Subarctic catchment water storage and carbon cycling – Leading the way for future studies using integrated datasets at Pallas, Finland

Subarctic ecohydrological processes are changing rapidly, but detailed and integrated ecohydrological investigations are not as widespread as necessary. We introduce an integrated research catchment site (Pallas) for atmosphere, ecosystems, and ecohydrology studies in subarctic conditions in Finland that can be used for a new set of comparative catchment investigations. The Pallas site provides unique observational data and high‐intensity field measurement datasets over long periods. The infrastructure for atmosphere‐ to landscape‐scale research in ecosystem processes in a subarctic landscape has recently been complemented with detailed ecohydrological measurements. We identify three dominant processes in subarctic ecohydrology: (a) strong seasonality drives ecohydrological regimes, (b) limited dynamic storage causes rapid stream response to water inputs (snowmelt and intensive storms), and (c) hydrological state of the system regulates catchment‐scale dissolved carbon dynamics and greenhouse (GHG) fluxes. Surface water and groundwater interactions play an important role in regulating catchment‐scale carbon balances and ecosystem respiration within subarctic peatlands, particularly their spatial variability in the landscape. Based on our observations from Pallas, we highlight key research gaps in subarctic ecohydrology and propose several ways forward. We also demonstrate that the Pallas catchment meets the need for sustaining and pushing the boundaries of critical long‐term integrated ecohydrological research in high‐latitude environments.

dynamics and greenhouse (GHG) fluxes. Surface water and groundwater interactions play an important role in regulating catchment-scale carbon balances and ecosystem respiration within subarctic peatlands, particularly their spatial variability in the landscape. Based on our observations from Pallas, we highlight key research gaps in subarctic ecohydrology and propose several ways forward. We also demonstrate that the Pallas catchment meets the need for sustaining and pushing the boundaries of critical long-term integrated ecohydrological research in high-latitude environments.
K E Y W O R D S biogeochemistry, catchment, hydrology, isotopes, measurements, subarctic 1 | INTRODUCTION High-latitude areas are facing an unprecedented transition in the near future as warming air, precipitation changes, and sea ice cover reductions continue to drive broad-scale and long-term change across the northern areas Bekryaev et al., 2010;Bintanja & Selten, 2014;Buchwall et al., 2020;Cohen et al., 2014;Ernakovich et al., 2014;Lee et al., 2011;Liston & Hiemstra, 2011;Neumann et al., 2019). Such transitional changes result in loss of freshwater ice (MacGreggor et al., 2020;Prowse et al., 2006;Sharma et al., 2019), with associated climate and ecological repercussions (Brown & Duguay, 2010). Long-term scientific monitoring programs are critical for detecting changes, understanding the baseline conditions, and quantifying the magnitude and frequency of change at high latitudes (Buchwall et al., 2020;Laudon et al., 2017;Luo et al., 2011). It is particularly important to have integrated research programs with crossdisciplinary approaches, uniting multiple scientific fields and enabling comprehensive analyses across boundaries in the Arctic landscape.
The Arctic water cycle is currently undergoing marked changes in three key components: the cryosphere, atmosphere, and hydrosphere Vihma et al., 2016). Changes in the Arctic water cycle reflect complex interactions in Arctic landscapes, processes, and ecosystems. Examples include retreating seasonal snow cover, increasing temperatures and precipitation, and shifts in precipitation from snow to rainfall (Bring et al., 2016;Hansen et al., 2019;Krasting et al., 2013;Pulliainen et al., 2020). Seasonal soil frost and permafrost conditions are predicted to change substantially Ala-Aho et al., 2021a, 2021bBiskaborn et al., 2019;Wang et al., 2019), modifying hydrological pathways and biogeochemical processes (Czimczik & Welker, 2010;Lupascu et al., 2018;Nowinski et al., 2010;Serikova et al., 2018) and causing infrastructure failure . The consequences of warming are likely to be severe in subarctic systems in the near future as slight changes in temperature can alter the magnitude and timing of snow accumulation and melt (Carey et al., 2010;Mioduszewski et al., 2014;Richter-Menge & Druckenmiller, 2020). Increased temperatures and the shift in growing seasons will cause rapid changes in the hydrosphere and ecohydrological connections within subarctic catchments, for example by amplifying lake bed drying and forest and shrub encroachment, and modifying CO 2 and CH 4 feedbacks (Blanc-Betes et al., 2016;Ives et al., 2012;Shen et al., 2014;Tape et al., 2012).
