Impact of City-Wide Diesel Generator Use on Air Quality in Quito, Ecuador, during a Nationwide Electricity Crisis

Abstract

As climate change intensifies and 2023 sets the record for the hottest year, the Amazonian region faces severe drought, leading to a hydroelectricity crisis. This study examines the effects of using environmentally harmful backup power solutions, which are all too common in developing countries—specifically, diesel-powered generators—on urban air quality in Quito, Ecuador, during the hydroelectric crisis of 2023/2024. The impact of the blackouts on air pollution is assessed by comparing it to a reference period before the crisis and to an earlier year with similar weather conditions. The findings indicate that extended blackouts (up to 8 h per day) considerably increased concentrations of SO₂ (180%), CO (43%), NO₂ (39%), and PM_2.5 (20%) compared to ‘normal’ conditions. Conversely, O₃ pollution levels decreased (−6%). Our findings suggest potential respiratory and cardiovascular health risks for the urban population, highlighting the urgent need for improved energy infrastructure and cleaner backup power solutions in the developing world. Addressing these challenges is critical for ensuring a healthier and more sustainable urban future amidst climate change projections.

1. Introduction

When the World Meteorological Organization (WMO) announces 2023 as the hottest year on record [1], an extraordinary drought in the Amazon River basin adds to the extensive list of negative effects of anthropogenic climate change [2]. The severity of this drought has been heightened by the El Niño phenomenon, which typically reduces rainfall and increases temperatures in the region. This has led to record low levels in the Amazon River and its basin, disrupting transportation, communities, wildlife, and crucially, hydroelectric power production. In countries like Brazil (80% dependency), Colombia (79%), Venezuela (68%), Bolivia (32%), and Peru and Ecuador (55%), hydroelectricity is a key power source, leading to power shortages starting in June 2023 and stretching well into April 2024 [2,3].

Human activities, responsible for climate change, also contribute to toxic global air pollution, posing risks to both the environment and public health. According to the World Health Organization (WHO), air pollution causes about 7 million premature deaths each year, with 88% of those deaths occurring in low- and middle-income countries due to weaker air quality regulations [4,5,6,7]. Toxic air pollution primarily originates from the combustion of fossil fuels in industries, transportation, and energy production [8]. The primary global source of air pollution is energy and heat production [9], underscoring the global urgency to shift towards renewable energy sources like hydroelectricity. However, hydroelectric power depends on a reliable water supply, which is increasingly threatened by changing climatic conditions. Furthermore, climate change is expected to increase the frequency and intensity of extreme weather events such as droughts, heavy rainfall, floods, cyclones, and wildfires [10]. To meet the ever-increasing energy demand regardless of climate conditions, in the event of grid failure, power generation is often replaced by the use of fossil fuel-powered generators [11].

Various studies, particularly those from developing countries, have documented the negative impacts of generator use on air quality. This is mainly due to outdated technologies and low-quality diesel fuel. For example, Iraq’s unreliable national grid has led to elevated levels of pollutants, including volatile organic compounds (VOCs), sulfur dioxide (SO₂), hydrogen sulfide, nitrogen dioxide (NO₂), and particulate matter (PM) at the University of Technology in Baghdad due to autonomous power generation [12]. Similarly, the extensive use of diesel generators in Nigeria, which provide electricity to 90% of businesses and 30% of homes, has been associated with severe air pollution [13,14,15]. In Nepal, the use of supplementary generators during the dry season also adversely affects air quality [16]. Predictive models for urban centers in the U.S. estimate that generator use could decrease ozone (O₃) levels while increasing primary and secondary PM_2.5 (PM with aerodynamic diameter ≤ 2.5 µm) concentrations [17,18]. Additionally, a switch from oil to gas power generation results in significant air quality improvements [19]. Similar modeling studies have raised concerns about emission accumulation in street canyons, particularly when exhaust is released near tall buildings, even with newer technology generators [20]. In contrast to the circumstances in developed countries, approximately 1.5 billion people globally experience thousands of hours of power outages each year [21], leading to serious health risks due to multiple toxic diesel-generated pollutants [14,22,23,24,25].

