Upwelling-induced trace gas dynamics in the Baltic Sea inferred from 8 years of autonomous measurements on a ship of opportunity

data analysis are done. Moreover, the overarching scientific objectives addressed by the study need to be given. 5) Section 3 ‘Results and Discussion’: Moreover, I am wondering why the authors do not discuss the effects of the ongoing environmental changes of the Baltic Sea (such as warming, changing wind patterns etc.). An important question to be addressed might be: Are there any trends detectable for the upwelling-induced CO2/CH4 fluxes during the course of the study which after all 140 covers eight-years? If yes, what are the main factors causing this trend? Jacobs et al. present 8-years of underway surface CO2 and CH4 measurements from the Baltic Sea. They assess the role of upwelling on surface gas concentrations and fluxes on seasonal time-scales, and describe typical annual cycles, as well as anomalies. The paper is very well written, and thoroughly describes regional and temporal differences in CO2 and CH4 concentrations, showing clearly the influence of upwelling and temperature. The methods used appear to be robust, and well-explained, with careful consideration of potential sources of error. The data set itself is of tremendous value, and the interpretation is well-done and could be applied to other regions with underway CO2 and/or CH4 systems. Although it could be argued the paper lacks clear objectives or motivation, I propose the value of this paper is in the methodological development used to extrapolate discrete underway data, thus improving its already high-resolution. Additionally, the development of a robust technique for identifying upwelling and linking it with observations on these spatiotemporal scales is well done and could be of value to others interpreting similar data sets, which could facilitate more robust extrapolation of such measurements in regions sorely lacking data. This makes for a valuable contribution to understanding the importance of upwelling on temporal and spatial variability in CO2 and CH4 flux, and a delight to read. Abstract. Autonomous measurements aboard ships of opportunity (SOOP) provide in situ data sets with high spatial and temporal coverage. In this study, we use 8 years of carbon dioxide (CO 2 ) and methane (CH 4 ) observations from SOOP Finnmaid to study the inﬂuence of upwelling on trace gas dynamics in the Baltic Sea. Between spring and autumn, coastal upwelling transports water masses enriched with CO 2 and CH 4 to the surface of the Baltic Sea. We study the seasonality, regional distribution, 5 relaxation, and interannual variability of this process. We use reanalysed wind and modelled sea surface temperature (SST) data in a newly established statistical upwelling detection method to identify major upwelling areas and time periods. Strong (cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58) surface (cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58)(cid:58) expressions (cid:58)(cid:58)(cid:58) of upwelling events are most frequently detected around August after a long period of thermal stratiﬁcation, i.e. limited exchange between surface and underlying waters. We found that these strong upwelling events with large SST excursions shape local trace gas dynamics and often lead to near-linear relationships between increasing trace gas levels and 10 decreasing temperature. Upwelling relaxation is mainly driven by mixing and (cid:58) , modulated

as eutrophication (HELCOM, 2018), changing wind patterns, and, thus, upwelling intensity (BACC II Author Team, 2015), and has also been shown to encounter a decrease of oxygen on a large number of coastal sites over the last decades (Caballero-Alfonso et al., 2015). The Baltic Sea, therefore, offers the potential to study feedbacks between anthropogenic and climatic drivers and upwelling-induced greenhouse gas fluxes early and with a high signal intensity." (Line 28) 50 The following paragraphs were only slightly changed, based on, e.g. suggestions by Reviewer #2. We moved the paragraph on SOOPs and on the instrumentation (which was the first one before) towards the end of the introduction.
We then summarise our motivation and give reasons to do these kind of studies and develop new methods in the Baltic Sea, using SOOP data: 55 "The large extent to which the Baltic Sea is influenced by climatic and anthropogenic forces and the availability of the presented eight-years data set of SOOP Finnmaid and of high-resolution models (Placke et al., 2018; Gräwe et al., 2019) make the Baltic Sea a unique study site to detect feedbacks early and to develop methods and process understanding that can be used to analyse long-term data sets with respect to, e.g. upwelling-induced trace gas dynamics. The SOOP strategy allows us to investigate the influence of coastal upwelling on surface pCO2 and cCH4 60 in the Baltic Sea on a large spatial and temporal scale without issues of bad coverage of seasonality due to (biased) individual RV-based studies. Furthermore, methods developed here can possibly be used for the treatise of upwelling in regions that are more important for global trace gas fluxes and budgets." (Line 90) As requested, we revised the presentation of our objectives, as well: In this study, we: 65 -present a method to identify upwelling events along the SOOP track based on wind and modelled SST data, -compare upwelling-induced trace gas dynamics within several regions in the Baltic Sea, -examine the relaxation of upwelling events over time with a focus on underlying processes, -discuss interannual variability within the data set with a focus on controlling mechanisms of seasonality and highlight the importance of upwelling to understand CO2 and CH4 dynamics in the Baltic Sea, 70 -test whether a long-term trend can be inferred from the analysis of this eight-years data set, and -demonstrate the potential of extrapolating trace gas observations based on modelled SST data and show ways to estimate air-sea trace gas fluxes on a broader spatial scale under extremely variable conditions.

