Environmental Variables

Tropospheric NO ₂ Column (CTNO ₂ )¹

CTNO₂ data were obtained from the tropospheric monitoring instrument (TROPOMI) on board the Sentinel-5 Precursor satellite (^{Eskes et al., 2019}). Typical long-term exposure to NO2 was defined as a 3.5-month period, that is, from January 1 to March 14, 2020, one day before the lockdown period in Peru was declared (^{Diario Oficial El Peruano, 2020}). Data for the period March 16 to April 20 were also analyzed to identify changes in NO₂ concentration. Average NO₂ concentrations for each period were calculated with the Google Earth Engine platform. The resulting raster datasets were exported and processed in a geographic information system (QGIS) to calculate the average per pixel for each region of Peru.

Vertical Air Flow - Omega (FVA)

The FVA parameter is analyzed in the troposphere; we used omega data produced by NCEP/NCAR Reanalysis 1 (National Centers for Environmental Prediction/National Center for Atmospheric Research) and downloaded from the web platform of the Physical Sciences Laboratory of the National Oceanic and Atmospheric Administration/ Earth System Research Laboratories (NOAA/ ESRL) (https://psl.noaa.gov/, last visited April 30, 2020). The data analyzed were the omega monthly means for 2019 at 850 hPa (~1500 m elevation). The working hypothesis was that regions where negative omega values prevail would be favored with better air circulation and dispersion of pollutants, which would prevent their stagnation on the surface by carrying them to higher elevations. In contrast, atmospheric pollutants will have a longer permanence in regions where omega values are either positive or close to 0 (atmospheric stability).

Source: Compiled by the authors.

Figure 1 Environmental and social variables included in the AHP to evaluate the risk of spread of SARS-CoV-2 in the regions of Peru. 

Population with No Access to Basic Sanitation Services (MEE)

Data used were the number of inhabitants per region with no access to sanitation services through the public sewerage network. These data were obtained from ^{INEI (2016)}.

Percentage of Solid Waste Disposed of in Open Dumps (RSB)

This variable denotes the percentage of solid waste that is disposed of in sites lacking proper sanitary measures (open dumps). Data on the percentage of solid waste disposed of in open dumps in each region of Peru were obtained from ^{Orihuela Paredes (2018)}.

Social Variables

Vulnerable Population (VP)

Data on the size of vulnerable population were obtained from chapter 1 of the Perfil Sociodemográfico del Perú, 2017 (Sociodemographic Profile of Peru, 2017) compiled by the national statistical agency, the Instituto Nacional de Estadística e Informática (National Institute of Statistics and Informatics, INEI). The vulnerable population amounts to 38.29% of the total population of Peru and includes children under 14 years of age and adults aged 60 years or older. These segments of the population are deemed vulnerable because they fall into economically inactive age classes supported by economically active age groups (^{INEI, 2017}). In addition, according to ^{CDC-MINSA (2020)}, the +60-year age class is related to an increased mortality risk from COVID-19.

Number of Hospitals (NHP)

Peru has 19,859 healthcare facilities (^{INEI, 2018}) spread across the 25 regions. Data on the number of hospitals, the main type of healthcare facility receiving people infected with COVID-19, were used for this study. These data were obtained from INEI (http://m.inei.gob.pe/estadisticas/indice-tematico/health-sector-establishments/, last accessed on May 17, 2020). The data were expressed as number of hospitals per 100,000 inhabitants based on the population size recorded in the latest population census.

Monetary Poverty (PM)

PM data were provided by ^{INEI (2019)}. These data classify the regions into five classes that denote the percentage of the population that cannot afford a basic food basket or has no food.

Environmental Variables

Figure 2 shows the distribution maps of tropospheric NO₂ before and after the lockdown period in the regions. Figure 3 shows the map of vertical air flow-omega (Pa/s), and Figure 4a shows the number of positive COVID-19 cases, CTNO₂, and FVA. The largest number of positive COVID-19 cases were found in Lima (20,048 cases), Callao (2,933), Lambayeque (1,814), Piura (960), and Loreto (881). Some of these regions also showed high mean tropospheric NO₂ concentrations prior to the lockdown period: Callao (41 μmol/m²), Lima (16 μmol/m²), and Lambayeque (10 μmol/ m²). In addition, vertical air flow in these regions was close to 0.

Figure 2 Changes in the concentration of tropospheric NO2 in the regions of Peru. Panels a1 and a2 show NO2 concentration before the lockdown period in Peru and the Lima regions, respectively. Panels b1 and b2 show NO2 concentrations after the lockdown period for Peru and Lima, respectively. 

Figure 3 Vertical air flow - omega (Pa/s) at 850 mb. Panel a shows vertical air flow for a large part of South America. Enlargements in panels b, c, d, and e show vertical air flow characteristics in the regions of Peru. Values close to 0 Pa/s can be observed in b, c, and d. 