Understanding the complex linkages and interactions between water inputs, storages, water turnover times and outputs, and how hydrology controls biogeochemical cycles and vegetation water usage represents a major challenge in ecohydrology (Guswa et al., 2020).
However, substantial progress has been made using water isotope forensics (Ala-Aho et al., 2021a, 2021bBailey et al., 2019;Jespersen et al., 2018;Penna et al., 2018). A better understanding of hydrological processes is especially important in subarctic mosaic landscapes with high variability in connectivity between diverse landscape elements (Laudon & Sponseller, 2018;Lyon et al., 2010). It is important to gain a better understanding of how large-scale changes affect ecohydrological processes and to predict and mitigate these effects, particularly in subarctic and Arctic regions where ongoing changes are rapid (Bring et al., 2016). Long-term research stations offer valuable potential for monitoring ecohydrological processes and supporting modelling efforts, yet there are few stations with a full catchmentscale platform for ecohydrological studies in the subarctic region (Laudon et al., 2017). Due to the shortage of synchronous measured data on the climate-cryosphere-hydrology-ecology system in subarctic regions, and the highly variable spatial seasonal connectivity and disconnectivity within subarctic catchments, a clear understanding of their ecohydrological functioning is currently lacking (Jencso et al., 2009;Tetzlaff et al., 2007;Wainwright et al., 2011).
Hydrological processes are key factors that control carbon (C), nitrogen (N), and greenhouse gas (GHG) exchange and balances (Blanc-Betes et al., 2016). Lateral carbon fluxes, which occur in the form of dissolved organic (DOC), dissolved inorganic carbon (DIC) and particulate C, act as a key link between terrestrial and aquatic ecosystems (Csank et al., 2019;Giesler et al., 2014;Zimmer & McGlynn, 2018). To date, the few studies combining high-frequency atmospheric and aquatic flux measurements at the individual forest or peatland sites indicate that the proportion of aqueous flux in the net C balance may be considerable (Nilsson et al., 2008;Öquist et al., 2014;Worrall et al., 2009), but in other systems, it comprises only a small fraction of catchment C fluxes (Tomco et al., 2019). Furthermore, recent studies highlight the role of headwaters as hotspots of CO 2 losses from streams and show that groundwater contributes disproportionately to CO 2 in headwater streams (Duvert et al., 2018;Rocher-Ros et al., 2019).
Here, we synthesize new and existing data to produce a conceptual ecohydrological model for the Pallaslompolo catchment in Arctic Finland (hereafter Pallas). Pallas is an established research site for atmospheric, ecosystem, and environmental research . We propose ways forward and provide research suggestions for ecohydrology studies in the subarctic. A better understanding of subarctic ecohydrological processes at temporal and spatial scales can be a means of predicting and managing ecohydrological change. We also introduce the subarctic Pallas research catchment as a platform for atmospheric, terrestrial, and ecohydrological research, and a successful example of the benefits obtained from scientific cooperation and integration across disciplines.

| SITE DESCRIPTION
The Pallas area has a long monitoring history. The first weather station in the Pallas region was established in 1935, atmospheric research started in 1991 with air pollution monitoring, and in 1994 a Global Atmosphere Watch (GAW) station was established . The first eddy covariance flux station for monitoring ecosystem-atmosphere exchange of GHG was established at Kenttärova forest in 2002, followed by flux stations at Lompolojänkkä wetland (2005), Sammaltunturi fell mountain top (2011), and Pallaslompoloniemi lakeshore Aurela et al., 2015). The Pallas station is situated -in the sub-arctic region, in an area comprising forests, wetlands, lakes, and treeless fells . Pallas is part of the long-term Integrated Carbon Observation System (ICOS) network in Finland, which produces high-resolution gradient measurements of carbon dioxide (CO 2 ), water vapour, and energy exchange fluxes, together with several other ecosystem variables. Additionally, GHG fluxes are measured at various upland locations using, and in the stream and lake using floating chambers. The subcatchment Pallaslompolo, which is part of Lake Pallasjärvi catchment (total area 105.2 km 2 ), is part of the national research infrastructure in Finland (INAR RI Ecosystems) and in the wider circumpolar region. It contributes to over 15 European and global research and monitoring programs .
