Climate impact of Finnish air pollutants and greenhouse gases using multiple emission metrics

We present a case study where emission metric values from different studies are applied to estimate global and Arctic temperature impacts of emissions from a northern European country. This study assesses the climate impact of Finnish air pollutants and greenhouse gas emissions from 2000 to 2010, as well as future emissions until 2030. We consider both emission pulses and emission scenarios. The pollutants included are SO2, NOx , NH3, non-methane volatile organic compound (NMVOC), black carbon (BC), organic carbon (OC), CO, CO2, CH4 and N2O, and our study is the first one for Finland to include all of them in one coherent dataset. These pollutants have different atmospheric lifetimes and influence the climate differently; hence, we look at different climate metrics and time horizons. The study uses the global warming potential (GWP and GWP), the global temperature change potential (GTP) and the regional temperature change potential (RTP) with different timescales for estimating the climate impacts by species and sectors globally and in the Arctic. We compare the climate impacts of emissions occurring in winter and summer. This assessment is an example of how the climate impact of emissions from small countries and sources can be estimated, as it is challenging to use climate models to study the climate effect of national policies in a multi-pollutant situation. Our methods are applicable to other countries and regions and present a practical tool to analyze the climate impacts in multiple dimensions, such as assessing different sectors and mitigation measures. While our study focuses on short-lived climate forcers, we found that the CO2 emissions have the most significant climate impact, and the significance increases over longer time horizons. In the short term, emissions of especially CH4 and BC played an important role as well. The warming impact of BC emissions is enhanced during winter. Many metric choices are available, but our findings hold for most choices.


Introduction
The Paris Agreement and its target of "holding the increase in the global average temperature to well below 2° C above pre-25 industrial levels and pursuing efforts to limit the temperature increase to 1.5° C above pre-industrial levels" (UNFCCC, 2015) provides an important framework for individual countries to consider the climate impacts and mitigation possibilities of its emissions. Globally CO2 and greenhouse gas emissions are key components in achieving the targets of the agreement, but the role of short-lived climate forcers (SLCFs) should also be studied as additional drivers of the surface temperatures.
The climate effect of emission reductions of air pollutants, particularly black carbon and tropospheric ozone, have been a 30 focus of research in last few years (Shindell et al. 2012, Bond et al. 2013, Smith and Mizrahi, 2013, Stohl et al. 2015. Since air pollutants can either cool or warm the climate on different timescales depending on the species, emission reduction policies from a climate perspective have to be designed to take into account the net-effect of multiple pollutants (UNEP/WMO 2011, Stohl et al. 2015. The pollutants considered to have most climate relevance are termed Short-lived Climate Pollutants (SLCP) or Short-lived Climate Forcers (SLCF), depending on the context. However, there is no common 35 agreement on the definition of SLCPs or SLCFs. In this study we use the terms as in the IPCC Global Warming of 1.5°C Special Report (IPCC 2018) where: (1) SLCFs refer to both cooling and warming species and include methane (CH4), ozone (O3) and aerosols (i.e., black carbon BC, organic carbon OC and sulfate), or their precursors, as well as some halogenated species; (2) SLCPs refer only to the warming SLCFs. Policies focusing on SLCPs have been suggested as supplements to greenhouse gas reductions (UNEP/WMO 2011, Shindell et al. 2012, Rogelj et al. 2014, Stohl et al 2015, Shindell et al. 40 2017. Modelling studies by UNEP/WMO (2011) and Stohl et al. (2015) suggested that the climate response of SLCF mitigation is strongest in the Arctic region. The Arctic region is of particular interest, since in the past 50 years the Arctic has been warming twice as rapidly as the world as a whole, and has experienced significant changes in ice and snow covers as well as 45 permafrost (AMAP, 2017). AMAP (2011 and as well as Sand et al. (2016) demonstrated that emission reductions of SLCFs in the Northern areas have the largest temperature response on the Arctic climate per unit of emissions reduced, with the Nordic countries (Denmark, Finland, Iceland, Norway, and Sweden) and Russia having the largest impact when compared with the other Arctic countries, Unites States and Canada. 50 Shindell et al. (2017); Ocko et al. (2017) have argued for assessing both near-and long-term effects of climate policy.
However, comparing the climate impacts of SLCFs, CO2, and other pollutants is not straightforward. Emission metrics are one way of enabling a comparison as they provide a conversion rate between emissions of different species into a common unit, for example CO2-equivalent emissions. Common emission metrics are the Global Warming Potential (GWP) (IPCC, 1990) and the Global Temperature change Potential (GTP) (Shine et al., 2005). The GWP compares the integrated radiative 55 forcing (RF) of a pulse emission of a given species relative to the integrated RF of a pulse emission of CO2. Since the UNFCCC reporting procedure uses the GWP with a 100 yr time horizon (GWP100) as a reporting guideline, it has become the most common metric to report greenhouse gas emissions. The GTP is an alternative to GWP and it compares the temperature change at a point in time due to a pulse emission of a species relative to the temperature change of a pulse emission of CO2. The GTP combines the changes in the radiative forcing induced by the different species with the 60 temperature response of the climate system and thus has been argued to relate better with climate effects (Shine et al., 2005).
Both GWP and GTP focus on the global response, while the temperature impact can also be analyzed on a regional scale, i.e. the Arctic, applying Regional Temperature change Potential (RTP) (Shindell and Faluvegi, 2010). Even for a uniform forcing, there will be spatial patterns in the temperature response. The metrics can be presented in absolute forms of radiative forcing (AGWP) or temperature perturbation (AGTP, ARTP) as well as normalized to the response of CO2 (GTP, GWP, 65 RTP). Especially for short-lived species, the climate impact depends on the location and timing of the emissions, which is reflected in the RTPs as well as in the global response for GTP and GWP. On a global scale, Unger et al. (2009)  In this study we assess the climate impact of Finnish air pollutants (SO2, NOX, NH3, NMVOC, BC, OC and CO) and greenhouse gas emissions (CO2, CH4 and N2O) in the past (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010) and until 2030, according to a baseline emission projection. We utilize emission metric values from several new studies relevant for Finland.

