key: cord-0957744-twy07z05
authors: Matthew, Olaniran Jonathan; Eludoyin, Adebayo Oluwole; Oluwadiya, Kehinde Sunday
title: Spatio-Temporal Variations in COVID-19 in Relation to the Global Climate Distribution and Fluctuations
date: 2021-03-19
journal: Spat Spatiotemporal Epidemiol
DOI: 10.1016/j.sste.2021.100417
sha: 5fa84d3f192b2392df2e8bfba652e90f257ca0ef
doc_id: 957744
cord_uid: twy07z05

This study investigates the spatio-temporal variations in the occurrence of COVID-19 (confirmed cases and deaths) in relation to climate fluctuations in 61 countries scattered around the world from December 31, 2019 to May 28, 2020. Logarithm transformation of the count variable (COVID-19 cases) was used in a multiple linear regression model to predict the potential effects of weather variables on the prevalence of the disease. The study revealed strong associations (-0.510 ≤ r ≤ -0.967; 0.519 ≤ r ≤0.999) between climatic variables and confirmed cases of COVID-19 in majority (68.85%) of the selected countries. It showed evidences of 1 to 7-day delays in the response of the infection to changes in weather pattern. Model simulations suggested that a unit fall in temperature and humidity could increase (0.04 -18.70%) the infection in 19.67% and 16.39% of the countries respectively while a general reduction (-0.05 —9.40%) in infection cases was projected in 14.75% countries with a unit drop in precipitation. In conclusion, the study suggests that effective public health interventions are crucial contain the projected upsurge in COVID-19 cases during both cold and warm seasons in the southern and northern hemispheres.

In rapid succession, the first COVID-19 case (a severe acute respiratory syndrome coronavirus 2), which was first reported on December 31, 2019 in the city of Wuhan, China, has caused an outbreak of human-to-human transmission globally. The disease was declared a Public Health Emergency of International Concern (PHEIC) on January 30, 2020 and then a pandemic on March 11, 2020 by the World Health Organization, WHO (Chan et al., 2020; Lai et al., 2020; Li et al., 2020) . Since the inception of the disease, concerted efforts have been made to have a better understanding of the genomics, hosts, modes of transmission and epidemiological link of the disease (Sahu et al., 2020) . Previous studies have revealed that a significant number of respiratory infectious diseases display seasonal patterns in their incidence. However, the impact of climate variability and other extrinsic factors on COVID-19 transmission is still a subject of debate.

Climate is one of many factors likely affecting the spread of the virus (Briz-Redón and Serrano-Aroca, 2020; Di Pietro et al., 2020; Mishra and Wargocki, 2020) and that the host's behaviour (Kraemer et al., 2019) and population density (Geoghegan and Holmes, 2017) are important predictors of the capacity of the virus to spread (Araujo and Naimi, 2020) . Recent studies suggested that infected humans can be asymptomatic and transmit the virus to others, generating substantial uncertainties regarding the overall risk of epidemic outbreaks under a variety of different climate, ecological and social settings (Li et al., 2020) . Specifically, Araujo and Naimi (2020) submitted that immediate physical environment can mediate human-to-human transmission of COVID-19 and that unsuitable climates can cause the virus to destabilize quickly, hence reducing its capacity to become epidemic. Further, the strong association of COVID-19 to a sharp North/South climate gradient, with a faster spread in warm and cold temperate climates have been reported (Araujo and Naimi, 2020; Briz-Redón and Serrano-Aroca, 2020; Di Pietro et al., 2020; Méndez-Arriaga, 2020) . Similarly, Mishra and Wargocki (2020), Sajadi et al. (2020) and Wang et al. (2020a) suggested a close relationship between the incidence of COVID-19 epidemics and climate with countries in high latitudes (characterized by temperate and/or continental climate) exhibiting a high incidence of the disease. On the other hand, Briz-Redón and Serrano-Aroca (2020) and O'Reilly et al. (2020) stressed that since local transmission of the disease has been confirmed to span all climatic zones, further studies on the impact of climate variability, on transmission is vital to increase understanding of the factors underlying the spread of the disease across the globe. Also, scholars have argued that atmospheric pollution could significantly contribute to the anomalous variability of COVID-19 depending on concentrations and chronicity of exposure (Conticini et al., 2020; Fattorini and Regoli, 2020; Zhu et al., 2020) .