The Pallas catchment (total area 4.42 km 2 ) of the Lompolonjängänoja stream is located next to Pallas-Ylläs National Park in northern Finland ( Figure 1). The catchment drains to Lake Pallasjärvi and ranges in altitude from 268 m to 364 m a.s.l. Forest soils at Pallas are Podzols that consist of gravelly sandy and sandy tills, with an average organic layer depth of 3-4 cm, distinct eluvial Ae and enriched Bf/Bs horizons of varying thicknesses, and soil organic matter content that gradually decreases from $10% in topsoil to less than 1% at 90 cm depth. Hydraulic conductivity for forest soils ranges from 1.25Á10 À5 m/s to 1Á40 À6 m/s at 5 cm depth and decreases (7.66Á10 À6 to 9.78Á10 À7 ) at 30 cm depth. Although we do not have undisturbed soil samples for hydraulic conductivity analysis from deeper soil, a decrease in mean porosity from 0.53 at 5 cm depth to 0.26 at 60 cm depth and dry bulk density increase from 0.90 g/cm 3 to 1.89 g/cm 3 indicates that hydraulic conductivity may decline rapidly due to soil consolidation bellow 50 cm. Soil thickness is thin in high altitude areas ($1-3 m), but increases towards the catchment outlet. The landscape varies from coniferous forests to various types of mires: open fens, treed mires with typical poorly growing trees, and paludified forest areas.
The mires are characterized by hummock and hollow microtopography and can be considered 'pristine', although there are some old ditches in the catchment area in one of the upslope mires. Manual peat thickness measurements were conducted in some mire parts in different catchment locations and the maximum peat thickness found was 3.85 m and median 1.35 m. The average degree of humification of peat (depth 0-3 m) in the area is H4 (range H1-H7 in the top 80 cm of peat), based on the von Post scale (Hobbs, 1986). Hydraulic conductivity of the peat, determined using a direct-push piezometer with a falling head, varies between 1Á10 À3 and 1Á10 À8 m/s from the surface (10 cm) to deeper layers (1.6 m).
Hydraulic conductivity is relatively high (>1Á10 À5 m/s) for the top 60 cm of peat, indicating that the structure of the peatland matrix can promote lateral subsurface flow in the upper part of the peat profile. The Pallas site does not have permafrost but nearby areas have palsa mire formations containing permafrost layers.

| Climate and precipitation data
The climate of Pallas is characterized as subarctic with persistent snow cover during winter. The mean annual temperature at F I G U R E 1 Overview of the Lompolonjängänoja catchment at Pallas, northern Finland. The study catchment drains into the freshwater Lake Pallas. In addition to the gauging station at the outlet of the catchment, hydrological fluxes are monitored across the catchment. A network of stations providing meteorological, soil, and snow measurements supports hydrological research Kenttärova weather station is 0.4 C (2003-2019), ranging from

| Eddy covariance flux measurements at Lompolojänkkä mire, Kenttärova, and Pallaslompolonniemi
Ecosystem-atmosphere GHG exchange and meteorological conditions (precipitation, air temperature, potential evaporation, incoming and reflected solar radiation) are measured by the FMI at the Kenttärova forest site, Lompolojänkkä mire, and Pallasjärvi lake shore within the Pallas catchment, using eddy covariance flux stations ( Table 1) (Li-Cor, Inc.) CO 2 /H 2 O analyzer at the same height using an 8-m long heated inlet tube and a flow rate of 5-6 L min À1 . Synthetic air without CO 2 is used as a reference gas. The fluxes are calculated as 30-min averages from the 10 Hz data following standard protocols including coordinate rotation, peak removal, data signal synchronization, compensation for water vapour fluctuations, and correction for highfrequency attenuation (Aurela et al., 2009;Aurela et al., 2015;Lohila et al., 2015). Supporting meteorological measurements including air

| Peat and stream water CH 4 concentration measurements at Lompolojänkkä mire
Daily mean concentration of methane (CH 4 ) in the peat was monitored at Lompolojänkkä mire. Measurements were obtained using two PTFE tubes (length 50 m, internal diameter 8 mm, external diameter 10 mm) buried in the peat to a depth of $0.1-0.15 m, in two areas $500 m 2 north and west of the eddy covariance tower, respectively.