75
Finnish emissions and their climate response are relatively small compared with emissions from larger regions, let alone the globe. Therefore it is challenging to use climate models to study the climate effect of national policies and to analyze the role of each pollutant and sector. This study demonstrates a method to overcome this challenge by the use of emission metrics.
The method is applicable in other countries or regions as well and has been used in connection with the Norwegian work on SLCPs (Norwegian Environment Agency, 2014) (Hodnebrog et al., 2014). 80 The Methodology section describes the construction and background data of the emission inventory and the future scenario as well as the emission metrics used. In Results we describe the emissions and their climate impacts first focusing at the historical emissions (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010) and then at a future projection until 2030. We also discuss separately the regional temperature effect of emissions on the Arctic and compare the results obtained with different metric studies. In the 85 Conclusions section we will summarize the main findings and conclude on the major scientific and policy relevant messages.
The objectives of this study were to (1) produce an integrated multi-pollutants emission dataset for Finland for 2000 to 2030, (2) compare multiple climate metrics and assess their suitability for a Northern country like Finland, (3) estimate the climate impact of Finnish air pollutants and greenhouse gases for the period 2000 to 2030 utilizing selected climate metrics, and (4) 90 suggest a set of global and regional climate metrics to be used in connection with Finnish SLCF emissions.

Methodology
Finland is one of the Nordic countries situated, between latitudes 60°N and 70°N. It has a population of 5.5 Million people with an average population density 17.9 inhabitants per square kilometer (for comparison: EU average is 117 inhabitants per square kilometer). Although much of the population is concentrated to the South of the country, the scarce population 95 compared with the size of the country makes transport of goods and people an important activity. The Northern location of the country in turn results in a high demand for energy to heat households, and the economy is largely based on energyintensive industry.

Emissions
The historical emissions of SO2, NOX, BC and OC in 2000, 2005 are estimated based on the data in the Finnish 100 Regional Emission Scenario (FRES) model (Karvosenoja, 2008). Emissions of NH3, VOC, CO2, CH4 and N2O are from the national air pollutant and greenhouse gas emission inventories as reported to the UNFCCC and the UNECE Convention on Long-Range Transboundary Air Pollution (CLRTAP). The CO emission data is estimated with the GAINS model (http://gains.iiasa.ac.at; Amann et al. 2011). The data sources by pollutant are presented in Table 1. Emissions of CO2 are presented according to the IPCC guidelines, which assume biomass as carbon neutral. However, this definition is disputed 105 and e.g., Cherubini et al. (2011) present emission metric values that account for CO2 emissions from biomass. Although the historical emission data emanates from different data sources (Table 1), they have been checked for consistency and are based essentially on the same statistical sources. We aggregated the data and performed specific analyses for the following eight major economic sectors: energy production (ENE IND), industrial processes (PROC), road transport (TRA RD), offroad transport and machinery (TRA OT), domestic combustion (DOM), waste (WST), agriculture (AGR), and other 110 (OTHER). The assumptions about the future energy use, transport and other activities in Finland follow Finland's 2013 National 115 Climate and Energy Strategy (Ministry of Employment and the Economy, 2013) and its' baseline scenario that fulfils the agreed EU targets and specific national targets for share of renewables and emission reductions in the non-ETS sector. Table   2 shows the primary energy consumption by fuel in Finland in 2010 and 2030. The 2013 National Climate and Energy Strategy assumes the future prevalence of wood heating to remain at 2011 level, which is estimated to lead to a decreased wood consumption, due to increasing energy efficiency in housing. The future emission projection was estimated with the 120 Finnish Regional Emission Scenario (FRES) model which used the activity estimates from the 2013 National Climate and Energy Strategy (Table 2) as a basis.

