key: cord-0975558-ryt3lzum
authors: Sarkodie, Samuel Asumadu; Owusu, Phebe Asantewaa
title: Impact of meteorological factors on COVID-19 pandemic: Evidence from top 20 countries with confirmed cases
date: 2020-08-22
journal: Environ Res
DOI: 10.1016/j.envres.2020.110101
sha: eb4c0dfda5bf899635cab73552795c7c9ee4b739
doc_id: 975558
cord_uid: ryt3lzum

The global confirmed cases of COVID-19 have surpassed 7 million with over 400,000 deaths reported. However, 20 out of 187 countries and territories have over 2 million confirmed cases alone, a situation which calls for a critical assessment. The social distancing and preventive measures instituted across countries have a link with spread containment whereas spread containment is associated with meteorological factors. Here, we examine the effect of meteorological factors on COVID-19 health outcomes. We develop conceptual tools with dew/frost point, temperature, disaggregate temperature, wind speed, relative humidity, precipitation and surface pressure against confirmed cases, deaths and recovery cases. Using novel panel estimation techniques, our results find strong evidence of causation between meteorological factors and COVID-19 outcomes. We report that high temperature and high relative humidity reduce the viability, stability, survival and transmission of COVID-19 whereas low temperature, wind speed, dew/frost point, precipitation and surface pressure prolong the activation and infectivity of the virus. Our study demonstrates the importance of applying social distancing and preventive measures to mitigate the global pandemic.

The novel coronavirus (COVID-19) aka severe acute respiratory syndrome coronavirus 2 (SARS-26

CoV-2) has received much attention due to its impact on the environment, socio-economic 27 development and health outcomes (Sarkodie and Owusu, 2020a ). The first incidence of COVID-19 28 occurred in Wuhan city, China which subsequently spread across countries and was declared a global 29 pandemic by the World Health Organization (WHO) (WHO, 2020). As of May 02, 2020 [6:32 am 30 (GMT+2)], 3,344,099 confirmed cases, 238,663 deaths and 1,053,342 recovery cases had been 31 reported (Lauren, 2020) . Several preventive and social distancing measures have been instituted 32 across countries to contain the spread of COVID-19 (Wilder-Smith and Freedman, 2020). These 33 containment measures are somewhat related to reducing the human-to-human transmission of 34 COVID-19 that might have been acquired through carriers driven by meteorological factors (Li et 35 al., 2020) . Several studies have thus far examined the relationship between COVID-19 and 36 meteorological factors such as temperature , humidity (Liu et al., 2020) , and air 37 pollution . However, there is no single study on other useful meteorological factors 38 such as, inter alia, wind speed, dew/frost point, disaggregate temperature, precipitation and surface 39 pressure. It is reported that COVID-19 has inherent genetic variability that leads to high mutation 40 rates, hence, affecting its adaptation (Martinez, 2020) . Thus, assessment of all useful impact of 41 meteorological factors is essential to empirically understand the virus. 42

Contrary to previous attempts, this study for the first time develops conceptual tools based on novel 43 panel estimation techniques across the top 20 countries with confirmed cases. We expand existing 44 literature to include eight meteorological factors and assess its impact on COVID-19 confirmed 45 cases, deaths and recovery cases. Our empirical results provide new perspectives to understanding 46 the survival, viability, stability and transmission of the virus. 47

To correctly model the complexities of COVID-19 outcomes and its relationship with 72 meteorological factors, we utilized several panel estimation techniques. First, we assessed the 73 characteristics of the dataset using descriptive statistical tools. We applied data normalization 74 technique following the procedure presented in Sarkodie et al. (2020) to correct negative data inputs 75 without losing the structural attributes of the data. 76

Second, we examined potential cross-sectional dependence in panel data. Due to global common 77 shocks (Eberhardt and Teal, 2011) such as, inter alia, pandemics like COVID-19, and climate 78 change, panel data usually suffer from correlation across countries. Hence, ignoring this challenge in 79 model estimation affects statistical inferences. Thus, we controlled for cross-sectional dependence 80 following the procedure presented in Pesaran (2004) . 81

Third, assessing the stationarity properties of panel dataset is useful to avoid spurious statistical 82 interpretations. We utilized the estimation procedure expounded in Pesaran (2007) to investigate the 83 stationarity of the data using CIPS and CADF panel unit root techniques. 84