The tubes were automatically sampled once a day, with a flow rate of 2 L min À1 . It was assumed that the concentration in the tubes equilibrated with that in the peat during the 24 h after sampling. During the 3-min sampling, the CH 4 concentration was analysed with an Agilent 6890 N gas chromatograph (Agilent Technologies, Inc., Santa Clara, CA). For a more detailed description see Lohila et al. (2010). The concentration of CH 4 in stream water was monitored 17 times during the snowmelt period (April 27 to May 132 009) near gauging station PL6, using the headspace technique (Ding et al., 2002). In brief, immediately after sampling of three parallel stream water samples collected at the same time, 50 mL of water was flushed with 50 mL of synthetic air for 2 min, and the concentration in the headspace gas was analysed using the same gas chromatograph as above. The CH 4 concentration in water was calculated by applying Henry's law (Ding et al., 2002

| Data analysis
We conditions, but increasingly dry summer periods are expected in arctic areas (Bring et al., 2016). Thus isotopes can provide a tool to assess evaporative influence also in sub-arctic conditions. In other years, evaporation from the wetlands appears to be minimal based on dexcess values, possibly due to the cooler summer temperatures, coldair drainage at night, and high levels of water transported from the surrounding forest, through the wetlands and into the stream.
Values of DOC and DIC increase along the stream continuum from PL3 until PL5 (Figure 3). It is typically assumed that DIC is primarily controlled by GW inputs, whereas DOC is linked to surface water connections such as overland flow and runoff patterns (Giesler et al., 2014;Nydahl et al., 2020), as seen in the data (Figure 3) Arctic Ocean (Stroeve & Notz, 2018). These changes were recently quantified using the Pallas water vapour isotope monitoring program , which showed a direct link between extreme snowfall in northern Europe and increasing atmospheric moisture sourced from ice-free Arctic seas. However, the current understanding of how shifts in precipitation (timing and volume) and moisture sources affect Arctic hydrological processes at the catchment-scale is limited.
Meteoric and stream water isotope forensics ( can reveal seasonal changes and also sub-daily variation in the ecohydrological process in these highly responsive subarctic catchments.
Stable water isotope measurements, and the latest technology for high-frequency measurements in particular, provide new ways forward to better understand high-latitude catchments.

| Limited catchment dynamic storage controls hydrological responses
Overall isotopically-distinct precipitation types (rain and snowfall) allows direct seasonal partitioning of streamflow inputs using end-member mixing and splitting analyses (Kirchner & Allen, 2020), revealing that only small percentages of rain (8 ± 1%) and snowfall (5 ± 1%) are discharged from the Pallas catchment during the snowy season (Novearly May). Conversely, 40 ± 6% of rainfall and 24 ± 8% snowfall leave the catchment during the rainy season (late May-Oct), whereby season demarcation uses a rain-snow threshold of 1 C (Jennings et al., 2018, described in details in Dai, 2008. Based on these analyses, 69 ± 9% of streamflow during the snowy season originates from rainfall and 31 ± 9% from snowfall. During the rainy season, 70 ± 9% of streamflow comes from rainfall and 30 ± 9% from snowfall, which is similar to the snowy season. In these calculations, the isotope data and the precipitation data were used from the years 2014-2018.
These values are in accordance with the general water balance for Pallas. Despite the rapid catchment response and small STS storage, end-member splitting analysis indicated the dependency of the Lompolonjängänoja stream on inter-seasonal water storages in the catchment. Based on Kirchner and Allen (2020), the rate of evapotranspiration (ET) of precipitation differs from season to season.
For instance, 71 ± 9% of snowy season precipitation is lost as evapotranspiration, while 24 ± 8% and 5 ± 1% contribute to rainy and snowy season discharge, respectively. In the rainy season, a smaller fraction of precipitation (52 ± 7%) is lost to evapotranspiration, while 40 ± 6% and 8 ± 1% contribute to rainy and snowy season discharge, respectively. Annual ET derives almost equally from snowy season precipitation (48 ± 7%) and rainy season precipitation (52 ± 7%). Similar trends are also evident in the dual-isotope plot (Figure 2). For example, the stream data for spring and early summer 2014 show relatively low δ 18 O values during peak flow from snowmelt, followed by a rapid increase from À15.5‰ to À12‰ coincident with a large rainfall event (Figure 3(c)), and indicative of limited STS catchment storage capacity and an extensive network of rapid surface flow paths. Indeed, the Pallas hillslope is dominated by till with declining hydraulic conductivity to a depth of 30 cm, thus promoting overland strong flow formation as seen in stream discharge and isotopic data. Additionally, shallow soil structures maintain only limited soil and groundwater storage. Our study thus support previous findings conducted in boreal and sub-arctic sites (Sterte et al., 2021) and highlight the utility of stable water isotopes as hydrologic tracers in high-latitude catchments (Lyon et al., 2018).