Emission metrics
This work studies Finnish emissions with several climate metrics and focuses particularly on three of them, the Absolute Global Warming Potential (AGWP) (IPCC, 1990), Absolute Global Temperature change Potential (AGTP) (Shine et al., 2005), and Absolute Regional Temperature change Potential (ARTP) (Shindell and Faluvegi, 2010). AGWP at time horizon H for emissions of pollutant i in emission season s from emission sector t is defined as 130 where RF is the time-varying radiative forcing given a unit mass pulse emission at time zero. Since two recent studies Aamaas et al., 2017) have separated between emission during summer (May-October) and winter (November-April), we make this separation when possible. AGTP is given as 135 ( 2) IRFT(H-t) is the temperature response, or impulse response function for temperature, at time H to a unit radiative forcing at time t. The ARTP is similar to AGTP, but gives the temperature response in latitude bands m: where Fl,i,s,t(t) is the radiative forcing in latitude band l and RCSi,l,m is matrix of unitless regional response coefficients based on the ARTP concept (Collins et al., 2013). In one of the papers (Sand et al., 2016), RCS differ for some of the different sectors, such as BC emissions in the Nordic countries from the domestic sector have about 15 percent higher sensitivity than BC emissions from energy and industry. Aamaas et al. (2017) do not provide this information on a sector 145 level, and we must therefore use the same RCS for all emission sectors.
The ARTP method divides the world into four latitude bands: southern mid-high latitudes (90-28° S), the Tropics (28° S-28° N), northern mid-latitudes (28-60° N), and the Arctic (60-90° N). We will focus on the temperature response in the Arctic, as well as the global mean response. 150 Some of the studies separate the net response for a pollutant into various processes. For the aerosols, the radiative efficiencies often include the aerosol direct and 1st indirect (cloud-albedo) effect. In addition, BC deposition on snow and semi-direct effect may also be considered for BC. The ozone precursors build on the processes short-lived ozone effect, methane effect, and methane-induced ozone effect, as well as the aerosol direct and 1st indirect effects. 155 All these emission metrics (AGWP, AGTP, ARTP) can be normalized to the corresponding effect of CO2, where M is GWP, GTP, or RTP: (4) 160 For GWP, we have included an additional analysis with the newly suggested metric GWP* (Allen et al., 2016;Allen et al., 2018). They argue for an alternative use of GWP to better compare CO2 and SLCFs, which can be done by comparing the cumulative warming of CO2 with the emission level change of SLCFs. For CO2 and N2O, we have calculated GWP*(H) based on Equations 1 and 4, which lead to CO2-equivalent emissions for pollutant iL between time t1 and t2: 165 For SLCFs, the CO2-equivalent emissions are ΔEiS is the change in emission level for SLCP iS between time t1 and t2. We have compared emissions for the 2000-2030 period and with a time horizon of H=100 years. 170 The pollutants we include in our analysis (SO2, NOX, NH3, NMVOC, BC, OC, CO, CO2, CH4, and N2O) have very different atmospheric lifetimes and impact pathways. For the GHGs (CO2, CH4, and N2O), we use the climate metric parameterization in IPCC AR5 (Myhre et al., 2013), but with an upward revision of 14% for CH4 to account for the larger radiative forcing calculated by Etminan et al. (2016). The atmospheric decay of CO2 is parameterized based on the Bern 175 Carbon Cycle Model (Joos et al., 2013) as reported in Myhre et al. (2013). We assume that the relative temperature response pattern in the four latitude bands is the same for all the GHGs, and we base our calculations on the latitude pattern for CH4 in Aamaas et al. (2017).
For all the other pollutants (SO2, NOX, NH3, NMVOC, BC, OC, and CO), we use several recent studies that are relevant for 180 the emission location, Finland Aamaas et al., 2017;Sand et al., 2016). We have examined how metric values from all those studies can be used for Finnish emissions and compared those, but we will mainly present the combinations of these studies that we think combine the strengths of the different datasets. For a general and global view, we have used the GWP and GTP values from Aamaas et al. (2016). The rest of the paper utilizes ARTP values from Aamaas et al. (2017) to estimate temperature responses, with scaling from Sand et al. (2016) for temperature responses in the Arctic. 185 Aamaas et al. (2017) is our starting point as this study has the full set of emissions and separate between summer and winter emissions.