Although research has been conducted in other parts of the world, to the best of our knowledge, no studies have examined the impact of electricity crises on air quality in Latin America. This study aims to fill that gap by analyzing the impact of the fall of 2023 and spring of 2024 energy crises on air quality in high-elevation Quito, Ecuador, focusing on five key pollutants, namely carbon monoxide (CO), NO₂, SO₂, O₃, and PM_2.5. The impact is quantified by comparing air quality during the crises to a reference ‘normal’ period, either a few months before the event or to diurnal trends from previous years with similar weather conditions.

2. Methodology

2.1. Study Site

Ecuador is among the least urbanized nations within South America, the world’s second most urbanized continent (UN, 2019) (Figure 1a). Quito, the capital city, is located on a narrow plateau on the Pichincha volcano (elev. 4800 m above sea level, m.a.s.l.) in the Andes mountain range (Figure 1b). Quito is the highest constitutional capital in the world, with an average elevation of 2815 m.a.s.l. [26]. The metropolitan area is rapidly growing, now spanning 372.39 km² and housing over 2,872,351 residents [27,28]. Due to the use of low-quality fuels, the city struggles with chronic air pollution [5]. Quito experiences a mild, spring-like climate with two main seasons, wet (September–May) and dry (June–August) [29].

The Secretariat of Environment at the city mayor’s office has managed an air quality monitoring network since 2004 to track atmospheric pollutants and meteorological conditions. The network consists of nine stations located in different parts of the city, representing areas with varying elevations, population densities, traffic levels, and industrial activities (Figure 1b). This study analyzed data from all nine sites, which were (1) San Antonio (elev. 2428 m.a.s.l., 78°26′53.3″ W, 0°0′31.5″ S); (2) Carapungo (elev. 2660 m.a.s.l, coord. 78°26′50″ W, 0°5′54″ S); (3) Cotocollao (elev. 2739 m.a.s.l, coord. 78°29′50″ W, 0°6′28″ S); (4) Tumbaco (elev. 2355 m.a.s.l., coord. 78°24′12.4164″ W, 0°12′53.334″ S); (5) Belisario (elev. 2835 m.a.s.l, coord. 78°29′24″ W, 0°10′48″ S); (6) Centro (elev. 2820 m.a.s.l, coord. 78°30′36″ W, 0°13′12″ S); (7) Camal (elev. 2840 m.a.s.l, coord. 78°30′36″ W, 0°15′00″ S); (8) Chillos (elev. 2453 m.a.s.l, coord. 78°27′36″ W, 0°18′00″ S); and (9) Guamani (elev. 3066 m.a.s.l., coord. 78°33′5″ W, 0°19′51″ S).

2.2. Pollution Data Collection and Analysis

Concentrations of CO, NO₂, SO₂, O₃, and PM_2.5 were measured hourly using U.S. Environmental Protection Agency (EPA) standard equipment, adhering to EPA guidelines [30]. CO levels were measured using the Thermo Fisher Scientific 48i (Waltham, MA, USA) based on infrared absorption (EPA No. RFCA-0981-054). NO₂ was monitored with the Thermo Fisher 42i, utilizing NO_x analyzers and the chemiluminescence method (EPA No. RFNA-1289-074). The Thermo Fisher Scientific 43i was employed to measure SO₂, using ultraviolet fluorescence (EPA No. EQSA-0486-060). O₃ was detected with the Thermo Fisher 49i, employing the ultraviolet absorption method (EPA No. EQOA-0880-047). For PM_2.5, Thermo Scientific continuous monitors FH62C14-DHS 5014i were used based on the beta-ray attenuation method (EPA No. EQPM-0609-183).