*Number of persons infected with COVID-19 in the Lima region is shown divided by 10.

Figure 4 Behavior of environmental variables in relation to the cumulative number of positive COVID-19 cases at the regional level in Peru. a) behavior of CTNO2 and FVA in relation to the number of people infected with COVID-19; b) behavior of MEE and RDB in relation to the number of people infected with COVID-19. 

Figure 4b shows that Loreto and Ucayali were the regions where MEE accounted for over 40% of the total population; these were also among the ten regions with the largest numbers of positive COVID-19 cases. In contrast, more than 98% of the inhabitants of Callao, Lima, and Tacna has access to basic sanitation services. However, the Callao and Lima regions had the largest number of people affected by COVID-19.

Figure 4b shows that RSB is over 65% in the regions with the largest number of positive COVID-19 cases (Lima, Callao, Lambayeque, Piura, Loreto, among others). In general, more than 45% of solid waste is disposed of in open dumps in almost all regions, except for Callao, where it is 0%, as no such spaces exist therein (^{MINAM, 2020c}).

Social Variables

Figure 5a shows that, on average, NHP ranges between 1 and 2 in all regions, even in the five regions with the largest number of positive COVID-19 cases. PV was lower in the five regions with the largest cumulative number of positive COVID-19 cases. PV ranged between 30% and 45% in all regions.

*Number of people infected with COVID-19 in the Lima region is shown divided by 10.

Figure 5 Behavior of social variables in relation to the cumulative number of COVID-19 cases at the regional level in Peru, including the constitutional province of Callao. a) Behavior of NHP and PV in relation to the number of COVID-19 cases; b) behavior of PM in relation to the number of COVID-19 cases. 

Figure 5b shows that Lima, Callao, La Libertad, Piura, and Loreto are the regions with the largest number of positive COVID-19 cases and these fall in classes 2 to 4 of PM (on a 5-class scale, where 1 are the regions with lowest PM, and 5 the regions with highest PM).

Analytical Hierarchy Process

Susceptibility to SARS-CoV-2

Table 2 shows the relative weights estimated from the pairwise comparison matrix. CTNO₂ and FVA are the most important variables, with relative weights of 0.435 and 0.407, respectively, followed by MEE with 0.106 and, finally, RSB with 0.052. The consistency ratio index (CR) was 0.032, indicating a fair degree of con sistency in the weight assigned to the variables analyzed.

The four driving environmental variables were included into an index of susceptibility (IS_SC2) of the regions to the SAR-CoV-2 virus; the index was constructed as the weighted sum of the four variables, as shown in Equation 1.

Vulnerability to SARS-CoV-2

Table 3 shows the relative weights assigned to the vulnerability variables through the AHP. NHP was the most important variable, with a relative weight of 0.584, followed by PV with 0.297 and PM with 0.118.

The three variables analyzed were integrated into an index of vulnerability (IV_SC2) to the SARCoV-2. The index was constructed as the weighted sum of the three variables, as shown in Equation 2.

The product of the susceptibility and vulnerability values yields values representing the risk level to the SAR-CoV-2 virus (^{CENEPRED, 2015}). Figure 6 shows regional-level maps of susceptibility, vulnerability, and risk to the SAR-CoV-2 virus; values in these maps were classified into four levels (low, medium, high, and very high) for easier interpretation (^{Pourghasemi, Pradhan and Gokceoglu, 2012})landslide locations were identified by aerial photographs and field surveys, and a total of 78 landslides were mapped from various sources. Then, the landslide inventory was randomly split into a training dataset 70 % (55 landslides.

Figure 6 Maps showing levels of a) susceptibility, b) vulnerability, and c) risk to the spread of the SARS-CoV-2 virus. 

Lambayeque, Callao, Tumbes, and Lima showed a very high susceptibility to the spread of SARS-CoV-2. In addition, La Libertad, Piura, Loreto, Ancash, Cajamarca, Amazonas, Ica, San Martín, Huancavelica, and Pasco showed a high susceptibility. Most of these regions are located in the northern part and the central coast of Peru. The rest of the regions exhibit a medium or low (Cusco region) susceptibility. All the regions with very high susceptibility are located on the coast of Peru.

Huancavelica, Huánuco, Ucayali, Amazonas, and Loreto showed a very high vulnerability to the spread of SARS-CoV-2. Four other regions located in southern Peru (Moquegua, Apurímac, Cusco, and Arequipa), two in central Peru (Pasco and Callao), and four located to the north (Tumbes, Piura, Cajamarca, and Ancash) exhibited a high level of vulnerability. All the other regions showed medium to low vulnerability.

