Temperature and rainfall records were retrieved from the National Oceanic and Atmospheric Administration and the Turkish Met Office [9,10].

Istanbul water consumption and reservoir level data were obtained from the Istanbul Municipality [11–13].

Data in Türkiye is published by the Turkish Statistical Institute each year [8]. We used data for the years for which we had information on water use.

Climate projection data (temperature, rainfall and warm spell duration index) is presented on the World Bank’s Climate Change Knowledge Portal (CCKP), including the Coupled Model Intercomparison Project (CMIP) Phase 5 and Phase 6 collections, which serve as a repository of model runs with consistent emission scenarios to standardise some of the output generated by diverse modelling groups. As the spatial scale of a global climate model’s output is too large to characterise climate over small and specific areas, downscaling to a finer resolution is necessary. Thus, statistically downscaled climate variables over the studied region were downloaded from the World Bank’s CCKP [14]. We chose RCP 2.6 as a low emission scenario (one where the Paris Agreement was met), RCP 4.5 as an intermediate emission scenario, RCP 6.0 as a stabilising emission scenario and the higher RCP 8.5 scenario if the Paris Agreement was not successful at reducing global emissions. We selected four different periods to examine (2020–2039, 2040–2059, 2060–2079 and 2080–2099). In order to look at the specific location of Istanbul, the outcomes of CIMP5/6 ensemble model for RCP 2.6, RCP 4.5, RCP6.0 and RCP 8.5 were extracted [14]. We used the Portal as it is one that policymakers and government agencies can get ready access to and might use for planning adaptations to climate change.

As part of our international comparisons (see below) we used the National Hydrological Monitoring Programme’s Monthly Hydrological Summaries, including water reservoir levels across the UK [15].

Additional analysis of air quality and traffic movements was used to support the widely held contention that the Covid-19 pandemic would lead to changes human behaviour in Istanbul (and other major cities). All these data sources/sets can be freely accessed from the link provided in the references [16] at https://doi.org/10.5522/04/19122179.

We used information gathered from the news and social media as well as information issued by local government bodies or published in authoritative reports (e.g., from water utilities and consultancies) as well as material in the academic literature. We used this range of sources to try to capture the fast-changing circumstances of the pandemic. To increase the likelihood that our findings might be generalised, we drew on information from other countries, including the UK, the United States, Germany, Italy and Brazil.

Figure 4 shows that water use in Istanbul has varied month by month from 2000 to 2022. Water consumption in Istanbul has nearly doubled. It has gradually increased since the 2000s (as shown by the shift from green to blue and darker blue) from an average of 1.6 million m³/day to 3.1 million m³/day. Water use was consistently higher in the summer. The overall increase in water use may be attributed to population increase, industrialisation and rising standards of living and/or lifestyle choices (such as leisure uses) that go alongside increasing consumption levels in general. Additionally, Fig. 7 shows that over the 21 years the population increased by about 40%, while water consumption rose by 70%.

Although population and water consumption (Fig. 8) both increase over time (and therefore show a strong Pearson correlation of r = 0.98, n = 20, P < 0.001), there have been periods when water consumption grew very little, for example, between 2005 and 2007–2009. This can be linked to low rainfall and drought periods in 2005 and 2007 [19], with people and businesses adjusting their behaviour to reduce consumption following communications from governmental bodies and the media about water shortages.

As has been recorded for many major cities (see below in the section on international comparisons), there were marked impacts on human activity in the early stages of the pandemic due to lockdown restrictions.

In order to obtain some separate evidence (other than that on water resources and consumption) that the pandemic restrictions affected human activity in Istanbul, we examined air quality parameters, and these suggest that air pollution in Istanbul was lower in 2020 than in 2019, indicating decreased human activity consistent with Covid-19 restrictions. Air quality parameters such as particulate matter (PM)10, sulfur dioxide (SO₂) and carbon monoxide (CO) provide good evidence of these changes [12]. Annex–I in [16] shows how the levels of these parameters decreased in Istanbul. After confirmation of the first cases in Türkiye in March, air quality improved as pollutant levels fell in late March when national restrictions (working remotely, closing restaurants and full lockdowns) were announced and applied. This indicates less commercial activity overall.

In addition to air quality parameters, Fig. 9 shows how monthly sensor-based numbers of vehicles at Kadikoy, Istanbul Metro, have changed since 2016. The station (near the Optimum Shopping Centre) is located on one of the main connections from Asia to Europe through the Eurasia Tunnel. As the road is actively and intensively used during the day, vehicle movements could be an indicator of how Covid-19 restrictions affected transportation in Istanbul. There was a sharp decrease in April and May 2020, during the national Covid-19 restrictions (working remotely, closing restaurants and full lockdowns).