Meteorological data were acquired using complete weather stations with automatic sensors at each study site. Wind speed was measured using MetOne instruments. Relative humidity, temperature, and precipitation were measured using Thies Clima equipment. Solar radiation was measured using Kipp & Zonen radiometers (OTT HydroMet B.V., Delft, The Netherlands).

Hourly data for criteria pollutants (CO, NO₂, SO₂, O₃, and PM_2.5) and meteorological conditions (wind speed (WS); temperature (T); solar radiation (SR); precipitation (Rain); and relative humidity (RH)) were collected from 1 January 2014 to 30 April 2024. For the daily time series analysis, data from 1 April to 26 October 2023 were considered to represent business-as-usual or ‘normal’ conditions, meaning periods without power outages. To investigate daily pollution behavior during the two electricity crises—the fall 2023 (27 October–18 December 2023) and spring 2024 (16–30 April 2024) crises—averages of air quality and meteorological variables during these periods were compared to those under ‘normal’ conditions. For the analysis of diurnal trends, hourly data were averaged over the same dates as the electricity crises for each year from 2014 to 2023. A paired t-test analysis was carried out to compare the atmospheric (meteorology and air pollution) differences between a crisis period and the ten years preceding the blackout, in order to conclude if the difference is statistically significant or not. The t-tests was performed on 24 h diurnal trends in meteorological and air quality conditions. After completing the ten pairwise comparisons, a single year presenting no significant meteorological differences was selected for assessing the true impact of the crisis on the concentration of each air pollutant. The selected year is qualified as the ‘normal’ period in the manuscript.

The percent changes in air pollutant concentrations were calculated using Equation (1), where $\overline{b}$ and $\overline{n}$ represent the mean concentration of each pollutant during the blackout and ‘normal’ periods, respectively.

$$\% change={\displaystyle \frac{\left(\overline{b}-\overline{n}\right) 100}{\overline{n}}}$$

(1)

Given the highly variable blackout schedule across different districts of Quito, we focused on understanding the overall impact on the city’s air quality by analyzing data averaged across all nine monitoring sites. The fall 2023 crisis involved blackouts of 2–4 h on average, while the spring 2024 crisis saw blackouts extending to 4–8 h. During the fall of 2023, no district experienced power cuts during nighttime hours (18:00–6:00) or weekends for security reasons, but this rule was lifted in the spring of 2024 due to a more extreme deficit and the severe drought in the Ecuadorian Amazon region. This is another reason why the analysis was conducted on city-wide averages rather than by individual sites, weekdays, or daytime.

The analyses, figure creation, and percent change calculations were performed using Microsoft Excel software version 16 (Microsoft Corporation, Redmond, WA, USA).

3. Results and Discussion

3.1. Daily Average Time Series Analysis

Figure 2 presents the time series—daily averages and standard deviations—of all studied atmospheric variables (Figure 2a–e) from 1 April 2023 to 30 April 2024. To facilitate the observation of the changes in air quality, Figure 2f–j were added to represent the average values of the studied atmospheric pollutants for each period of interest, namely (i) ‘Normal’ conditions (i.e., 1 April–26 October 2023; gray columns); (ii) fall 2023 electrical crisis (i.e., 27 October–18 December 2023; orange columns); and (iii) spring 2024 electrical crisis (i.e., 16 April–30 April 2024; green columns). According to Figure 2 and the summarized statistics in

Table 1. The percent change from ‘normal’ conditions (1 April–26 October 2023) for two blackout crises (fall 2023: 27 October–18 December 2023; and spring 2024: 16–30 April, 2024) for criteria pollutants (i.e., NO₂, SO₂, CO, O₃, and PM_2.5) and meteorological variables (i.e., average daily cumulative precipitation (Rain), average daily temperature (T), average daily solar radiation (SR), average daily wind speed (WS), and average daily relative humidity (RH)).

