values over Bokaro (**a**, **b**, **c**), Dhanbad (**d**, **e**, **f**), and Ranchi (**g**, **h**, **i**), respectively
The relationship between the NDVI and the LST changes depending on the season and the time of the day (Sun & Kafatos, 2007). The association between LST and NDVI is commonly reported to be negative (Deng et al., 2018; Guha et al., 2020); however, it can be negative in the summer and positive in the winter and early spring seasons (Kaufmann et al., 2003; Sun & Kafatos, 2007). In this study, we have also investigated the changes in these parameters during the lockdown period. Representative scatter plots between the LST and NDVI are shown in Fig. 11 for May (2019/20) and April (2021) (Table 6). A negative correlation could be noticed in these two parameters, and an increase in the negative correlation is seen in these values, which is consistent with our other results. As our results showed a decrease in LST and an increase in the NDVI/P~v~ value, thus, the correlation between LST-NDVI turned more negative.Fig. 11Scatter plot linear regression of LST–NDVI relationship for **A** Bokaro, **B** Dhanbad, and **C** Ranchi in (1) 2019, (2) 2020, (3) 2021Table 6Analysis of LST–NDVI relationship based on Pearson’s correlation coefficientsStudy areasMonthsPearson’s correlation coefficient (*r*)201920202021BokaroJanuary − 0.35 − 0.32 − 0.37February − 0.48 − 0.45 − 0.33March − 0.34 − 0.43 − 0.47April − 0.20 − 0.45 − 0.53May − 0.35 − 0.65November − 0.30 − 0.49DhanbadJanuary − 0.10 − 0.41 − 0.30February − 0.25 − 0.36 − 0.23March − 0.33 − 0.40 − 0.19April − 0.29 − 0.35 − 0.45May − 0.47 − 0.60November − 0.44 − 0.57RanchiJanuary − 0.37 − 0.25 − 0.32February − 0.45 − 0.43 − 0.45March − 0.17 − 0.31 − 0.45April − 0.41 − 0.43 − 0.49May − 0.52 − 0.69November − 0.30 − 0.48
## Conclusions
In this paper, we looked into how the lockdown affected the surface and vegetation indices, which are UHI markers. The change in anthropogenic activity over this area is principally responsible for these alterations. We looked at the variations in the LST, NDVI, and P~v~ values as well as the relationship between the LST and NDVI. The prime finding of the work is as The LST values decreased significantly over study sites in the lockdown months; this decrease was maximum over mining sites (i.e., Bokaro and Dhanbad). LST values decline in a range of 13–19% over these sites during the duration of the lockdown. Aerosol emissions, particularly black carbon aerosols, are significantly reduced by restricted mining activities and the accompanying transport processes. This has impacted the environment and possibly contributed to the decline in LST. Decreased concentrations of absorbing aerosols at the surface result in reduced LST values in mining site areas.The NDVI and P~v~ values increased substantially over all sites in the lockdown period in the year 2020. About 15% was noticed in the NDVI values over all the sites. P~v~ values also improved in the same range as the NDVI value. The increase in NDVI readings is probably due to a decrease in LST values and coal particle concentrations. The lockdown may have increased farming efforts, another likely factor contributing to the rise in NDVI readings.The negative correlation between LST and NDVI increased significantly during the lockdown period compared to 2019.The statistical significance of changes in the LST, NDVI, and P~v~ values compared to 2019 is checked using the ANOVA and Tukey test, which indicated the difference in mean LST/NDVI/P~v~ was statistically significant for the lockdown period compared to the pre-lockdown period.This study shows that lockdown on anthropogenic activities has significantly improved the UHI indicators, i.e., reduction in LST values and improvement in NDVI/P~v~ values. These improvements were slightly higher over the mining sites (i.e., Bokaro and Dhanbad) than non/less mining sites (i.e., Ranchi).The findings of this study show that lockdowns had a beneficial effect on the study locations. Reduced LST and rising NDVI/Pv readings point to more vegetation in the study area. This argues that urban planners should create vegetation on the existing bare plains to reduce urban heating.
## Supplementary Information
Below is the link to the electronic supplementary material.Supplementary file1 (DOCX 28 KB)