Data underpinning the visualisation in Fig. 4 show that in March 2020, when Covid-19 started to spread in Türkiye, water consumption was higher than in March 2019 (from 2.73 million to 2.85 million m³/day). This increase of about 120,000 m³/day (or about 3%) could reflect additional water use due to the increased emphasis on hygiene measures such as hand washing or street disinfection (although it is uncertain if this later increase is included within the figures). This might appear to be a slight increase. However, in March and April, when the Turkish government decided to move to a lockdown in big cities, including Istanbul, substantial numbers of people decided not to stay in a highly populated city and went to their hometowns or summer houses. The movement of people seemed substantial such that the Mayor of Bodrum, a coastal town in the Muğla area, stated ‘Please do not come to Bodrum; Bodrum is full’. Many vehicles entered Bodrum that were registered in Istanbul or Ankara [20]. The first Covid-19 case in Bodrum was confirmed in a person coming from Istanbul before the start of travel restrictions among the big cities in Türkiye [21]. These temporary internal migrations probably explain why consumption in Istanbul fell in April by an average of 42,210 m³/day when traffic movements were also at a minimum.

From June, when Covid-19 cases started to decrease in Türkiye as a whole, some people began returning to Istanbul, and water consumption gradually increased by 141,609 m³/day to the end of the summer in August. Continuing public health measures such as an increased emphasis on hand washing and the sanitisation of public spaces with water-based disinfectants (Fig. 10) could all be expected to lead to increased water consumption in comparison to use in recent years.

Expressing the water use data in terms of consumption per head per day, as in Fig. 11, creates further insights, especially when the first Covid-19 pandemic year of 2020 is compared to the same data for the three years preceding the pandemic (data from earlier years was not used to avoid complications linked to the rising water use rates in Istanbul).

Water use in 2020 (blue line) was higher than the average of 2017–2019 (black dashed line) for several months between March and October (an extra 2.56 L per person per day was consumed across the year as a whole, assuming a constant population for Istanbul although we know this was not the case due to the scale of temporary migrations). A similar effect was found in other countries as well (see Discussion). In contrast, during the period people left Istanbul temporarily (April and May 2020), water use was lower than the 2017–2019 average – for instance, in April, it was 2.73 L/person per day lower than in March. As people gradually returned to Istanbul from the second quarter of 2020, there was a steady increase in water consumption, with consumption in July–October higher than the 2017–2019 average. In the pandemic year of 2020, along with rising summer temperatures, which generally cause additional water usage, water consumption continued to climb above the three-year average right through until September, when, usually, consumption would have fallen back. This suggests that the Covid-19 pandemic placed an additional exertion on Istanbul’s water resources (about 2% greater across the year as a whole assuming a constant population). Perhaps the most striking finding was that in 2020 water consumption per head in Istanbul did not decline in July, August and September as it had in earlier years. The difference in this period between 2020 consumption and the three-year average was about 10 L/person per day, equivalent to approximately 150,000,000 L/day (or 150,000 m³/per day). An additional complexity of interpretation is that in the last three years, religious holidays (for Islam) were in August and September. This might have an impact on water consumption as some people went to their hometown during the religious holiday.

Yilmaz et al. [7] provide information on the water resources of Istanbul, which are largely reservoir-based. Periodically, reservoir levels are low, suggesting that Istanbul is close to experiencing drought conditions should rainfall be lower than usual. Monthly water reservoir levels (as % of live capacity) from 2005 are given in Fig. 5. In most years, reservoirs are at their highest levels in late-winter and spring months and lowest in the autumn because of the increased consumption during the summer and lower rainfall together lead to resource depletion. Reservoir levels were at their lowest level in this record in 2007, 2008 and 2014. These years are shown clearly in the visualisation of Fig. 5. In the pandemic year of 2020, the highest reservoir level – a nevertheless relatively low 69% was observed in April, when human activity was at its lowest and public health measures were at their maximum extent. Normally, an increase in reservoir levels is expected with increased rainfall through the end of the year. However, the decreases in reservoir capacity that started in the spring months of 2011, 2013, 2019 and 2020 saw limited recovery (Fig. 12). In the last two years (2019–2020), capacity did not recover throughout the year, so even without the impact of Covid-19, there was a case to be made for taking a precautionary approach and encouraging water saving. There are indications in this limited dataset that, effectively, near-drought conditions impacting Istanbul may occur every 6–7 years. Within the data for our limited time period: a severe drought occurred in 2007/8, and not dissimilar conditions occurred in 2014/15, and late in 2019 and 2020 and perhaps into 2021.