No studies have presented climate metric values specific for Finnish emissions. The default choice would be to use climate 190 metric values based on global emissions, while we believe using smaller emission regions near or including Finland is more representative than applying the global average. The most relevant emission regions in the three selected studies are Europe (consisting of the Western Europe, Eastern members of the European Union, and Turkey, up to 66°N) for Aamaas et al. (2016); Aamaas et al. (2017) and the Nordic countries for Sand et al. (2016). The Nordic countries is a smaller region and geographically more representative for Finland than Europe. Therefore we have calculated ratios between metrics for the 195 Nordic region vs. Europe in the Sand et al. (2016) and used those ratios to scale the metric values from Aamaas et al. (2017) to better represent Finnish emissions. However, Sand et al. (2016) provided climate metric values only for the Arctic response, their set of pollutants was limited to BC, OC, and SO2, and for the ozone precursors they included only a combined response. To solve this we have used averages, such as taking a weighted average of the different emission sectors for each pollutant and assuming that NH3 can be scaled by an average of BC, OC and SO2. The scaling we have done for the 200 Arctic responses are 2.22 for BC, 3.09 for BC deposition in snow, 2.32 for OC, 1.94 for SO2, 2.16 for NH3, and 1.00 for NOx, CO, and NMVOC. This scaling for the Arctic will also increase the global responses, but not affect the coefficients for the other temperature response bands.
For all the pollutants, the IRF for temperature comes from the Hadley CM3 climate model (Boucher and Reddy, 2008). 205 Hence, our temperature calculations are based on a climate sensitivity of 3.9 K warming for a doubling in CO2 concentration.
Most emissions stay relatively constant throughout the year, while the changing seasons result in much larger emissions from the domestic sector in winter than in summer. We account for this seasonality for those metric datasets compatible with this, 210 otherwise, annual emission and metric values are applied.
The global and regional temperature responses of Finnish emissions are estimated by convolving ARTP values with emissions. For an emission scenario E(t), the global temperature response is based on AGTP values. Similarly, the temperature responses in latitude bands can be estimated by replacing AGTP with ARTP values. As noted, the ARTP method divides the world into four latitude bands, and thus the global temperature response can also be estimated by using the ARTPs and taking the area-weighted global mean basing on the results for the latitude bands. As the forcing-response coefficients are different and the ARTP concept can better parameterize varying 220 efficacies, the estimate global temperature response may vary whether based on AGTPs or ARTPs.
Our climate impact dataset can be analyzed in many different dimensions, such as for different time scales, for different emission sectors, for different processes, for pulse or scenario emissions. We show some examples. As we focus on near term climate change and the global and regional temperature, most of the discussion in this paper utilizes ARTP for the mean 225 warming in the first 25 years after a pulse emission, as recently proposed by Shindell et al. (2017). Mean(ARTP(1-25)) is the average temperature response over the time period, which differ from ARTP(25) being a snapshot at the time horizon of 25 years. It has similarities with the iGTP concept introduced by Peters et al. (2011). We want to point out that our choice of metric is not based on a thorough scientific analysis, but rather a subjective choice to study the near-term climate impacts and the importance of short-lived species in more detail. To balance the choice we compare it with some other known 230 climate metrics. Reductions of CO2 and SO2 take place to large extent in the energy production sector following the reduction of energy consumption of fossil fuels, i.e. coal, oil and peat (Table 2).