% change from ‘normal’	CO	NO2	SO2	O3	PM2.5
Blackout Fall 2023	1.11	12.16	83.08	−9.77	−6.36
Blackout Spring 2024	92.68	34.58	160.83	−28.66	12.98
% change from ‘normal’	RH	T	WS	SR	Rain
Blackout Fall 2023	10.73	0.47	−9.58	−4.95	45.17
Blackout Spring 2024	20.73	−3.9	−31.42	−24.3	182.77

, the more extreme power cuts, and thus longer generator use, during the spring 2024 crisis greatly affected air quality in Quito.

A marked increase in CO concentrations from 0.5 to 1 mg m⁻³ was observed during the spring 2024 electrical crisis, when blackout periods lasted 2–4 times longer (up to 8 h/day) than the fall 2023 crisis (Figure 2a,f). The latter showed little change from ‘normal’ conditions (Figure 2f). CO concentrations increased by up to 93% during the spring 2024 crisis compared to ‘normal’ conditions (Table 1). A small (7.6%) increase in CO concentrations was reported in other studies investigating the impact of generator use on air pollution [13]. This increase might be due to generator emissions, but it can also be further affected by the reduction in wind speed during this period (Table 1), which reduces urban pollution ventilation (R = −0.36, p < 0.05,

Table 2. Correlation matrix (with significance analysis p < 0.05 marked by *) for criteria pollutants’ (i.e., NO₂, SO₂, CO, O₃, and PM_2.5) hourly concentrations and meteorological variables (i.e., cumulative precipitation (Rain), temperature (T), solar radiation (SR), wind speed (WS), and relative humidity (RH)) averaged over 9 sites in Quito for the study period (1 January 2014–30 April 2024).

	NO2	SO2	CO	O3	PM2.5	Rain	T	SR	WS	RH
NO2	1 *
SO2	0.41 *	1 *
CO	0.67 *	0.35 *	1 *
O3	−0.36 *	0.02	−0.43 *	1 *
PM2.5	0.30 *	0.31 *	0.36 *	0.12 *	1 *
Rain	0.06 *	−0.05 *	0.01	0	−0.04 *	1 *
T	−0.27 *	0.10 *	−0.24 *	0.83 *	0.18 *	−0.10 *	1 *
SR	−0.22 *	0.23 *	−0.15 *	0.74 *	0.22 *	−0.12 *	0.82 *	1 *
WS	−0.40 *	−0.12 *	−0.36 *	0.70 *	0	−0.02 *	0.70 *	0.47 *	1 *
RH	0.26 *	−0.08 *	0.26 *	−0.78 *	−0.11 *	0.19 *	−0.84 *	−0.74 *	−0.72 *	1 *

). Additionally, other meteorological factors could be considered, such as increased relative humidity (due to an increase in combustion) (R = 0.26, p < 0.05, Table 2). Temperature and solar radiation both show a significant but very weak negative correlation with CO concentrations. In this high-elevation tropical city, the following three factors are interrelated: more sun leads to more heat, more convective mixing, and thus higher wind speed (for all significant and strong positive correlations, see Table 2).

A slight increase in NO₂ concentrations can be observed, with a more pronounced increase during the more extreme spring 2024 electricity crisis (+34.6%) (Table 1, Figure 2b,g). Similar to CO, NO₂ significantly correlates with several meteorological variables. Increased NO₂ concentrations are weakly associated with low wind speed (R = −0.40, p < 0.05), solar radiation (R = −0.22, p < 0.05), and temperature (R = −0.27, p < 0.05), as well as higher relative humidity (R = 0.26, p < 0.05) (Table 2). It is important to emphasize that the 24 h average concentrations of NO₂ exceeded the WHO recommendations (25 µg m⁻³) in the Ecuadorian capital during the spring 2024 electricity crisis, indicating potential negative effects on respiratory health. A similar study in Iraq linked diesel generator use to increased concentrations of NO_x [12].