Next, we examined the panel data for potential heterogeneity and applied the panel-based causality 85 test for the heterogeneous panel. Like cross-sectional dependence, panel data also suffer from 86 heteroskedasticity due to different characteristics across countries. Hence, we controlled for 87 heterogeneity using the modified Wald (MWALD) test for groupwise heteroskedasticity in the fixed-88 effects regression model (Greene, 2000 The descriptive statistical features of the data series are presented in Table 1 Granger causality test for panel with heterogenous slope. The results in Table 3 Table 4 . The estimated coefficient on T2MDEW is positive for confirmed 209 cases and deaths but negative for recovery cases. Meaning that a percentage increase in dew/frost 210 point intensifies confirmed cases and deaths by ~11% (p-value<0.01) but declines recovery cases by 211 0.10% (p-value<0.01). The coefficient on T2M is negative for confirmed cases and death but positive 212 for recovery cases. We observe that a percentage increase in temperature declines confirmed cases 213 and deaths by 0.13% (p-value<0.01) and ~0.11% (p-value<0.01), respectively but improves recovery 214 cases by 10% (p-value<0.01). To ascertain the degree of temperature (cold or warm) that affects 215 COVID-19 outcomes, we investigated disaggregate (minimum and maximum) effects of 216 temperature. The coefficient on T2M_MAX is negative for confirmed cases and death but positive 217 for recovery cases. Thus, increasing maximum temperature by 1% declines confirmed cases by 218 0.13% (p-value<0.01) and deaths by 0.11% (p-value<0.01) but increases recovery cases by ~10% (p-219 value<0.01). In contrast, a percentage increase in minimum temperature upsurges both confirmed 220 cases and deaths by ~10% (p-value<0.01) but declines recovery cases by 0.10% (p-value<0.01). The 221 coefficient on RH2M is negative and statistically significant for confirmed cases and death while the 222 coefficient is positive for recovery cases. In the same way, increasing relative humidity by 1% 223 declines confirmed cases and deaths by ~0.08% (p-value<0.01) but intensifies recovery cases by ~4% 224 (p-value<0.01). 225

While the coefficient on WS2M is positive for confirmed cases and death, we observe a negative 226 coefficient for recovery cases. The empirical evidence shows that a percentage increase in wind 227 speed increase confirmed cases and deaths by almost 2% (p-value<0.01) while recovery cases decline 228 by 0.16% (p-value<0.01). The corresponding coefficient on PRECTOT and PS is positive and 229 statistically significant at 1% level for confirmed cases and deaths but negative for recovery cases. 230 An increase in precipitation by 1% declines recovery cases 0.07% and increases confirmed cases and 231 deaths by 1% and 0.86%, respectively. A percentage increase in surface pressure escalates confirmed 232 cases and deaths by 62% and 48% while it declines recovery cases by 0.07%. 233 234 without human-to-human transmission and community spread through skin contact. 296

Our results confirmed a causal relationship running from the incidence of COVID-19 cases to 297 deaths. In conjunction, while high temperature and high humidity declines COVID-19 deaths and 298 surges recovery, other meteorological factors such as minimum temperature, wind speed, surface 299 pressure, dew/frost point and precipitation increase the rate of deaths from COVID-19 and negate 300 the rate of recovery. Apart from the impact of meteorological conditions on the survival of viruses, 301 there are cases of the reported effect of weather conditions on the human immune system. For 302 example, cold regions with limited exposure to sunlight are reported to have many cases of Vitamin 303 D deficiency compared to tropical regions (Cannell et al., 2006) . Low levels or lack of Vitamin D 304 affects the anti-microbial peptide system responsible for the regulation of human immune response 305 (Fuhrmann, 2010) . Exposure to extremely dry and cold weather conditions is reported to modify the 306 human immune response, hence, the host becomes susceptible to pathogen-causing infections 307 (Fisman, 2007) . This perhaps explains why the majority of COVID-19 attributable deaths reported 308 has potential risk factors or underlying health conditions (Covid et al., 2020) . It is reported that 309