Emissions
CH4 emissions have declined mostly due to developments in waste sector. Amounts of methane recovered from landfills 240 have increased during the study period following EU and national regulations. Methane emissions from landfills have also declined because the energy use of municipal solid waste has increased instead of landfilling; a development that is expected to continue also until 2030. Another factor explaining the declining emissions by 2030 is the prohibition of disposal of organic wastes to landfills after 2016.

245
The transport sector is responsible for the decline of the emissions of CO, NOX, VOC as well as the particle species, black carbon (BC) and organic carbon (OC). The modernization of the vehicle fleet and consequent introduction of stricter emission controls required by the EURO-standards explain the decline in CO, NOX and NMVOC emissions. The standards do not directly regulate BC or OC emissions, but since they are the main constituents of the regulated particulate emissions, reductions in emissions of BC and OC are expected, especially after the introduction of the diesel particulate filters for on-250 road light duty vehicles from 2010 onwards. The stoves and boilers in the residential sector will remain significant emitters of several pollutants, since the regulation following the European Union Ecodesign directive will not have major impact by 2030, due to the relatively long lifetime of Finnish heaters (Savolahti et al. 2016).
NH3 and N2O emissions remain relatively stable throughout the study period, since either much of the emission reductions 255 have already taken place before the study period or no major changes are expected in the main emission sectors (agriculture for NH3).   As we focus on near term climate change and the global and regional temperature, the remaining paper utilizes ARTP with a 290 time horizon of mean(GTP(1-25 yrs)), as proposed by Shindell et al. (2017). ARTP values are applied, following the argumentation by Aamaas et al. (2017) that ARTPs may give a better estimate of the global impact than AGTPs since they account for varying efficacies with latitude to a larger degree. The ARTP(1-25 yrs) used in this study are presented in Table   3. GWP* could also be a basis to estimate global temperature changes, but that would not give us regional temperature changes. 295

Climate impacts by emission sector
This section discusses the global temperature response of the emissions by pollutant and emission sector based on weighted ARTP values (Aamaas et al., 2017). The general findings described in the following paragraphs would be similar with AGTPs, and similar figures based on the AGTP values  can be found in Figure A1  following the baseline projection, respectively. The sum of all sectors is given in Figure 4D. The pollutant mix varies for the different sectors. CO2 is the most important pollutant for combustion in energy production and industry (ENE IND) and road transport (TRA RD), while methane is most important for waste (WST) and agriculture (AGR) sectors. BC emissions cause more than two thirds of the warming and increasing with time in the domestic sector (DOM) and a significant share of the warming in the on-road (TRA RD) transport as well as off-road transport and machinery (TRA OT) sources. Rest of the 305 warming effect for these sectors is due to CO2 emissions from fossil fuels, especially diesel and light fuel oil. Wood is an important fuel in the domestic sector, and since this study considers wood fuel as CO2 neutral, the CO2 warming effect is not as pronounced as, for example, in the on-road transport sector. Organic carbon has been the most important cooling agent in domestic and the transport sectors, as fuelwood does not contain much sulfur, and it has been phased out from liquid fuels in the transport sector as well. Overall, SO2 is the major cooling pollutant, mainly due to emissions from energy production 310 (ENE IND) and industrial processess (PROC). Agriculture is an important source of ammonia (NH3), which has a cooling effect (Fig. 4A-4C and Fig. 5A) via its participation in formation of cooling atmospheric aerosols like ammonium sulphates and nitrates.
Year 2000 was relatively warm and 2010 relatively cold in Finland, which is reflected as a higher use of coal, peat and wood 315 fuels in 2010, and consequently also as higher emissions for some species. The baseline projection will lead to emission reduction of all pollutants between 2010 and 2030, from more than a 50 percent reduction of BC to a small reduction for N2O ( Fig. 1 and Table A1). Because of climate policies, CO2 is reduced following 330 the declining use of fossil fuels ( Table 2, Fig. 4B, 4C and 4D). The SO2 emissions continue their decline between 2010 and 2030, particularly in the ENE IND sector, which leads to additional warming, but only partly offsetting the reduced CO2 ( Fig. 4B, 4C and 4D). In the on-road (TRA RD) and off-road (TRA OT) transport sector, particularly the warming effect from the SLCFs declines, because the new vehicles, in order to comply with the European emission legislation, are equipped with efficient emission reduction technologies ( Fig. 4B and 4C). The amount of domestic wood combustion is expected to 335 decrease in the baseline due to improved energy efficiency in housing, which is the main reason for the reduced SLCF emissions in the sector (Fig. 4B and 4C). However, when interpreting these results it is important to note that the prevalence of domestic wood combustion has been increasing during the 2000's and the future wood use in households is challenging to predict. Therefore the emissions from the domestic sector should be considered uncertain. This is demonstrated in a sensitivity analysis of future particle emissions from the domestic sector presented by Savolahti et al. (2016). Also the 340 methane emissions in the WST sector continue their decline ( Fig. 4B and 4C). As a consequence of the emission changes, the net-temperature impact of 2030 emissions is 35 percent lower compared with the 2010 emissions (Fig. 4D). Practically all sectors but AGR contribute to the reduced warming (Fig. 4B, 4C).