As one of the main sources of SO₂ is the internal combustion of fossil fuels, especially those with high sulfur content, it is no surprise that there were substantial increases in levels of this pollutant during both periods of electricity crisis (Figure 2c,h). This is because electrical power generators use poor-quality diesel, as is common in most Latin American countries [5]. It is also evident that the standard deviation of this pollutant is large, suggesting high variability among the sites or dates (e.g., some days with blackouts and some without, a common trend during the fall 2023 crisis, Figure 2c, orange shaded area). Table 1 shows that SO₂ concentrations increased by 83% during the fall 2023 crisis and by 161% during the spring 2024 electrical crisis. This is a major increase, implying possible health effects, even if the levels remain well below the healthy limits (40 µg m⁻³) (Figure 2c,h). Like CO and NO₂, wind appears to negatively very weakly affect SO₂ concentrations (R = −0.12, p < 0.05, Table 2). Solar radiation (R = 0.23, p < 0.05) and temperature (R = 0.10, p < 0.05) very weakly but significantly positively correlate with this pollutant.

Ozone concentrations, as observed in previous studies [31], show a decrease of 10% to 29% during the fall 2023 and spring 2024 electricity crises, respectively (Figure 2d,i, Table 1). This can be explained by increased NO_x emissions enhancing photochemical and fast reaction times of NO with O₃, leading to lower ozone levels [32]. Figure 2d shows that around the fall equinox (i.e., 23 September 2023), when the sun is directly over the equator (Quito), high solar radiation and low NO_x levels result in the highest ozone concentrations during the study period. The correlation between solar radiation and O₃ concentrations is confirmed in Table 2 (R = 0.74, p < 0.05). Interestingly, O₃ also shows a strong positive correlation with temperature (R = 0.83, p < 0.05) and wind speed (R = 0.70, p < 0.05), suggesting that elevated temperatures enhance ozone formation through accelerated photochemical reactions and that higher winds may help transport ozone precursors (i.e., natural VOCs) from nearby rural and natural areas surrounding this tropical city (Table 2) [33].

Finally, concentrations of PM_2.5 showed a slight reduction during the fall 2023 electrical crisis (−6.4%), while they increased during the spring 2024 crisis (13%) (Figure 2e,j, Table 1). The reduction in fine particle concentrations during the first crisis could be attributed to the overall decrease in human movement to avoid traffic congestion (e.g., non-functioning traffic lights) or because some employers opted to keep their businesses closed unless they had a backup generator. However, for many small businesses, which account for about 50% of the workforce [34], affording a generator is often not feasible. During the spring of 2024, 8 h blackouts pushed some businesses to shut down. Table 2 shows that PM_2.5 concentrations very weakly anticorrelate with relative humidity (R = −0.11, p < 0.05) and correlate with solar radiation (R = 0.22, p < 0.05) and temperature (R = 0.18, p < 0.05), suggesting the impact of wet removal and photochemical secondary PM formation factors on this pollutant. This pollutant is problematic in this high-elevation city (i.e., 70% oxygen availability compared to sea level), frequently nearing the violation of national annual standards (15 µg m⁻³) and unfortunately always exceeding the WHO’s stricter health standard (5 µg m⁻³) for annual averages [5]. In this study, it can be seen that the short-term 24 h WHO standard (15 µg m⁻³) is often violated, implying health effects (Figure 2e).

3.2. Diurnal Trend Analysis

For the fall crisis, the t-test analysis results show no significant difference in weather conditions between 2018 and 2023 (

Table 3. p-values for the t-test comparison between 2018 and 2023.

Factors	WS	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
p-value	0.08	0.47	0.58	0.07	0.0003	4.1 × 10−6	4.2 × 10−17	5.8 × 10−5	8.4 × 10−13

and

Table A1. Statistical significance test (t-test) of 24 h diurnal trends for all meteorological conditions (wind speed (WS); temperature (T); solar radiation (SR); precipitation (Rain); and relative humidity (RH)) and air pollutants (NO₂, SO₂, CO, O₃, and PM_2.5) during the fall 2023 (i.e., 27 October–18 December 2023) electricity crisis against the same dates (27 October–18 December) of 2014–2022, where values with a p < 0.05 significance are shaded in yellow. The dates with the most similar weather conditions to those of the 2023 fall hydroelectricity crisis (27 October–18 December 2023) are shaded in blue (27 October–18 December 2018).