(Line 97)
75 2) Section 2 'Data methods': I would like to suggest to move sections 2. 2 and 2.4-2.6 to the Appendix. The information given in these sections is relevant only for side aspects of the data analysis. (Please note that Fig. 9 is already mentioned in section 2.4, so the numbering of figures is not correct, it should appear as Fig. 5) This should indicate clearer which advancements have been made compared to previous studies. We also removed the corresponding statement in the abstract. 180 Minor points:

1) Section 2 'Data and Methods' (and throughout the rest of the text): The authors use the term 'saturation concentration' wh ich is misleading. This term should be replaced with 'equilibrium concentration'.
Equilibrium concentration is in fact the more suitable expression, thank you. We replaced it accordingly throughout 185 the manuscript, but stuck to the phrases "supersaturated / undersaturated" as these are clear within the context (we defined relative saturation as ratio of cCH4 to equilibrium concentration) and are used frequently in the literature in similar context.
2) Figure 1: Please indicate the location of the Uto station in the map.
3) P5L101-103: Please note that a concentration is only independent from temperature when it is given as mol kg-1. If it is given as mol L-1 (as in the ms) it is not independent from temperature. Moreover, the partial pressure is depending on the temperature when you refer to the partial pressure in equilibrium with the water phase. Please correct.

195
We corrected/clarified both statements. We replaced with: "Note that -neglecting the very small influence of temperature on water density -we can handle the concentration (of CH4 in nmol L -1 ) as a quasi-conservative parameter with respect to temperature changes in the discussion. In contrast, the partial pressure (of CO2) in equilibrium with the water phase is temperature-dependent (see also Sect.

3.3)." (Line 122) 200
At the end, we have the feeling that the manuscript has improved a lot as a result of your suggestions. The rewritten introduction should give better context and orientation than the old one, and the added information to discussion and conclusion should give more insight into the details and perspectives of our work. Thank you again for your 205 valuable and helpful review.
Thank you very much for the thorough review and your valuable recommendations and suggestions, which surely 210 helped to improve the manuscript.  of underway surface CO2 and CH4 measurements from the Baltic Sea. They assess the role of upwelling on surface gas concentrations and fluxes on seasonal time-scales, and describe typical annual cycles, as well as anomalies. The paper is very well written, and thoroughly describes regional and temporal differences in CO2 and CH4 concentrations,showing 215 clearly the influence of upwelling and temperature. The methods used appear to be robust, and well-explained, with careful consideration of potential sources of error. The data set itself is of tremendous value, and the interpretation is well-done and could be applied to other regions with underway CO2 and/or CH4 systems.
Although it could be argued the paper lacks clear objectives or motivation, I propose the value of this paper is in the methodological development used to extrapolate discrete underway data, thus improving its already high-resolution. Additionally, the 220 development of a robust technique for identifying upwelling and linking it with observations on these spatiotemporal scales is well done and could be of value to others interpreting similar data sets, which could facilitate more robust extrapolation of such measurements in regions sorely lacking data. This makes for a valuable contribution to understanding the importance of upwelling on temporal and spatial variability in CO2 and CH4 flux, and a delight to read.
Thank you for the encouraging words. The objectives you mentioned are indeed main foci of our study. As 225 Reviewer #1 pointed out, however, we had to emphasise these goals clearer in the introduction, which we now did.
I have only minor suggestions to improve clarity of figures (especially regarding choice of colors), and text. I recommend publication of the manuscript. We have spent many thoughts on this topic and all colour scales in the manuscript have been chosen with the colour-blindness issue in mind. The colour scales we used (https://cran.rproject.org/web/packages/viridis/vignettes/intro-to-viridis.html) are perceptionally uniform and robust to both colour-blindness and grey-scale printing. We would like to keep all those advantages. However, we do agree that 235 the differences between colours are rather small in Fig. 2, which is partly because we left out yellow. You find the new version of the plot below, which includes yellow (thereby widens the colour gradient) and has bolder lines to make the colours more distinguishable. We hope that this improved version will find your approval.