Vitamin D plays a critical role in reducing the risk of COVID-19 infections and death rates (Grant et The empirical results of the study have policy implications for preventing the incidence of COVID-317 19 ( Figure 3 ). The practice of social distancing (ID 1) measures including staying home (ID 2) and 318 preventive measures such as avoiding handshake (ID 3), using face mask (ID 4), handwashing with 319 soap and running water (ID 5) and applying 70% alcohol-based sanitizer (ID 6) -will help reduce 320 the spread of COVID-19. The safety measures in Figure 3 mitigate the effect of meteorological 321 factors in facilitating COVID-19 viral transmission through contact transmission (ID 1-3, 5-6), 322 droplet transmission (ID 1, 3-6), and airborne transmission (ID 4). 323

As a contribution to the global debate on the novel coronavirus (COVID-19), we examined the 325 nexus between meteorological factors and COVID-19 outcomes in the top 20 countries with 326 confirmed cases. Using daily data from 01/22/2020 to 04/27/2020, we utilized a battery of panel 327 estimation techniques to control the complexities of SARS-CoV-2. We found that, first, using 328 disaggregate temperature rather than average temperature provides more insight into understanding 329 the temperature-SARS-CoV-2 nexus. Second, due to potential heterogeneity, using all 330 meteorological factors in a single model lead to erroneous estimates, hence, producing spurious 331 inferences especially in panel data. Thus, controlling for individual-specific effects of meteorological 332 factors on COVID-19 is worthwhile. Our results confirmed a strong causal relationship between 333 meteorological factors and COVID-19 outcomes. While high temperature and high relative humidity 334 were found to reduce incidence cases, low temperature, wind speed, surface pressure, dew/frost 335 point and precipitation were found to facilitate the survival and transmission of COVID-19, hence, 336 increasing confirmed cases, deaths and reducing recovery rates. We propose laboratory examination 337 of the effect of wind speed, surface pressure, dew/frost point and precipitation on coronaviruses. 338 

Survival and disinfection of parainfluenza viruses on 348 environmental surfaces

Epidemic influenza and vitamin D

Effects of air temperature 352 and relative humidity on coronavirus survival on surfaces

The Effects of 354 Temperature and Relative Humidity on the Viability of the SARS Coronavirus

Preliminary estimates of the prevalence of selected underlying health 358 conditions among patients with coronavirus disease 2019-United States

Testing for Granger non-causality in heterogeneous panels

Econometrics for grumblers: a new look at the literature on cross 363 country growth empirics

Seasonality of infectious diseases

The effects of weather and climate on the seasonality of influenza: what we 366 know and what we need to know

Evidence that vitamin D supplementation could reduce risk of influenza and COVID-19 369 infections and deaths

Econometric analysis

The role of cathelicidin and defensins in pulmonary 372 inflammatory diseases

Vitamin D and respiratory infection in adults

Dynamic linkage between renewable and conventional energy use, 377 environmental quality and economic growth: Evidence from Emerging Market and Developing 378 Economies

Impact of meteorological factors on the 384 COVID-19 transmission: A multi-city study in China

Effects of 386 temperature variation and humidity on the death of COVID-19 in Wuhan

Compounds with therapeutic potential against novel respiratory 2019 389 coronavirus

2020). meteorological data sets

General diagnostic tests for cross section dependence in panels

A simple panel unit root test in the presence of cross section dependence

Testing for unit roots in heterogeneous panels

Foreign direct investment and renewable energy in 397 climate change mitigation: Does governance matter

Global assessment of environment, health and economic impact 399 of the novel coronavirus (COVID-19)

Investigating the cases of novel coronavirus disease

China using dynamic statistical techniques. Heliyon

Would everyone wearing face masks help us slow the pandemic?

Survival of human coronaviruses 229E and OC43 in 406 suspension and after drying onsurfaces: a possible source ofhospital-acquired infections

To mask or not to mask: WHO makes U-turn while US, Singapore abandon 409 pandemic advice and tell citizens to start wearing masks

Stability of Middle East respiratory 413 syndrome coronavirus (MERS-CoV) under different environmental conditions

Novel Coronavirus (2019-nCoV) -Situation Report -1. SITUATION REPORT -1

Isolation, quarantine, social distancing and community 419 containment: pivotal role for old-style public health measures in the novel coronavirus (2019-nCoV) 420 outbreak

Association between ambient temperature and COVID-19 infection in 122 422 cities from China

Association between short-term exposure to air pollution 424 and COVID-19 infection: Evidence from China