Significance 2016 vs. 2023 (t-test):
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
0.73271411	0.121224242	0.4034471	0.2229383	0.000150323	1.25301 × 10−8	0.019513412	2.11457 × 10−16	0.004798785	7.40495 × 10−15
Significance 2017 vs. 2023 (t-test):
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
0.901658963	9.2875 × 10−6	0.6216069	0.0116656	0.058501619	0.009352845	1.26583 × 10−5	2.08887 × 10−18	0.963067392	1.55299 × 10−12
Significance 2018 vs. 2023 (t-test):
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
0.081686061	1.05939 × 10−6	0.4727719	0.5809547	0.068301132	0.000265125	4.11366 × 10−6	4.16475 × 10−17	5.83605 × 10−5	8.48342 × 10−13
Significance 2019 vs. 2023 (t-test):
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
0.065993722	5.06822 × 10−12	0.9667778	0.0674505	2.50945 × 10−10	1.3087 × 10−5	1.54764 × 10−7	1.45916 × 10−12	0.288449622	5.48409 × 10−9
Significance 2020 vs. 2023 (t-test):
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
0.179093131	9.19675 × 10−18	0.131328	0.8188844	8.81362 × 10−7	7.5871 × 10−5	0.00068069	2.75758 × 10−11	0.000131646	1.76364 × 10−18
Significance 2021 vs. 2023 (t-test):
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
0.000457447	2.42922 × 10−13	0.0033442	0.044246	3.46931 × 10−12	4.39527 × 10−5	1.75428 × 10−7	9.34735 × 10−9	0.095808079	0.008426095
Significance 2022 vs. 2023 (t-test):
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
0.384330499	1.40348 × 10−15	0.0216292	0.0003671	0.037371245	0.01213938	0.000196215	5.5538 × 10−17	4.36336 × 10−6	3.63129 × 10−23

, Appendix A). The p-value for the meteorological comparison is consistently above 0.05, indicating that 2018 can be used as a valid reference for comparing air quality to 2023. This latter comparison reveals a substantial difference in terms of urban pollution, further supporting the negative impact of this second energy crisis on air quality.

Figure 3 shows the diurnal trends in pollutant concentrations averaged over 27 October–18 December in both 2018 and 2023, corresponding to the fall 2023 electricity crisis. The 24 h trends for 2018 represent baseline conditions with weather patterns very similar to those during the electricity crisis, allowing for a more accurate estimation of the percent change from ‘normal’ to crisis conditions. The data showed that the largest estimated increase was in SO₂ concentrations (+47.8%, Figure 3b), followed by O₃ (+14%, Figure 3d). In contrast, the other three pollutants showed a decrease in concentrations; PM_2.5 and CO decreased by 29.9% (Figure 3e) and 28.7% (Figure 3c), respectively, while NO₂ saw a reduction of 8.8% (Figure 3a).

Although these results may seem surprising, they can be partially explained by the reduced activity of smaller businesses, many of which may not have opened during the crisis, as discussed earlier (Section 3.1). The increase in SO₂ concentrations is likely due to the heightened use of high-sulfur diesel in backup generators, which were more frequently used during the crisis. On the other hand, the overall reduction in other combustion-related pollutants such as PM_2.5, CO, and NO₂ could be attributed to decreased human activity and vehicular traffic, resulting in lower emissions. The increase in O₃ concentrations is particularly interesting and can be attributed to a phenomenon known as the ‘titration effect’. Normally, NO reacts with O₃ in the atmosphere, resulting in the formation of NO₂. During both crises, there was an increase in NO₂ (Figure 2g) and a corresponding reduction in O₃ (Figure 2i) concentrations. In contrast, during the COVID-19 pandemic, reduced traffic and industrial activity led to lower NO₂ levels and consequently higher ozone levels in some regions [35].