240
Figure 2 already contains an oxygen profile. However, we do realize that we did not define "cO2" because we do not need the term elsewhere and that it is rather easy to be confused with "CO2". Therefore, we replaced the legend label with "O2 concentration" to make it clearer (see above). Indeed, the two sentence parts do not really fit together and we revised the sentence according to your recommendation: "The typical seasonality of surface carbon dioxide (CO2) partial pressure (pCO2) in the Baltic Sea features a minimum 250 during spring and one or more subsequent minima throughout summer" (Line 37)   Abstract.
Autonomous measurements aboard ships of opportunity (SOOP) provide in situ data sets with high spatial and temporal coverage. In this study, we use 8 years of carbon dioxide (CO 2 ) and methane (CH 4 ) observations from SOOP Finnmaid to study the influence of upwelling on trace gas dynamics in the Baltic Sea. Between spring and autumn, coastal upwelling transports water masses enriched with CO 2 and CH 4 to the surface of the Baltic Sea. We study the seasonality, regional distribution, 5 relaxation, and interannual variability of this process. We use reanalysed wind and modelled sea surface temperature (SST) data in a newly established statistical upwelling detection method to identify major upwelling areas and time periods. Strong :::::: surface ::::::::: expressions ::: of upwelling events are most frequently detected around August after a long period of thermal stratification, i.e. limited exchange between surface and underlying waters. We found that these strong upwelling events with large SST excursions shape local trace gas dynamics and often lead to near-linear relationships between increasing trace gas levels and 10 decreasing temperature. Upwelling relaxation is mainly driven by mixingand : , modulated by air-sea gas exchange, : and possibly primary production. Subsequent warming through air-sea heat exchange has the potential to enhance trace gas saturation. In 2015, quasi-continuous upwelling over several months led to weak summer stratification, which directly impacted the observed trace gas and SST dynamics in several upwelling-prone areas. :::: Trend ::::::: analysis :: is ::: still ::::::::: prevented :: by ::: the :::::::: observed :::: high ::::::::: variability, :::::::::: uncertainties ::::: from ::: data ::::::::: coverage, ::: and :::: long ::::: water :::::::: residence ::::: times :: of ::::: 10-30 :::::: years. We introduce an extrapolation method based 15 on trace gas -SST relationships that allows us to estimate upwelling-induced trace gas fluxes in upwelling-affected regions. In general, the surface water reverses from CO 2 sink to source and CH 4 outgassing is intensified as a consequence of upwelling.