Subsurface soil was fully saturated during the snowmelt runoff period, but moisture content declined markedly afterward (data not shown), indicating that the majority of snowmelt inputs produce runoff or generate groundwater recharge. In contrast, only large summer precipitation events (>20 mm) increased soil moisture content, indicating that summer precipitation either evapotranspired or produced direct runoff through preferential flow. Following larger rainfall events, catchment soils drain quickly (within days), indicating that soils in the upper hills do not store moisture between precipitation events.
Soil moisture measurements indicate that the upper layer (top 5 cm) responds more quickly than lower layers (30 cm) to snowmelt or rainfall inputs, demonstrating flow through the till layers' (cf. Bishop et al., 2011), and enabling quick flow along with the upper more permeable soil layer in the forest areas. These observations are further supported by the minimal lag (within minutes) between rainfall events and water level rise in peatland piezometers, as well as the simultaneous increase in stream discharge due to rapid rainfall runoff from the hills and peatlands directly into the stream. Hence, these data further support that Pallas catchment storage is limited by the relatively shallow, young soil deposit layers. Similar findings have also been observed in previous studies which emphasize how shallow soil deposits exert a strong influence on catchment-scale hydrological and geochemical responses Peters & Driscoll, 1987;Verry & Timmons, 1982). However, to our knowledge this study is one of the few to demonstrate such storage processes in sub-arctic or arctic conditions (Lyon et al., 2018;Sterte et al., 2021;Tetzlaff et al., 2018) and by utilizing multi-year stable water isotope data. In The TOC concentrations remain relatively steady during winter (less than 5 mg l À1 ), as discharge remains stable and is dominated by groundwater inputs (see parts 3.1 and 3.2), indicating minimal inputs of TOC to the stream during the winter months due to reduced connectivity with the wider catchment. Similar observations of groundwater influence on TOC have been made e.g., at permafrost sites (Lamontagne-Halle et al., 2018). However, as snowmelt proceeds, TOC concentrations increase substantially with increasing discharge (Figure 4(a)). Snowmelt dominates carbon fluxes in subarctic streams, and is estimated to be responsible for 55% of DOC flux in Arctic rivers (Finlay et al., 2006). Snowmelt leads to activation of carbon-rich flow paths, causing substantial increases in terrestrially derived in-stream organic and inorganic carbon (Giesler et al., 2014). Export of terrestrial carbon to the stream has a strong causal link with aquatic CO 2 emissions, with increased processing of terrestrially derived DOC linked to fluxes at the mire site  and in the larger Pallas catchment  suggest that hydrological pathways strongly controls the exchange rate and direction of CH 4 exchange. At the mire site, distance to the stream, which controls the area of oxic/ anoxic zone in the peat by bringing O 2 -rich water to the mire, is reported to be the most important factor explaining the spatial variability . For the whole catchment, CH 4 fluxes are reported to be best predicted by topographical wetness indices, Sentinel-1 SAR soil moisture data, and the digital terrain model .
While the strength and variability of CH 4 fluxes are known to be strongly related to hydrology (Knox et al., 2019;Räsänen et al., 2021;Zhang et al., 2020), the seasonal dynamics in CO 2 and CH 4 fluxes typically follow the patterns in vegetation activity and, both directly and indirectly (through phenology), the patterns in temperature. On the other hand, the non-growing season fluxes can be more directly related to hydrological processes, particularly at sites dominated by active surface flow patterns, such as Lompolojänkkä mire. Eddy covariance measurements at Lompolojänkkä show direct interactions between -snow melt period and CH 4 fluxes. In the example of spring 2009, the first warm spell in mid-April, which resulted in depletion in snow cover of approximately 10 cm in a few days, caused a significant CH 4 emissions pulse from the snowpack ( Figure 5). An effect of the first warm spell was not seen in any other GHG-related variables, so it was evident that the first peak of CH 4 was released from the snowpack during the first melting period. In CO 2 , no such peak was observed, which might be related to the strong interaction between CO 2 and water, based on complex inorganic chemical processes including dissolution of CO 2 and subsequent dissociation of CO 2 into other carbonic species, while CH 4 is much less soluble in water.