For the spring crisis, the t-test analysis results show no significant difference in meteorological conditions between 2021 and 2024 (

Table 4. p-values for the t-test comparison between 2024 and 2021.

Factors	WS	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
p-value	0.06	0.18	0.1	0.3	1.2 × 10−14	2.8 × 10−11	2.6 × 10−8	0.0028	0.0007

and

Table A2. Statistical significance test (t-test) for all meteorological conditions (wind speed (WS); temperature (T); solar radiation (SR); precipitation (Rain); and relative humidity (RH)) and air pollutants (NO₂, SO₂, CO, O₃, and PM_2.5) during the spring 2024 (i.e., 16–30 April 2024) electricity crisis against the same dates (16–30 April) of 2014-2023, where values with a p < 0.05 significance are shaded in yellow. The dates with the most similar weather conditions to those of the 2024 spring hydroelectricity crisis (16–30 April 2024) are shaded in blue (16–30 April 2021).

Scheme 2018. vs. 2024 (t-test):.
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
0.001973114	0.782767444	0.402305	0.004434197	3.52757 × 10−5	7.78859 × 10−11	1.64874 × 10−9	2.29182 × 10−14	8.54807 × 10−8	0.7107841
Significance 2019 vs. 2024 (t-test):
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
0.000354938	0.886634182	0.792312	0.007665521	0.000418526	2.84269 × 10−12	5.88657 × 10−11	9.41587 × 10−11	7.98113 × 10−6	0.107326151
Significance 2020 vs. 2024 (t-test):
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
0.001353019	0.000158636	0.019122	0.086738617	2.76089 × 10−6	2.8494 × 10−15	1.58462 × 10−13	1.8069 × 10−14	9.55938 × 10−11	3.3887 × 10−8
Significance 2021 vs. 2024 (t-test):
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
0.05728857	7.49586 × 10−16	0.183563	0.095205486	0.29661775	1.24425 × 10−14	2.82848 × 10−11	2.5951 × 10−8	0.002842942	0.000749829
Significance 2022 vs. 2024 (t-test):
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
6.92666 × 10−7	1.57496 × 10−11	0.57811	0.176987379	1.68197 × 10−5	1.30335 × 10−13	4.97226 × 10−10	2.77932 × 10−15	2.17824 × 10−7	4.6702 × 10−9
Significance 2023 vs. 2024 (t-test):
WS	T	SR	Rain	RH	NO2	SO2	CO	O3	PM2.5
0.009869577	1.16509 × 10−10	0.805852	0.69507086	0.003325807	9.77833 × 10−14	5.53481 × 10−12	6.25711 × 10−18	0.098811781	4.02886 × 10−11

, Appendix A). The p-value for the weather comparison is consistently above 0.05. It means that 2021 can be used as a valid reference for comparing air quality to 2024. This latter comparison reveals a substantial difference in atmospheric pollution, supporting the hypothesis of the negative impact of the crisis on air quality.

Figure 4 presents the diurnal trends in pollutant concentrations averaged over 16–30 April 2024, during the spring 2024 electricity crisis, compared to the ‘normal’ conditions of the same period in 2021. The year 2021 serves as a reference due to its similar meteorological conditions, allowing for a robust estimation of the percent changes in pollutant levels during the crisis. Diurnal trend analysis indicates an increase in all air pollutant concentrations except ozone, which decreased by 5.7% (Figure 4d). Similar results were obtained from the daily average period analysis (Section 3.1). The largest change in both analyses was observed for SO₂ (+180%, Figure 4b), followed by CO (+43.3%, Figure 4c), NO₂ (+38.8%, Figure 4a), and PM_2.5 (+20%, Figure 4e).