45
Map and bathymetry of the study area with typical routes of SOOP Finnmaid (grey) between Lübeck-Travemünde (L) and Helsinki (H). Boxes highlight seven regions (solid lines) in which we expect upwelling to occur and one in the open Gotland Sea (dashed lines) as reference (Table 1). Red dots mark the locations of wind data used for this study with red arrows indicating upwelling-favourable wind directions. We further marked the islands of Bornholm (B), Öland (Ö), Gotland (G), and Hiiumaa (Hi).
Local coastal upwelling increases both pCO 2 and CH 4 concentration (cCH 4 ) in the surface water of the Baltic Sea by introducing enriched water from below the summer thermocline ( Fig. 2) to the surface (Gülzow et al., 2013;Norman et al., 2013;Schneider et al., 2014b;Humborg et al., 2019;Stawiarski et al., 2019). However, since previous studies were limited to episodic events, only little is known about the seasonality and regional distribution of upwelling-induced trace gas signals in 80 the Baltic Sea and their relaxation. Ferry-based measurements enable studies on upwelling in the Baltic Sea on larger scales, which has already been demonstrated for physical parameters in the Gulf of Finland (Kikas and Lips, 2016).
Upwelling events in the Baltic Sea are common, but irregular since they depend on wind conditions (Lehmann and Myrberg, 2008). Westerly to south-westerly winds prevail in the Baltic Sea area, which enhances the possibility of upwelling near southern and south-eastern coasts (see red arrows in Fig. 1). These upwelling events have a typical lifetime of several days up to one 85 month with sharper horizontal gradients compared to oceanic upwelling (Lehmann and Myrberg, 2008). In summer, sea surface temperature (SST) may decrease by more than 10 • C during an upwelling event, while salinity changes are usually below 0.5 (Lehmann and Myrberg, 2008). The influence of upwelling decreases in autumn and winter, when no seasonal stratification is present. While for oceanic upwelling regions, it is known that upwelling may trigger extreme primary production through nutrient transport, the influence of upwelling on primary production in the Baltic Sea is still poorly constrained (Lehmann and 90 Myrberg, 2008). However, upwelled waters characterised by low N/P ratios have been reported to fuel cyanobacteria blooms during nitrogen limitation (Vahtera et al., 2005;Lips et al., 2009;Wasmund et al., 2012). Yet, time lags of about three weeks are possible :::: have :::: been ::::::: reported : for this feedback with an initial decline of phytoplankton biomass (Vahtera et al., 2005;Wasmund et al., 2012). As an explanation for this delay, Wasmund et al. (2012) propose that the initialisation of a cyanobacteria bloom requires mixing of biomass-rich surface water with phosphate-rich upwelled water.
We present most of the findings by use of illustrating examples, but provide more information in the appendix, supplement, 130 and data set.
2 Data and methods

Measurements aboard SOOP Finnmaid
SOOP Finnmaid is equipped with a variety of sensors to survey the surface water of the Baltic Sea between Lübeck-Travemünde in Germany and Helsinki in Finland ( Fig. 1). Parameters including SST, salinity, pCO 2 , and cCH 4 are logged every minute.

135
The data set used for this study is filtered ::::: refers :: to ::: the :::: time :::::: period from May to September and within the regions defined in Table 1 and consists of 482 transects from 2010 to 2017 with about 395,000 observations. ::::: 2017.
We post-calibrated xCO 2 and xCH 4 using a single-point calibration to a standard gas at near-atmospheric concentrations.
These standard gas measurements were performed automatically when leaving the harbour to yield a drift correction between transects. The measurement system aboard SOOP Finnmaid also includes a LI-COR 6262 CO 2 /H 2 O analyser with a separate equilibrator. Even though these additional CO 2 data are not presented here, they provided cross-validation and quality 155 6 control. Presenting both CO 2 and CH 4 measurements from the same instrument ensures best consistency between the two trace gases. Therefore, minor deviations from the previously published CO 2 data set in SOCAT (Surface Ocean CO 2 Atlas, https://www.socat.info, Bakker et al., 2016) exist, which is a combined product of both setups. In the study area and period, these differences in pCO 2 have a median of 0.75 µatm and an interquartile range (IQR) of 2.1 µatm.
We further used monthly-averaged atmospheric CO 2 and CH 4 data to calculate atmospheric equilibrium conditions. For the 160 closest distance to observations from SOOP Finnmaid, we utilised atmospheric data from Utö station (Finnish Meteorological Institute, Helsinki) starting in March 2012. Prior to that or to fill gaps in the Utö series, atmospheric data from Mace Head station (National University of Ireland, Galway) via the NOAA ESRL Carbon Cycle Cooperative Global Air Sampling Network (Dlugokencky et al., 2019a, b) were used and both data sets were matched to those of Utö by linear regression (Fig. A2 for details). The atmospheric data are displayed as atmospheric partial pressure for CO 2 or as saturation ::::::::: equilibrium : concentration 165 calculated from SST and salinity for CH 4 . We also plotted relative CH 4 saturation, which is the ratio of cCH 4 to saturation ::::::::: equilibrium : concentration.
We used the SST output of the numerical ocean model GETM (General Estuarine Transport Model) for the Baltic Sea.