The Callao and Tumbes regions showed a very high risk level. Piura, Loreto, Lambayeque, Huancavelica, Amazonas, Cajamarca, Ucayali, Huánuco, Ancash, Moquegua, Pasco, Ayacucho, San Martín, La Libertad, and Apurímac were under a high risk. All other regions exhibited a medium risk level.

Environmental Variables

CTNO₂ maps for the regions of Peru showed notable reductions in NO₂ levels from pre- to post- lockdown periods. The magnitude of these reductions as up to three times higher in regions with high population density, intense industrial activity, and large motor car fleet such as Arequipa, Callao, Lima, and Piura. The highest tropospheric CTNO₂ values were recorded in coastal regions (north, center, and parts of the south), some of them in Lima, Callao, Lambayeque, and Piura, regions where the largest numbers of positive COVID-19 cases were recorded.

The lowest CTNO₂ values were found in the Amazonas region - one of the ten regions with fewest confirmed cases of COVID-19. This suggests a likely direct relationship between CTNO₂ and the number of confirmed COVID-19 cases, as shown by ^{Ogen (2020)}. In addition, ^{Wu et al. (2020)} mentioned that long-term exposure to poor air quality might exacerbate the COVID-19 symptoms, and even the mortality risk from this disease.

The highest FVA value, <-0.02 Pa/s (close to 0) was found in Loreto, one of the five regions with the largest numbers of positive COVID-19 cases. In contrast, the lowest FVA value (> -0.16 Pa/s) was observed in Puno, one of the five regions with fewest positive COVID-19 cases. This suggests a likely direct relationship between the regions with more stable FVA (which favors the buildup of pollutants) and the number of positive COVID-19 cases.

In summary, we found that high CTNO₂ values and FVA values very close to 0 Pa/s prevailed in the regions with the largest numbers of positive COVID-19 cases. That is, conditions of atmospheric stability prevail in those regions, which boost the buildup of atmospheric pollutants, including NO₂, near the ground. ^{Wang et al. (2020)} mentioned that environmental conditions may be linked to the rate of spread of SARS-CoV-2 and the severity of the disease.

Regions such as Loreto and Ucayali showed MEE values above 40%. According to ^{SUNASS (2015)}, these two regions lack wastewater treatment plants with disinfection treatment. The regions on the Peruvian coast, where the largest numbers of positive COVID-19 cases have been observed, show MEE <15%. These findings do not support a relationship (direct or inverse) between MEE and the number of positive COVID-19 cases. Nevertheless, the lack of access to basic sanitation services could give rise to public health issues related to COVID-19 contagion, as the virus has been reported to be present in feces from infected people (^{Ahmed et al., 2020}; ^{Medema et al., 2020}; ^{Rosa et al., 2020}; ^{Wu et al., 2020}; ^{Wurtzer et al., 2020}; ^{Zhang et al., 2020}).

The Loreto and Ucayali regions showed RSB values over 40%. Although no direct relationship was found between this factor and the number of positive COVID-19 cases at the regional level, special precautions should be taken in the management of solid waste from people infected with COVID-19 as this is a potential transmission route (^{Association of Cities and Regions for Sustainable Resource Management, 2020}; ^{SWANA, 2020}). ^{Van Doremalen et al. (2020)} pointed out that the time the virus remains in stainless steel, plastic, cardboard, and copper waste materials, is possibly related to the number of positive COVID-19 cases, since the virus can remain active in these types of solid wastes for several hours. Therefore, the solid waste recycling measure adopted by the Peruvian government (through decree D.S. N° 080-2020-PCM) may increase the number of cases of COVID-19 infections in workers involved in solid waste management.

Analytical Hierarchy Process

This study used the AHP to combine environmental and social variables. We found a high or very high risk of spread of SARS-CoV-2 mostly in northern and central Peru, as well as in some southern regions. This results from the high vulnerability and susceptibility levels found in those regions.

The results from our model might be limited by the relatively few variables analyzed; nevertheless, this is the first study that examines environmental and social variables for Peru and showing, in relatively simple terms, results that are consistent with those reported by ^{CDC-MINSA (2020)}. The latter study found that 60% of the regions of Peru should strengthen their care and control of COVID-19 cases; our study found that 68% of the regions are at high or very high risk of spread of SARS-CoV-2. In addition to the areas identified by ^{CDC-MINSA (2020)}, our results suggest that the San Martín and Ayacucho regions should also be included among the priority areas. The study by ^{Yaser Burhum (2020)} showed that, as of 16 May 2020, all regions of Peru had an R index greater than 1, indicating that SARS-CoV-2 was still in the propagation stage. This could be related to the influence of environmental and social variables such as those analyzed in our study - and others not yet analyzed- on the virus propagation.