The substantial spike in SO₂ concentrations can be attributed to the widespread use of high-sulfur diesel in backup generators, which were operated for up to eight hours a day during this crisis compared to the shorter two-hour outages in the fall of 2023 that resulted in a 48% increase. The rise in CO (+43.3%) and NO₂ (+38.8%) further underscores the impact of prolonged generator use, as these pollutants are closely associated with combustion processes in diesel engines. The increase in CO, a byproduct of incomplete combustion, suggests that many of these generators may not be functioning optimally, leading to inefficient fuel combustion and higher emissions. The 20% increase in PM_2.5 concentrations likely reflects the combined effect of increased emissions from both generators and additional vehicular traffic due to non-functional traffic lights. Conversely, the 5.7% decrease in O₃ concentrations can be explained by the higher NO₂ levels. NO₂ reacts with O₃, leading to its breakdown and resulting in lower ambient ozone levels during the crisis. Normally, NO₂ breaks down ozone in a reaction that reduces ambient O₃ levels. The higher NO₂ concentrations during the crisis could have led to more significant ozone titration, thereby slightly lowering O₃ levels. This effect is somewhat opposite to what was observed during the Fall 2023 crisis, where a reduction in NO₂ led to an increase in O₃. The differing behaviors in O₃ concentrations between the two crises highlight the complex interplay between NO₂ and O₃ in urban air quality.

Overall, the data suggest that the extended use of backup generators during the spring 2024 crisis significantly exacerbated air pollution in Quito. As climate change intensifies, many regions will experience more severe weather, leading to societal disruptions, human and material losses, and economic damage. This study, by examining urban air quality deterioration during climate-induced hydroelectricity crises, emphasizes the urgent need for improved climate adaptation planning to ensure a sustainable energy future. This type of extreme, but not unusual (El Niño), climate- and weather-related event must be considered in future scenarios. To address the energy shortfall studied here, high-sulfur diesel generators had to be widely used in urban areas to meet the power demands. This situation, all too common in the developing world, must be addressed in national mitigation plans and tackled at the international level, as it is a consequence of rising greenhouse gas concentrations. It is imperative to resolve this issue, given the recorded increase of up to 180% in certain toxic pollutants in densely populated urban areas, which poses significant future respiratory and cardiovascular health risks to the population.

4. Conclusions

This study offers a detailed examination of air quality variations in Quito, Ecuador, during the fall 2023 and spring 2024 hydroelectricity crises, both influenced by climate change and intensified by the 2023/24 El Niño phenomenon. The analysis reveals that extended power outages significantly worsened air quality in this high-elevation tropical city. During the more extreme spring 2024 crisis, where blackouts extended up to 8 h daily, CO levels nearly doubled compared to ‘normal’ conditions, primarily due to the widespread reliance on diesel generators. This surge in generator use also led to a 38.8% increase in NO₂ concentrations, raising concerns about respiratory health risks, particularly given the exceedance of the World Health Organization (WHO) guidelines. The study further identifies a dramatic 180% rise in SO₂ levels during the spring 2024 crisis, highlighting the detrimental impact of using high-sulfur diesel in generators. This sharp increase underscores the localized air quality challenges posed by poor-quality fuels during prolonged power outages. Additionally, PM_2.5 concentrations rose by 20% during the spring 2024 crisis, reflecting the combined effect of generator emissions and additional vehicular traffic, as blackouts turn off traffic lights. These sustained elevated PM_2.5 levels, already above the WHO recommendations, present significant long-term health risks. While the concentrations of most of the pollutants increased, O₃ concentrations decreased by 5.7% during the spring of 2024, likely due to the higher NO₂ levels leading to ozone titration. This finding contrasts with the fall 2023 crisis, where a reduced NO₂ resulted in higher O₃ levels, demonstrating the complex interplay of urban chemistry. The study underscores the pressing need for cities to prepare for more frequent and severe climate-related disruptions to electricity supplies, which may exacerbate urban air quality issues. To mitigate these impacts, it is crucial to develop and implement comprehensive strategies, including improved energy infrastructure and cleaner backup power solutions. Addressing these challenges is vital for ensuring a healthier and more sustainable urban future, particularly in high-elevation areas like Quito.