175
The model has a horizontal resolution of 1 nautical mile and 50 vertical terrain-following levels. The uppermost level has a maximum thickness of 50 cm to properly represent the SST and ocean-atmosphere fluxes. The model run covers the period from 1961 to 2019. For a detailed analysis of the model performance see Placke et al. (2018) and Gräwe et al. (2019). For the present run, we restarted the model in 2003, but changed the atmospheric forcing to the operational reanalysis data set of the German Weather Service (DWD), with a spatial resolution of 7 km and a temporal resolution of 3 h (Zängl et al., 2015).

Identification of upwelling events
Based on the statistical analysis of upwelling in the Baltic Sea by Lehmann et al. (2012), we defined major upwelling areas that SOOP Finnmaid crosses as foci of this study ( Fig. 1 and Table 1). We excluded the Arkona Basin and the Mecklenburg

190
Bight (areas west of 14 • E) because strong wind may trigger vertical mixing through the entire water column in these shallower Table 1. Upwelling areas crossed by SOOP Finnmaid (abbreviated and long name), their boundaries (latitude/longitude) including a specification if the respective box is not rectangular (Fig. 1), their upwelling-favourable wind direction and the distance between the track of SOOP areas, thereby eliminating the usual decoupling between sediment and surface water and greatly enhancing surface trace gas concentrations (Gülzow et al., 2013). Thus, it is impossible to disentangle the influence of wind-induced upwelling in these areas by the method proposed here. We included an area in the open Gotland Sea, which should not be directly influenced by upwelling due to being far from the coast (> 40 km, Table 1), for comparison. Furthermore, we only considered data from May 195 to September each year, when upwelling-induced SST signals can be observed (Lehmann et al., 2012), which is -together with a wind criterion -the basis of the detection method we used.
According to Lehmann et al. (2012), we defined an upwelling event as: upwelling-favourable wind, i.e. the wind component projected parallel to the coast ( Fig. 1) exceeding 3.5 m s −1 for two days, causing a temperature drop by more than 2 • C in the respective box. Both criteria (wind and ∆SST) were evaluated per day and box and visualised in yearly plots, which also 200 display data coverage of SOOP Finnmaid (Fig. 4, A3, and A4).
We calculated the wind criterion as running mean of upwelling-favourable wind speeds of the last 48 h with a temporal resolution of 3 h. The criterion is considered as "met" if at least 4 of the 8 mean values per day exceed the threshold of 3.5 m s −1 .
The ∆SST criterion was calculated as difference between median and minimum model-SST in the respective area, since 205 a local upwelling event will lower the minimum SST while the median remains relatively stable. To achieve a more robust median, the boxes were selected to extend beyond the actual upwelling areas. This results in a pronounced increase in ∆SST during upwelling events, while the criterion is mostly below the 2 • C threshold otherwise. This calculation can be based on These verification measures were calculated according to Jolliffe and Stephenson (2003). The differences in number of events correctly identified depending on which subset of SST data is used are explained by the fact that upwelling events start near the 215 coast and then propagate seawards and, thus, SST drops along the sub-transect are delayed and often smaller. The first approach (using all SST data within the upwelling box) does not incorporate this lag, but is usually better aligned with the wind criterion for the same reason. Therefore, the appropriate method choice depends on the desired use: Including more spatial coverage of SST data would be appropriate to analyse the occurrence of upwelling in a certain region statistically. However, we chose to use only SST data along the SOOP route to amplify the agreement with the in situ SST and trace gas measurements. We 220 provide a more detailed method assessment in Sect. 3.1.
Note that a large-scale upwelling event triggers a drop in median SST, but due to increased spatial variability during those events, the sensitivity of ∆SST is usually still sufficient to exceed the threshold of 2 • C in these cases (e.g. Fig. 4 and A3 on ca. 10 August 2016).