Given the evidence above (in Figs 5,12–14) that Istanbul may be subject to rising temperatures and variable rainfall and thus drought-like conditions relatively frequently, we examined four climate change emission scenarios (RCP 2.6, RCP 4.5, RCP 6.0 and RCP 8.5) at four different time periods to better understand whether conditions would be right in the future for such events to become more frequent or more intense. Projected temperature and rainfall patterns compared to the reference period (1986–2005) have been placed in the University College London (UCL) Research Data Repository [16].

Overall, the climate projections given in Table 1 suggest that temperatures will rise in Istanbul, but rainfall will not increase. Rainfall may decline most under those scenarios where temperatures move to potentially very high values (the mid- or even high-40s°C). With current pressures and limited water resources, these locally expected trends in temperatures and rainfall may have significant impacts on water availability over the upcoming decades especially if there is lower rainfall at times of the year when reservoirs are currently replenished. Given the long-term implications for increased water use in Istanbul as a result of higher temperatures and given that the population might increase further, dealing with the public health aspects of a future pandemic might indeed become even more challenging, especially if it occurred in a drought year. Much may depend on how long warm and/or dry spells last in the future. It is appreciated that the argument here is complex and depends on a range of social technological and economic issues, but it would seem wisest to plan for more water use to cope with a pandemic-like public health emergency.

Because it appeared that temperatures in Istanbul would rise substantially in the future under some scenarios we thought that the Warm Spell Duration Index (WSDI), defined as the count of days with at least six consecutive days when daily Tmax > 90th percentile of the reference period (1986–2005) [26], would be a useful indicator of periods when water use might be high or conditions were favourable to drought. Figure 15 shows the historical Türkiye WSDI for 1986–2005 and the projected WSDI by 2060 under four emission scenarios. Even in a low emission scenario (RCP 2.6), warmer periods seem likely to last about nine times longer than in the reference period (1986–2005). This will likely place more pressure on water resources, as people under high temperature conditions tend to use more water for recreational and other activities.

Limited comparisons are possible between Istanbul and other areas at present due to a lack of comparative data for the pandemic period. This is due to natural–environmental variables, such as varying climatic patterns, as well as variations in governance, management and monitoring of water resources and systems in the areas being compared. These, in turn, might arise as the result of culturally (or politically) determined preferences, the initially adopted control paradigms and the associated strategies, and in addition a high level of power distance and uncertainty avoidance [27].

In all the examples where data was collected during the pandemic’s early phases, a combination of factors contributed to some changes in daily domestic patterns of water consumption. In these other studies, we looked particularly for factors that needed to be included in the modified DPSIR framework and for confirmation that factors included in the initial DPSIR framework (Fig. 3) were indeed of some importance. Table 2 summarises the main findings on how various countries have been affected by different phases of the pandemic and makes comments related to the final DPSIR framework (see final sections).

We could only access reservoir data for the UK from readily available data sources. Despite different weather patterns, there is evidence for a decline in reservoir-based water resources in both the UK (Fig. 6 from the London, UK area as an exemplar) and Türkiye (Fig. 4) during the period of the pandemic. Not all of this is necessarily due to it, but perhaps to a combination of lack of ability to replenish reservoirs and additional demands for supply.

Figure 6 shows how Thames Water reservoir levels (%) in London have changed since 1989. As with Istanbul, there are indications in this dataset that the lowest reservoir levels in the London area may occur every 6–7 years. The effect was most pronounced in 1990, with not dissimilar data for 1996/97, 2003, 2011 and 2017/18 as well. However, one difference between the Istanbul and London areas is that London reservoir capacities rarely reach the low levels seen in those that serve Istanbul, perhaps because London has the Thames catchment to call on as a freshwater resource.

As evidenced in Fig. 10, a lot of water has been used for disinfecting public spaces. Many kinds of cleaning, disinfecting and bleaching agents were probably used during the pandemic, some of which may have been used in quantities and amounts that were not expected when the national and international risk assessments were done for these materials. When released into the environment, in this case through surface water drains or directly onto soil near the areas disinfected, bleach and other chemicals can release chlorine that reacts with organic matter in the soil, water and air to form a range of compounds, some of which might be classed as halogenated compounds subject to restriction. Other materials, such as detergents, can have toxic effects on aquatic systems if they enter these systems at high enough concentrations. Thus, it is possible that such practices could have environmental impacts that have not yet been quantified as these compounds could be toxic to wildlife, carcinogenic or mutagenic, accumulate in the food chain and eventually impact humans. Not all water used for disinfection passes through a treatment facility before entering the natural environment. Thus, the products used for hygiene purposes during Covid-19 times may end up in rivers and in the sea.

Additionally, the disinfection products may also infiltrate the soil and impact land, plants and animals. As the purification process used by most water treatment plants is achieved by bacterial action, the introduction of chemicals in high concentrations (such as bleaches or disinfectants) could significantly impact treatment plants. More evidence is needed on these emergent issues linked to public health measures taken in the public realm of many urban areas.