The first melting peak was followed by a temporarily colder period (below zero temperatures in day and night) which ended on April 23-24, 2009, after which the temperature mainly stayed above zero. This period of above zero temperatures initiated a second Notably, the soil at 7 cm depth never showed signs of freezing, but there was a clear ice cover on top of the peat and inside the snowpack, created by the groundwater seepage to add water in the snow and on top of the ice throughout the winter. The soil temperature at 7 cm depth only started increasing after the spikes in CH 4 emissions and changes in peat and stream water CH 4 concentrations had ceased. This result suggests that the observed CH 4 emission peak was due to thawing of the surface peat as the accumulated CH 4 was released into the atmosphere. At the same time, CH 4 dissolved in the stream water, additionally causing a CH 4 concentration spike in stream water.
To our knowledge, these are the first direct observations of a thawing-related CH 4 peak from a peatland with contemporaneous peat and stream CH 4 concentration data measurements. Previously, thawing peaks have been reported from polygonal tundra in Alaska (Raz-Yaseef et al., 2017) and permafrost peatland in China (Song et al., 2012), while no peaks were observed at a Swedish mire with sporadic permafrost (Lakomiec et al., 2021). Similar to these studies, we also conclude that the thawing-related CH 4 peak was attributed to a release of CH 4 stored in peat under the ice and snow. In our data, the height of the spring thawing peak was at the same level with the summer maximum (30-min values, data not shown from the July-August). Quite divergent results were found by Song et al. (2012) and Raz-Yaseef et al. (2017), who reported that the thawing peak clearly exceeded or was lower, respectively, as compared to the summer maximum. It is likely that the size of the peak is greatly affected by e.g. the length of the winter, thickness of the ice cover and rapidness of the melting, and great year-to-year variation in the thawing peak size and even emergence can be expected.
The peak in emissions to the atmosphere observed for CH 4 was not recorded for CO 2 , probably due to similar chemistry-related water-CO 2 interactions as observed for the first peak. However, CO 2 emissions gradually started to increase during the final melt, which is potentially attributable to microbial decomposition of organic matter in the peat. At the time of melting, the mire surface was brown as the vegetation had not sprouted after winter. Only later, around May 20, was the first sign of CO 2 uptake observed, after the emergence of the first green leaves. While it is generally acknowledged that hydrology and particularly GW level impose important controls on GHG dynamics, especially in wetlands (Heiskanen et al., 2021), the above example demonstrates the importance of short-term hydrological patterns for GHG emissions. It can be concluded that in the spring-time the connection between hydrology and GHG emissions is mainly related to transport of gases, while during the summer hydrology has direct influence on the microbial processes producing GHGs.

| Fine-resolution and catchment-scale spatial analysis reveals the importance of connectivity in ecohydrological studies
Our snapshot analysis of Pallas UAS data highlights the importance of fine-scale spatial analyses in high-latitude catchments. In particular, it reveals how snow accumulation and melting are spatially heterogeneous across the catchment and can be mapped in high-resolution  When combined with water sampling and catchment modelling, thermal imaging analysis (in time and space) can reveal hydrological connectivity at the catchment scale.
These observations highlight the potential of UAS in highresolution monitoring of snow depth, melt speed and their variability, and in identifying groundwater seepage locations in subarctic land cover types: that is, information that can be used to analyse their relationship to ecohydrological conditions in the subarctic. UAS data could be quantified and combined with more detailed information about snow cover (such as density, isotope composition, etc.) acquired from in-situ measurements to extend their coverage for catchmentscale analysis. Snow coverage and depth, spatial accumulation, melting and the timing of these factors are highly important for subarctic ecohydrological processes.