Cumulative temperature development 2000-2030 350
While most of our study focuses on emission pulses, in this section we will discuss global temperature responses given a Similar figures based on AGTP values  are given in Figure A2 of the Supporting material. Figure 5 demonstrates why emission reductions of CO2 and other long-lived greenhouse gases are key for limiting the long-term 355 surface temperature increase. As more years are added, the relative importance of CO2 increases, since a large portion of it stays in the atmosphere for hundreds of years. This relative importance over time also occurs in case of N2O. The air pollutants become of less relative significance with time, which is mostly because of those pollutants being quickly removed from the atmosphere, but also because of the reduced emissions levels in the later period. Almost all sectors have a netwarming temperature response, with the exception of cooling from ENE IND sector for more than the first ten years and a 360 slight cooling from PROC sector until 2030 ( Figure 5B). Cooling from mainly SO2 emissions is offsetting the warming impact of CO2 from those sectors. Over time, ENE IND becomes the most influential sector, being the single largest contributor of CO2. BC is the most significant warming pollutant in the domestic sector and CH4 for the agriculture and waste sectors.

Seasonal temperature response from Finnish emissions 370
The estimated temperature response of Finnish emissions varies between the seasons. In Figure 6, we compare Finnish SLCF emissions for the year 2010 during summer (May-October) and winter (November-April). A decomposition into different atmospheric forcing processes is also included. When we do not consider CO2, N2O, and CH4, the pollutants give a net cooling for emissions in summer and a net warming of equal size for emissions in winter. The main driver for this is larger BC emissions in winter combined with much stronger response from the snow albedo effect. The reason is that more than 70 375 percent of the annual emissions in the domestic sector occur in winter. Another important difference is the much stronger cooling by SO2 in summer. Some pollutants have both warming and cooling processes, such as BC and NOX. The same process can be warming for one pollutant and cooling for another, for example NOX emissions remove CH4 from the atmosphere (cooling), while VOC and CO emissions add CH4 (warming). Changes in the methane concentration will also influence ozone, giving rise to the methane-induced ozone effect and reinforcing the methane effect. 380 For a ton of BC emission with the ARTP metric the wintertime impact is higher by more than 120 percent than the summertime. Almost 80 percent of the net impact for winter emission comes for BC deposition on snow. The annual impact of winter emissions of BC is almost 80 percent with the ARTPs. From a mitigation perspective, these estimates indicate that attention should be placed on reducing winter emissions of BC. 385

Arctic temperature response from Finnish emissions
The seasonal differences for the SLCFs are also clearly seen in the Arctic temperature response (Fig. 7). Finland is closely situated to the Arctic as practically the whole country is north of the 60°N latitude and a significant area lies north of the Arctic Circle. Figure 7 shows the Arctic (between 60° to 90°N) temperature response based on the ARTP metrics from Aamaas et al. (2017) and Sand et al. (2016). As a general observation the temperature responses are larger in the Arctic (Fig.  395 7) compared with the global ones (Fig. 6). The trends are similar, with net cooling of summer emissions and net warming of winter emissions. The Arctic warming in winter is up to about 3 times larger than the cooling in summer mostly due to the outsized impact of wintertime emissions of BC. However, during summer, the cooling by SO2 emissions outweigh the warming by BC emissions.  than the global average. If we apply the ARTP methodology for GHGs, the response in the Arctic is up to 50% larger than the global average due to stronger local feedback processes in the Arctic (Boer and Yu, 2003). The ozone precursors have similar or weaker efficacies in the Arctic compared with the GHGs. However, the aerosols and sulfur emissions stand out with the largest differences (Fig. 8). By applying ARTP values from Aamaas et al. (2017) with scaling from Sand et al. (2016), we find that Finnish emissions of SO2 and OC have a 300 percent stronger efficacy in the Arctic than the global 410 average, and even higher for BC with 700 percent. A limitation with this method is that the scaling from Sand et al. (2016) is only applicable for the Arctic temperature response, which add some uncertainties to these Arctic vs. global ratios. For BC, this amplification in the Arctic is even stronger for emissions occurring in winter. Hence, the results indicate that mitigation of Finnish BC emissions is especially beneficial for limiting Arctic warming.