Owing to the introduction of lockdown measures (including transportation, businesses, and industrial shutdowns) by governments around the world to limit the COVID-19

pandemic, air quality has improved significantly in major cities of the world (Bashir et al., 2020) . This is because reduced fossil fuel consumption will lower emissions into the atmosphere resulting in cleaner air (Sharif et al., 2020) . About 20 -50% in greenhouse gas emissions (e.g. CO 2 , O 3 , NO x and SO x ) or/and other air pollutants (such as PM2.5, PM10, BC and benzene) in Korea, China, Spain, Germany, Italy, USA and New York compared to preepidemic years were reported in the literature (Collivignarelli et al., 2020; Conticini et al., 2020; Gautam, 2020; Knowland et al., 2020; Wang et al., 2020b) . It has been argued that resumption of large-scale industrial activities after the epidemic will probably reverse the environmental changes, once the epidemic is controlled (Bashir et al., 2020; Bernauer and Slowey, 2020) .

Previous studies suggested that people under lockdown will more likely develop psychological problems like stress disorders, fear, depression, emotional fatigue, and insomnia than people who freely move (Brooks et al., 2020; Fofana et al., 2020) . Fattorini and Regoli (2020), thus, argued that environmental control should be integrated with human health protection into sustainable development as measures for controlling epidemics, on a long-term projection rather than a short-term, incidence-based measure. As new cases of COVID-19 are being confirmed daily around the world, there is a heightened concern, and pressure is mounting on researchers from various fields of study to improve our understanding of the factors underlying the spread of the disease. Some pertinent questions at the moment for COVID-19 mitigation strategies are: (i) Will the virus be less transmissible in hot and humid climates? (ii) Will changes in weather affect the transmission intensity of COVID-19? And (iii) will asynchronous seasonal global outbreaks occur in future if the spread of the disease continues to follow the current trend? The present study, therefore, investigates potential effects of spatio-temporal variations in world's geographical climate (in terms of humidity, temperature and precipitation) on the COVID-19 occurrence (incidence rate and prevalence rate) and its fatality rate. The study is designed to uncover certain underlying trends on the spread of COVID-19 concerning climate variability and link the results with findings from related previous researches with a view to providing useful information on the vulnerability of different climatic regions of the world to infection that could help to curtail its spread.

Laboratory-confirmed infection case series of daily local transmissions of COVID-19 across the globe from the date of first reported case till May 28, 2020 (12.00 GMT) were obtained from https://ourworldindata.org/coronavirus; https://covid19info.live/. The website contained the worldwide country by country and regional COVID-19 data on many aspects of the disease such as the number of infections and death and other epidemiological and surveillance records. In addition, global climate data (hourly time-series of temperature, humidity, and precipitation) of high-resolution ERA5 datasets (freely available at https://cds. climate.copernicus.eu/#!/home) were obtained and analysed. ERA5 reanalysis is the fifth generation of atmospheric reanalysis of the global climate developed by the European Centre for Medium-Range Weather Forecasts (ECMWF). It combines model data with observations from across the world into a globally complete and consistent dataset using the laws of physics with horizontal resolution of 0.25° × 0.25° (Copernicus Climate Change Service, 2017).

We analysed laboratory-confirmed infection case series of daily local transmissions of COVID-19 in 61 countries (representing 33.7% of countries with available COVID-19 data) in Asia, Oceania, Europe, North and South America, Arctic region and Africa from the date of first reported case in each country till May 28, 2020. Similarly, daily means of the climatic parameters over the selected countries during the study period were estimated. In order to ensure that the datasets that were analysed were valid and of good quality, data cleaning process was performed on both datasets to remove missing values. The local transmission ratio (LTR), defined by Méndez-Arriaga (2020) as the number of confirmed positive cases divided by the number of the effective contagion days since community onset of the disease, was estimated.

