Increased emphasis on local action for global change in the sustainability discourse encourages efforts of sub-national authorities to prioritize sustainability (Theobald et al., 2015; Wamsler et al., 2014; Xue and Tobias, 2015, among others). International initiatives to promote local sustainability still tend to focus on local production-based resource uses and emissions (Covenant of Mayors, 2017). However, trade-related environmental impacts are now increasingly analysed through e.g. concepts of water footprint and life-cycle, and tracking of consumption-based (as opposed to more common production-based) greenhouse gas (GHG) emissions (Mekonnen and Hoekstra, 2010; Gerbens-Leenes et al., 2012; Tukker, 2000; Stokes and Horvath, 2010; Caro et al., 2017; Yang et al., 2015). Moreover, in the wake of the adoption of the UN Sustainable Development Goals (SDGs) in 2015, the need for integrated and cross-scale coordination of sustainability actions is increasingly acknowledged (Nilsson et al., 2016, among others).

Two SDGs with particularly strong interactions are those focused on water (SDG 6) and energy (SDG 7). Both resources are crucial for human survival and prosperity, and both need to be managed more sustainably in the coming decades in order for the world to meet the related SDGs (Griggs et al., 2017). Direct (at the energy utility) and indirect (throughout the whole energy supply chain) uses of water for electricity production have been shown to be significant, not only on local but also on, regional continental and global scale (Destouni et al., 2013; Jaramillo and Destouni, 2015). The water-energy nexus has also been increasingly studied in the past decade (among many others: Hamiche et al., 2016; Hussey and Pittock, 2012; Bazilian et al., 2011; Scott et al., 2011). The ability to accurately assess both direct-local and indirect-remote interactions of water and energy uses will only increase in importance as local actors move towards implementing the SDGs. The present study addresses this water-energy nexus on different scales.

Specifically, we here quantify and analyse energy-related indirect-remote water use, i.e. the water-use required for extracting, processing, transforming and supplying energy for local use, using the 21 counties of Sweden as concrete case examples of such local energy use; Fig. 1 displays the 21 counties on the map of Sweden. In general, Sweden is considered to be the most advanced country towards meeting the SDGs (Sachs et al., 2016). As such, the present case study focus on and across different Swedish counties may provide some important insights on relatively advanced water-impact patterns associated with local energy choices towards long-term sustainability. The analysis is further broadened, to also consider the climate-related SGD (13), by relating the water-energy nexus results to the officially reported GHG emissions of each county.

The used data and obtained results of this study (direct water use, GHG emissions, and calculated energy-related indirect water use) are not overlapping in scope (the latter includes trade impacts, while the former two relate only to local activities). This scope inconsistency is intentional, in order to clarify the difference between data that are readily available for – and directly relate to the mandate of – county-level decision-makers (reported local direct water use and GHG emissions) and the indirect energy-related water impacts (indirect-remote water use for energy supply) that are calculated in this study and are significant for achievement of global sustainability.

In the following sections, the analytical approach and data sources are described, followed by presentation and discussion of results and main conclusions.

Figures 3 and 4 display the calculated indirect water-use requirements for the energy supply in the 21 counties of Sweden, in absolute and per-capita terms, respectively. The size of circles in these graphs corresponds to the relative volume of energy-related indirect freshwater use, with the x axis representing reported county emissions of GHGs and the y axis representing county-specific direct local water-use.

In terms of absolute values (Fig. 3), the most populous counties (Stockholm, Skåne and Västra Götaland) have the largest reported GHG emissions, as well as most of the largest reported direct water uses. However, also the county of Norrbotten (in terms of emissions) and that of Västernorrland (in terms of direct water use) display high absolute resource uses despite having much smaller populations. Västernorrland has the most water-intensive industry in Sweden, while Norrbotten is the county using the most coal of all Swedish counties in 2010, with thereby high associated GHG emission levels. These latter two counties stand out even more clearly in Fig. 4, where results are displayed in terms of per-capita units. In addition, the county with the smallest population, Gotland, stands out as the county with the highest per capita GHG emissions, due to significant fossil GHG emissions from the county's lime and cement industries (Region Gotland, 2016). However, the direct local water use in Gotland is limited, with this being a Swedish county that is particularly prone to water shortages and droughts.

Looking further at the results of indirect water use for energy supply (circle sizes in Figs. 3 and 4), the pattern differs from both that of reported GHG emissions and that of reported direct water use. Most notable here are the counties of Värmland and Gävleborg, with significantly larger energy-related indirect use of water than that of other Swedish counties. The reason is that the industries in Värmland and Gävleborg use large volumes of renewable liquid fuels (assumed to correspond to biodiesel or bio-oil in Table 1). As shown in Table 1, these fuels consume 70–90 times more water per unit of energy than corresponding liquid fossil fuels (and this is a modest water-use estimate compared to those of Gerbens-Leenes et al., 2009; IEA, 2012; among others).

Gotland's indirect per-capita use of water does not follow the pattern of its per-capita emissions of GHGs. As mentioned, this county has a relatively high share of fossil GHG emissions but uses almost no liquid biofuels. This choice makes the county's energy use relatively water efficient.