Discussion
The first objective in our study was to produce an integrated multi-pollutants emission dataset for Finland for 2000 to 2030. 420 We were able to achieve this aim, but it required the use of several data sources and studies that are not necessarily maintained at a regular basis. Future efforts should pursue to maintain the integrated multi-pollutant database developed for this work. This would require an integrated modelling environment, for example the Finnish Regional Emission Scenario model (FRES), and further work to fill in the gaps for the missing sectors and pollutants via developing relevant activity and emission factor databases into the FRES framework. 425 Our second set of objectives for this study was to compare different climate metrics and to assess their suitability for calculating the climate impact of a multi-pollutant emission set. Several air pollutants and greenhouse gases have detrimental impacts on global and regional climate, human health and wellbeing as well as crop yields (see i.e. Shindell et al. 2012).
Since the magnitudes and pathways of the effects differ between the constituents, integrated modelling is needed to 430 understand the consequences and form the basis for robust climate and air quality policies. This paper applied and compared various climate metrics to study the approximate integrated climate impact of Finnish air pollutant and greenhouse gas emissions globally and in the Arctic area. The results demonstrated that the relative impacts and importance of individual species as well as sectors can differ significantly between the studied temporal response scales, emission seasons as well as geographical scales for both emissions sources and temperature responses. Especially the warming or cooling impact of 435 SLCFs is sensitive to the studied time scale, with shorter time spans showing greater importance compared with GHGs.
Finnish emissions and their climate responses are relatively small; therefore it is challenging to use climate models to study the climate effect of national policies and to analyze the role of each pollutant and sector. This study demonstrated a method to overcome this challenge by utilizing emission metrics. All studied metrics provided interesting insights into the impacts of 440 Finnish emissions and which aspects could be emphasized when formulating mitigation strategies. We assessed that particularly the ARTP based metrics provided useful information. We preferred to use ARTP approaches to assess the impacts of Finnish emissions to both global and Arctic climate, because it includes the regional or latitudinal dimension of emission impacts in more detail. We also chose to use the mean(1-25 yrs) timeframe, since for the time being there is no established climate metrics for air pollutants, and this approach was recently suggested by Shindell et al. (2017) to be used in 445 connection with SLCFs. This is a subjective choice to study the near-term climate impacts and the importance of short-lived species in more detail. To our knowledge, we are the first to present metric values with mean(ARTP(1-25yrs)) and among the first to use the GWP*.
The use of ARTPs to study the impacts of Finnish emissions is useful for designing national emission mitigation strategies 450 also from a regional perspective. Finland is an Arctic country and a member of the Arctic Council, which is why there is high interest on understanding the Arctic impacts.
The third set of objectives was to estimate the climate impact of Finnish air pollutants and greenhouse gases utilizing the selected metrics. Our analysis across climate metrics, time horizons, pollutants and Finnish emission pathways demonstrated 455 that carbon dioxide emissions have the largest climate response also in the near term (10 to 20 years), and its relative importance increases the longer the time span gets. Hence, mitigation of carbon dioxide is crucial for reducing the climate impact of Finnish emissions. In the near or medium term, i.e. 25 year perspective, especially methane and black carbon have relatively significant warming impacts in addition to those of carbon dioxide. SO2, on the other, hand is an important precursor to light reflecting sulphate aerosol, thus having a cooling impact and offsetting part of the warming impact of the 460 other species.
Of Finnish emissions, the combustion in energy production and industry has the largest global temperature impact in the medium and long term due to significant carbon dioxide emissions, while sulfur dioxide emissions induce a shorter term cooling. Transport has the second biggest warming impact, and although that is expected to decrease notably by 2030 due to 465 stricter control on particulate and consequently black carbon emissions, it will remain a major source of carbon dioxide.
Emissions from domestic and agriculture sectors also have a considerable warming impact, and they will remain so, due to the relatively large respective emissions of black carbon and methane from the combustion of solid fuels, especially wood.
For all of the species the temperature response of Finnish emissions is generally stronger in the Arctic than globally, but 470 most significantly so in case of black carbon and sulfur dioxide. Results obtained with the ARTP metric indicated that especially mitigation of wintertime black carbon emissions are important for reducing the temperature increase in the Arctic.
Emissions of sulfur dioxide are expected to continue decreasing and this has many benefits (Ekholm et al. 2014). However, it will offset some of the climate benefits of the reduced carbon dioxide emissions, and this should be taken into consideration in climate assessments. 475 The fourth major objective of this study was to recommend a set of global and regional climate metrics to be used in connection with Finnish SLCF emissions. As a preparation for writing this paper we compared several climate metrics to be used in connection with Finnish SLCF emissions. We ended up relying on those presented in Aamaas et al. (2017) that in our understanding is currently the most complete set of climate metrics available for assessing the global and Arctic temperature 480 responses of European emissions. However, we have scaled those values with ratios from Sand et al. (2016) for the Arctic temperature response because that study provided ARTPs for Nordic emissions, which is more representative for the Finnish case. For the GHGs, we argue to apply the metric parameterization from IPCC AR5 (Myhre et al., 2013), but with an upward revision for CH4 (Etminan et al., 2016). The coefficients for mean(ARTP(1-25 yrs)) (see also Shindell et al. 2017) in Table 3 were used for assessing different mitigation pathways in a 25 year time span. This time window is relevant for policies that 485 focus on reducing global or Arctic warming in the near or medium term, from today and until 2040 or 2050. Corresponding mean(RTP(1-25 yrs)) values are shown in Table A2.  1-25yrs)) values for SLCF and GHG emissions.
The Arctic response for the GHGs is based on the latitudinal pattern for CH4. The annual average is based on emissions in 490 2010. Normalized values (CO2-equivalents) are shown in Table A2.
The assessed temperature impact of an emission dataset depends on the set of metrics available, as well as the applied metric setup, which bring uncertainties to the results. As there is no consensus on one individual set of metrics, especially in case of air pollutants, the results will differ between different studies. This work estimated the global and regional temperature 495 impacts of Finnish emissions based on methodologies in three recent papers (Sand et al., 2016;Aamaas et al., 2016;Aamaas et al., 2017). As all of these studies utilize partly the same radiative forcing datasets and partly similar general circulation models and chemistry transport models, we welcome other studies to complement the basis for our findings. Future work should continue to explore uncertainties and provide improved metrics.