Multiple regression analysis and lag correlation (0 -7 days) were performed to examine the association or relationship between the daily meteorological parameters i.e. temperature/ o C, relative humidity/% and precipitation/mm (as independent variables) and COVID-19 cases (dependent variable). Before fitting linear regression models and correlation, we first subjected the data to statistical test for collinearity among the variables and normality as suggested by Zar (1992) . As expected, the dependent variable (a count variable that did not conformed to these assumption) was transformed (using logarithm transformation) while the outliers (elements that are greater than the 3 scaled median absolute deviation (MAD) away from the median) that could introduce substantial errors in the outcomes of the analysis, were removed or smoothened (using MATLAB® command -filloutliers‖, version 2019a) before further analysis.

The variance stabilizing logarithm-transformation was performed on the dependent variable, Y as a function of time, t (days) using Eqn. (1) as fully described in Feng et al. 

Where β was the y-intercept, were the regression coefficients for the 3 independent variables (temperature, humidity, and precipitation) respectively and was the random error We adopted kriging interpolation algorithm (which accounts for uncertainties due to associated fit parameters in semivariogram as documented in Davis (1986) and de Smith et al. Globally, the spread of the disease was more pronounced in the northern hemisphere and the prominent direction of the transmission was from China to the western countries.

Estimated local transmission ratio (LTR) for each of the countries scattered over different regions are presented in Fig. 3 . In Africa, LTR values ranged between about 371 and 2 persons day -1 (Fig. 4a) . South Africa had the highest closely followed by Egypt (LTR = 275 persons day -1 ) while the least was recorded in Zimbabwe. India (LTR = 1,070 persons day -1 )

and Iran (LTR = 957 persons day -1 ) were the highest in Asia & Oceania while New Zealand (LTR = 2 persons day -1 ) was the lowest (Fig. 4b ). In Europe, United Kingdom had the highest LTR (2,365 persons day -1 ) closely followed by Spain (2,095 persons day -1 ) while Norway (74 persons day -1 ) and Bulgaria (22 persons day -1 ) were the lowest (Fig. 4c ). The USA in North America, Brazil in South America and Russia in the Arctic region had the highest LTR of 23943, 5280 and 3280 persons day -1 respectively. The results revealed that North American countries had the highest LTR of 26,334 persons day -1 . Next was Europe (8,991 persons day -1 ), closely followed by South America (9,108 persons day -1 ), Asia (4,692 persons day -1 ), and the Arctic Region (3,421 persons day -1 ) while Africa came last with LTR of 1,069 persons day -1 respectively (See supplementary Fig. 1 ).

The estimated global monthly climatology of surface air temperature ( o C), relative humidity (%) and total precipitation ( were extremely cold (< -30 o C) during this period. Furthermore, most parts of the word were very humid with relative humidity greater than 50% (Fig. 4b) . However, some islands in the Pacific, India, and Atlantic oceans, as well as a few places in Far East Asia (e.g. Bangladesh,

China and India), had very low relative humidity between 10 and 40%. The most humid inland regions (with humidity ≥ 80%) were found in the tropics and northern hemisphere. In addition, wet regions (with records of precipitation ≥ 10 mm) were found in the tropical parts of western (e.g. Nigeria, Senegal, Ghana) and southern Africa ( 

Scattered plot of COVID-19 cases in relation to selected weather parameters indicate non-normal distribution and erratic linear patterns between new daily COVID-19 cases and the parameters (Supplementary Fig. 2) . The logarithm-transformation of the dependent variable and removal of outliers, nonetheless, produced acceptable normal distribution and collinearity among the variables required for regression analysis as done in this study.

Consequently, the lagged correlation coefficient at 0 to 7-days period was evaluated to ascertain the level of weather-COVID-19 daily occurrences (Tables 1-6 ). The coefficients reveal strong (positive and negative) relationships between the weather parameters and In addition, some significant varying time-delay effects were obtained between climatic variables and COVID-19 infections. For instance, time-lag between change in temperature and maximum occurrence of infection was 1-day in Canada (North America; Table 4 ) and Argentina (South Africa; Table 6 ). In addition, the results revealed 2-day lagging in Ghana and South Africa (Africa); Poland (Europe); Iran and Qatar (Asia); Mexico (North America); and Chile (North America). Three-day lagging was obtained in Egypt, Pakistan and Peru Madagascar recorded 4-day lagging. Yemen and Kyrgyzstan had 5-day lagging to temperature effects while Bangladesh and the USA had 6-day time lag. Similar wobbling results were obtained for relative humidity and precipitation with time-lags ranging between 1 to 7 days in 16 and 10 countries respectively.