In Fig. 5, the national (Swedish) indirect water use (for fuel, electricity and heat production within Sweden) is removed from the results. This exacerbates the pattern seen in Fig. 4, exhibiting Gävleborg and Värmland as responsible for an even larger share of Sweden's indirect (imported) energy-related water use (Fig. 5). This is not surprising given the large water requirements of imported liquid biofuels (in the present study, no domestic liquid biofuels are considered available). It also emphasizes the large discrepancy that may prevail between the direct water use in local service-delivery systems and the non-local and indirect water use required for the local energy supply.

If direct water use were a good indicator of total water use (including indirect water from energy imports), the size of the circles in Figs. 3–5 should generally increase along the y axis, which is not the case. Similarly, and even though both water use and GHG emissions are tightly linked to the energy sector, the levels of GHG emissions are not correlated with the remote and indirect water use for the local energy supply (as circle sizes do not systematically increase (or decrease) along the x axis).

For further analysis of the joint water-use and GHG connections to energy, Fig. 6 displays all discussed water-use and emission results, as well as the total energy use in each county. Värmland and Västernorrland are then found to have similar profiles of per-capita energy use and per-capita GHG emissions (boxed part of Fig. 6). However, their indirect per-capita water uses for energy are far apart. In Värmland, the chosen low-carbon energy use consists to large degree of liquid biofuels, while the per-capita energy use of Västernorrland includes primarily electricity and biomass – two fuels that consume much less water than imported liquid biofuels. These different choices explain some of the discrepancy seen in Fig. 6 for indirect water-use per capita. However, in this context, it is again worth noting that the used data on energy-related GHG emissions and the calculated data on indirect water use have different scopes of measure. The former (GHG data) include local emission sources, including burning of fuels in vehicles and in electricity- and heat-production utilities, etc., locally as well as local non-combustion/energy related emissions from agriculture, waste facilities and similar. The latter (data on water use for energy supply) consider only the energy sector and the energy use in each county, including the full energy supply chain, both local and non-local. Although we can largely explain the patterns of GHG emissions and of indirect water use by investigating the specific mix of fuels used in a county, there is no unambiguous pattern of e.g. low (high) emissions corresponding to high (low) indirect water use. With that said, if Värmland's industries had chosen to use fossil fuels instead of biofuels, one would expect higher GHG emission levels in the county and lower indirect water use associated with the county's energy supply.

Finally, looking at the combination of direct water use and GHG emissions reported, Figs. 4 and 5 show that Blekinge has similar x/y-coordinates (per-capita direct water use and emissions, respectively) as Värmland. However, their indirect water uses for energy supply (circle size) are very different. As such, neither direct water use nor GHG emissions, or the combination of these, provides any clear clue to the indirect water use of a county. Specific case-by-case analysis of the considered energy fuels – and ideally a more focused analysis of the specific origin of these fuels – is needed to assess this type of externalized water impacts with greater certainty. Nevertheless, the present analysis supports existing work on water-energy resource interdependencies (Hussey and Pittock, 2012, among others) and the suggested importance of SDG interactions by Nilsson et al. (2016). The studied external (remote) water impacts of local energy strategies are significant, but difficult to trace with great precision and certainty from only local-level resource-use and emissions data.

As noted in the data section, limitations in both availability and comparability of data on water use in energy supply chains are significant. Although careful selection of water factors has been made for this analysis (Table 1), the uncertainties and temporal and spatial variations of reported energy-related water uses are large. As a consequence, the results presented here should be interpreted and used conservatively. This work primarily aims to highlight the un-intuitive (sometimes counter-intuitive) relations between direct and indirect environmental impacts of local energy choices, and the potentially large differences in indirect resource impacts (in our study focused on water) of regions with seemingly comparable local emissions profiles and direct resource uses.

Focusing on the water-energy nexus, this paper has identified quite different patterns of reported direct water use, GHG emissions and indirect water use for local energy use among Sweden's 21 counties. Direct water use is typically driven by population size and the presence of water-intensive industries. In contrast, energy-related indirect water use is primarily impacted by the choice of fuels used in the county – with liquid biofuels requiring orders of magnitude larger volumes of water than all other fuels used in the analysed energy systems.

Similar to the indirect water use studied in this paper, GHG emissions are tightly coupled to the energy sector. However, on a fuel-by-fuel basis, the water-use and GHG interactions with energy differ significantly. Conventional fossil fuels, with their high CO2 emissions, are orders of magnitude less water-intensive than the carbon-neutral biofuels.

As climate-change action plans are developed and adopted by an increasing number of sub-national authorities (to meet the Paris Agreement as well as the climate-related SDG 13), the goal conflict implied by a largely (even though not fully) inverse relation between GHG emissions and indirect water use should be acknowledged and further investigated. As no clear general relation could be determined in the present analysis of energy-related GHG emissions, indirect-remote water use and direct-local water use, further investigation of such relations are needed for different local-regional cases. Such further analysis is not least needed in Sweden, if the country is to remain a front-runner in reaching the highly integrated SDGs, with the freshwater SGD (6), the energy SDG (7) and the climate SDG (13) considered as equally important sustainability targets.

Finally, accounting for this type of cross-scale interactions in the monitoring of progress towards reaching the SDGs could potentially alter assessments of how different nation states or sub-national regions are performing, and should be considered when indicators for monitoring progress towards the SDGs are further developed.