This was an issue for the UK and probably other countries as well. As people stayed home more, increased flushing of inappropriate articles through the domestic toilet systems caused blockages in the sewerage systems [56–58]. Popular news outlets carried dramatic images of the increased blockages occurring as early as April 2020. It seemed that Covid-19 lockdowns were followed by a rise in blockages [59]. Media messages to only allow permitted materials into the sewage system accompanied such images.

Increased costs of sewer blockages (costing about £3M) were only one of the economic impacts of the pandemic on Thames Water, UK. As people lost their jobs and experienced financial difficulties, they were advised to contact their water companies for help with deferring their water and sewerage payments. This led to bills going unpaid. There were also increased costs of running water infrastructure as a result of changes in the pattern and size of the demand. Thames Water posted a pre-tax loss of £246.5M in its interim report for 2020/21 [60]. Many of the factors involved in this loss involved social/economic ones (help to less well-off domestic customers; loss of revenue from business customers during the pandemic; reassessment of risks partly to account for the pandemic’s wider implications) as well as other cost issues linked to supply, sewage and infrastructure. Parts of this loss, such as the increase in bad debt, were attributed directly to the pandemic. Similarly, Southern Water [61] refers to several impacts of the Covid-19 pandemic, including a 7% rise in water use from 127 to 136 L/person per day; leakage up by about 4%; increased operating costs of £2.7M together with redistribution in income from different customer sectors; help for vulnerable customers; and a need for re-analysis of a range of business risks. The full effects of the pandemic on the water sector may not be known for some time.

In Ankara, capital of Türkiye, during lockdowns, real-time usages were not billed due to governmental statements and restrictions (to protect workers and public health) and, consequently, municipalities’ revenues were disrupted, which might be even worse in a further pandemic or under currently unexpected pressures, such as those that might arise under climate change [62].

The events of the pandemic also suggested that people may need to take a wider view of water resources and act accordingly. For example, in order to act on advice to consume less water, they may need to reuse water used while cooking for watering plants, use less water when washing and showering and, if possible, plant drought-resistant seeds in gardens and parks [63].

By disrupting economies and causing millions of deaths globally, the Covid-19 pandemic has likely had a serious impact on progress toward the Sustainable Development Goals (SDGs) and further compromised the 2030 targets. Some argue it is important to introduce further cost-effective and innovative policies for achieving those SDGs [64]. More generally, there were concerns about preventing water contamination as the virus can survive up to several days, perhaps longer in low temperature regions. Even if sewage treatment methods and approaches such as chlorination and ultraviolet (UV) irradiation have the ability to eradicate Covid-19 in the water, the possibility of contamination may be high in areas where sewage is untreated [65]. Interestingly, water systems can also be part of the early warning systems for Covid-19 (or other types of pandemics or diseases). In many parts of the world, this form of monitoring for Covid-19 has been initiated (e.g., in the Netherlands) [66].

Emergencies, such as the Covid-19 pandemic [67], can compound the pressure on water usage and availability that are already under pressure from climate change in some parts of the world.

A similarity between Istanbul and the London/South East England area is that both regions are likely to experience higher temperatures and more periods of dry weather under climate change scenarios [68–70]. For some parts of the UK, there are substantial concerns about water resource and supply issues due to climate change alone, especially after 2050. Potential supply shortages of between 1220 and 3900 million L/day are possible for the UK – that is, a supply equivalent to that for between 8.3 and 19.7 million people [71]. Actions are in hand to address the climate issues [39,40] but issues linked to the pandemic may need to feature more explicitly than they do at present. Economic growth in South East England together with rapid demographic changes, driven in part by housing-led regeneration, are adding to water quantity and quality issues with a pause on housing development in place during 2022 that will have long-term impacts on a range of social issues.

For a range of reasons reservoir levels in various countries were at low levels during the pandemic; this was certainly so in parts of Türkiye and parts of the UK as this study indicates. We have not examined aquifer data as that is not a major factor relevant to the main case study area of Istanbul. Additional consideration of the operational margin needed to maintain a viable live capacity may be necessary given pandemic experiences and upcoming climate change impacts on precipitation patterns. New reservoirs are proposed for South East England [39,40] that allow for climate change, but perhaps not a future pandemic whose control measures extend into hot weather periods when demand for water increases, especially if this occurred in a period of drought.

The substantial changes in raw water use during the pandemic and the pressure on such compound the risks to water systems presented by climate change and other socio-economic events such as the current pandemic, and suggest that much greater use of digital technologies may be necessary for use in a reimagined water infrastructure system [41].