| WAYS FORWARD IN SUBARCTIC ECOHYDROLOGY
Northern ecosystems are highly sensitive to any changes in the cryosphere, and in particular to the loss of distinct temperature and precipitation regimes and the subsequent variation in ecohydrological conditions (Jyväsjärvi et al., 2015;Rolls et al., 2018). Even small changes in annual air temperatures or shifts in seasonal ranges can have drastic ecosystem-scale influences (Jennings et al., 2018) especially related to catchment hydrological responses and carbon processes. Additionally, an increasing proportion of precipitation falling as rain and/or changes in total volumes will rapidly shape ecohydrological conditions (e.g., snow, soil moisture, carbon transport, stream processes) already in the near future (Campbell & Laudon, 2019;Meriö et al., 2019). Most existing high-latitude research infrastructure is located in temperate or boreal regions, with less permanent infrastructure situated in subarctic or arctic areas (Laudon et al., 2017), and despite an increasing requirement to document baseline conditions and changing processes in the North (Burt & McDonnell, 2015). The Krycklan catchment in Sweden  has been a flagship for boreal hydrological and biogeochemical studies for decades. Yet, whilst the distance between the Krycklan and Pallas sites is less than 500 km, their climate, dominant vegetation, and geology types are markedly different. For instance, subarctic Pallas is located near the northern tree line, whereas Krycklan lies in the heart of the boreal landscape . Similarly, other long-term, high-latitude research catchment sites such as Toolik Field Station and Bonanza Creek LTER in Alaska (Medvedeff et al., 2021)  uio.no/), whereas the Pallas is northernmost existing research station without permafrost. Thus, Pallas bridges a critical gap in the global network of high-latitude catchment monitoring sites that are needed to improve current understanding of ecosystem changes and the hydrological and climatic processes driving them.
F I G U R E 6 (a) Unmanned aerial system (UAS)-based analysis of snow depth and calculated average snowmelt rate over a three-week period, revealing high spatial variation in melting rates at Pallas catchment. UAS-based analysis can reveal spatiotemporal variation in the melting speed of snow layers. (b) UAS thermal camera data analysis can detect groundwater seepage locations and catchment scale connections (bluish colour shows wettest parts of the peatland and groundwater seepage areas) We propose that future ecohydrological studies in subarctic regions focus on atmosphere-water-ecosystems interactions using combined measurements at broad spatial and temporal scales, and with linkages to physically-based catchment-scale and land-surface models. The novel technical capabilities emerging today allow a move from point measurements to spatial mapping (Figure 6), and from grab samples to high-resolution continuous monitoring in real-time ( Figures 4 and 5). Systematic monitoring of coupled atmospheric, biogeochemical, and ecohydrological processes facilitate a better understanding and quantification of how these complex systems interact and function today, and how they may change in the future under shifting seasonal patterns of synoptic climate, weather, snow, and sea ice in the subarctic.
The complex challenges of climate change and associated transformation of subarctic ecosystems call for more holistic science to inform local communities and policy for adaptive management (Prowse et al., 2015). Comprehensive ecohydrological science is needed to improve understanding feedbacks between climate-hydrology-ecosystem boundaries in subarctic conditions.
Our observations suggest that limited catchment dynamic storage in subarctic regions that are geologically similar to Pallas may have critical ecohydrological implications given future uncertainties in snowfall inputs , and especially the timing of snowmelt processes. Simultaneously, ecohydrological connectivity strongly influences local and landscape processes in terms of carbon (DOC and DIC) export and GHG emissions (Harms et al., 2020). The latest sensor technologies and high-frequency measurements can reveal highly-detailed processes as shown by our snap-shot analysis. Thus, pronounced variations in spatial catchment connectivity, such as during snowmelt and groundwater-surface water connections (Duvert et al., 2018;McKenzie et al., 2021) should be given more consideration as hotspots in the ecohydrological system using the latest technological advances. In studies on subarctic ecohydrology, there is a need to acknowledge the value of long-term monitoring for anticipating future changes (Laudon et al., 2017). Based on the analysis from Pallas and recent findings in other studies, we suggest that future studies in subarctic catchments should focus on: (a) analysis of pronounced seasonality (Post et al., 2019) in the subarctic ecohydrological regime and processes, (b) water dynamic storage, pathways, and soil moisture controlled by ecohydrological processes (Kuppel et al., 2020), and (c) ecohydrological controls regulating GHG and stream fluxes, especially in GW-SW transition zones such as riparian wetland areas .