500
Since the atmospheric lifetime of SLCFs is relatively short, their climate impact is more dependent on the emission region than with GHGs. Using Europe and the Nordic region as proxies for the emission region, as in this study, gives us a more representative picture of the Finnish case than would the global average. Further development of the metrics should use more precisely the geographical location of Finland as the emission region in order to provide more precise temperature estimates for the Finnish emissions. The snow albedo effect of BC is expected to be much larger for the northernmost emissions, as 505 indicated by a study for Norway by Hodnebrog et al. (2014). Future work should also focus on providing metrics for potentially missing species that could be important, for example dust aerosol.
Scientific literature has demonstrated that the climate impact of biomass combustion may depend on the timescale and forestry practices (i.e. Cherubini et al. 2011, Repo et al. 2012and Repo et al. 2015, which have not been a focus of this 510 study. Since the use of biomass for energy is important in Finland and will likely remain so in the coming decades, future studies could utilize metrics to study its climate impacts. This study has mostly focused on surface temperature metrics, however, other interesting impacts could be studied using the metric approach. For example Shine et al. (2015) has recently presented a new metric named the Global Precipitation change Potential (GPP), which is designed to gauge the effect of emissions on the global water cycle. 515 The understanding of the impact pathways of different pollutants has improved in recent years, which has led to further revisions of the climate impact estimates. Such development is expected to continue. The metric studies, however, are often based on earlier radiative forcing studies, and a time lag from new scientific understanding to this being reflected in the climate metrics exists. This study has utilized the latest metric studies, but there are already literature available, for instance 520 on BC, indicating that the temperature response may be smaller than demonstrated by the metrics used in this work (e.g., Stjern et al., 2017). As the understanding of the climate system improves, the estimated we give here for Finland should be revisited.

Conclusions 525
All studied metrics provided interesting insights into the impacts of Finnish emissions and which aspects could be emphasized when formulating mitigation strategies. We assessed that particularly the ARTP based metrics provided useful information, although one should not rule out the significance of the other temperature and radiative forcing based metrics, for the latter due to their relevance in connection with climate change mitigation work of the UNFCCC and IPCC. In the 530 future also other climate impact metrics should be explored and utilized. To enable such policy analyses an integrated multipollutant emission and metrics database, similar to the one used in this work, should be maintained.
Our analysis across climate metrics, time horizons, pollutants and Finnish emission pathways demonstrated that carbon dioxide emissions have the largest climate response also in the near term, 10 to 20 year time perspective, and its relative 535 importance increases the longer the time span gets. Hence, mitigation of carbon dioxide is crucial for reducing the climate impact of Finnish emissions. In the near or medium term, i.e. 25 year perspective, especially methane and black carbon have relatively significant warming impacts additional to those of carbon dioxide.
For all of the species the temperature response of Finnish emissions is generally stronger in the Arctic than globally, but 540 most significantly so in case of black carbon and sulfur dioxide. Especially the wintertime emissions are net warming, and even more so in the Arctic, mostly due to black carbon. The snow albedo effect of the Finnish BC emissions is found to be large and this phenomenon should be adequately included in the analyses. Since the atmospheric lifetime of SLCFs is relatively short, their climate impact is more dependent on the emission region than with GHGs. Our study demonstrated using the Finnish case that future studies and further development of the metrics should use precisely the geographical 545 location as the emission region in order to provide more precise temperature estimates.

Author contribution
KJK and MS compiled the emission data with supporting contributions from NK and VP. BA prepared the climate metrics databases and applied them to the emission data. KJK, BA and MS were leading the preparation of the manuscript. NK and VP acted as contributing authors. 550

Acknowledgements
This study has been financially supported by the Finnish Ministry of the Environment and the Ministry for Foreign Affairs of