Projected percentage changes in COVID-19 cases associated with a unit decrease in temperature ( o C), relative humidity (%) and precipitation (mm) are illustrated in Fig. 6 .

Generally, our results suggested that one-degree Celsius drop in temperature will alter 

The present study revealed that the total number of reported cases and death had reached 4,753,211 and 303,531 respectively by the end of the study period (May 28, 2020) in all the selected 61 countries scattered over the globe. In order of severity of the disease, the persons day -1 ) -Europe (8,991 persons day -1 ) -South America (9,108 persons day -1 ) -Asia (4,692 persons day -1 ) -Arctic region (3,421 persons day -1 ) -Africa (1,069persons day -1 ).

Results of spatial distribution of the infections suggested that spread of the disease was highest in the northern hemisphere (i.e. high latitudes, with temperate and/or continental climate) and the prominent direction of the transmission was mainly from the epicentre to Furthermore, there were strong evidences of 1 to 7-day delays in the response of the infection to changes in climate. This is an indication that the effects of climate variations may take 1 to 7 days to manifest in an infected individuals or it may be a manifestation of the incubation period of COVID-19 infections, which is 2-14 days. Previous studies have reported that delayed response to diseases, which may however vary with victim's age, body characteristics and underlying health condition may adversely affect the turnout of victims in seeking medical treatment and that the condition may be dangerous in countries with poor testing rate 

We Table 1 Lagged correlation coefficients at 0 to 7-days between the daily reported infected cases and temperature (TMP in o C), relative humidity (RHU in %) and precipitation (PRE in mm) over selected countries in Africa (* significant at p ≤ 0.05; bold value is an indicator of lagging). Table 5 Lagged correlation coefficients at 0 to 7-days between the daily reported infected cases and temperature (TMP in o C), relative humidity (RHU in %) and precipitation (PRE in mm) over selected countries in the Arctic region (* significant at p ≤ 0.05; bold value is an indicator of lagging). Table 6 Lagged correlation coefficients at 0 to 7-days between the daily reported infected cases and temperature (TMP in o C), relative humidity (RHU in %) and precipitation (PRE in mm) over selected countries in the South America (* significant at p ≤ 0.05; bold value is an indicator of lagging). 

Spread of SARS-CoV-2 Coronavirus likely to be constrained by climate

A brief review of socio-economic and environmental impact of Covid-19

COVID-19, extractive industries, and indigenous communities in Canada: notes towards a political economy research agenda

The psychological impact of quarantine and how to reduce it: rapid review of the evidence

A spatio-temporal analysis for exploring the effect of temperature on COVID-19 early evolution in Spain

A familial cluster of pneumonia associated with the 2019 novel coronavirus indicating person-to-person transmission: a study of a family cluster

COVID-19 in India: Are biological and environmental factors helping to stem the incidence and severity?

Lockdown for CoViD-2019 in Milan: What are the effects on air quality?

Can atmospheric pollution be considered a cofactor in extremely high level of SARS-CoV-2 lethality in northern Italy?

ERA5: Fifth generation of ECMWF atmospheric reanalyses of the global climate

Geospatial analysis: A comprehensive guide to principles techniques and software tools, 6 th edition

Standardizing effect size from linear regression models with log-transformed variables for meta-analysisBMC

COVID-2019: update on epidemiology, disease spread and management. Monaldi Arch

Temperature, humidity and latitude analysis to predict potential spread and seasonality for COVID-19

COVID-19 pandemic, oil prices, stock market, geopolitical risk and policy uncertainty nexus in the U.S. economy: Fresh evidence from the wavelet-based approach

Ethnic inequalities in time to diagnosis of cancer: a systematic review

High temperature and high humidity reduce the transmission of COVID-19

Severe air pollution events not avoided by reduced anthropogenic activities during COVID-19 outbreak

2020. Novel coronavirus

Biostatical Analysis, second ed

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

None