Calculation of theoretical relaxation and flux estimates 225
We calculated theoretical relaxation curves (Sect. 3.3) as follows: CO 2 system calculations were performed using the R package seacarb (Gattuso et al., 2019) with K 1 /K 2 from Millero (2010), K w /K f from Dickson and Riley (1979) and K S from Dickson (1990). pCH 4 was calculated from cCH 4 (Wiesenburg and Guinasso, 1979). Since salinity changes by upwelling in the Baltic Sea are usually small (Lehmann and Myrberg, 2008) and no calcifying organisms are present (Schneider et al., 2014a), we assumed a constant salinity of 7 and a total alkalinity (A T ) of 1600 (Müller et al., 2016). These along with the values for the 230 initial (upwelled) water mass (SST = 10 , pCO 2 = 700 , cCH 4 = 5 ) and the background water mass (SST = 20 , pCO 2 = 150 , cCH 4 = 3.5 ) used for mixing are typical for late-summer upwelling in the central box Go-SE (Fig. 7d,l). However, as we mainly discuss the shapes of the curves, which are unaffected by variation of the input variables over a reasonable range, Fig. 8 is used to discuss processes in all regions.
Fluxes were calculated according to Wanninkhof (2014). We approximated the Schmidt number dependence on salinity via 235 linear interpolation between the values for freshwater and seawater. In Sect. 3.3, we assumed constant wind speeds of 10 , SST = 10 , air temperature = 20 , relative humidity = 0.8, and relative cloud coverage = 0.8. In Sect. 3.5, we used the available wind data with 3 resolution to calculate daily fluxes.

Air-water equilibrator response times
Gas phase measurements using air-water equilibrators are subject to response times (Johnson, 1999), which depend on construction 240 and operation parameters of the setup, solubility of the respective gas, temperature and salinity (Webb et al., 2016). The e-folding time constants τ of the system aboard SOOP Finnmaid were determined to be 226 for CO 2 and 676 for CH 4 at room temperature using fresh water (Gülzow et al., 2011). Non-negligible response times lead to smoothed and delayed signals in both time and space, with a more pronounced impact on CH 4 than CO 2 . Corrections for temporal and spatial lag are used in profiling sensor applications (Fiedler et al., 2012;Bittig et al., 2014), but they are often neglected for surface trace 245 gas measurements. We demonstrate such a correction using the method described in Bittig et al. (2018) to illustrate potential advantages and practical issues (Fig. A1). In the illustrated example, SOOP Finnmaid travels from south-west to north-east through box Go-SE (Fig. A1h). Compared to SST measurements (Fig. A1a), the CO 2 and especially CH 4 signals are delayed and smoothed (Fig. A1b,c, black curves). In comparison, the corrected signals (Fig. A1b,c, red curves, most prominently around 18.5) respond earlier, are more pronounced, 255 exhibit more fine structure, and mirror the SST signal better, which is expected when entering a new water mass. The relationships between uncorrected trace gas signals and SST (Fig. A1d,f) feature hysteresis, which is reduced substantially after the correction (Fig. A1e,g). However, the method introduces artefacts like overshoots (e.g. Fig. A1b,c, low values around 19.15) and noise particularly if data density is low and/or τ is poorly characterised. This problem can be mitigated, but not solved, by applying additional smoothing before and/or after the correction (not done here).

260
Unfortunately, this response time correction only provided satisfactory results for a minority of cases. Elsewhere, the resulting noise degraded data quality and created additional hysteresis. We attribute this to the unknown dependence of τ on, e.g. temperature, salinity, and water/gas flows, all of which vary along a transect. This issue would be particularly influential for this study since, e.g. SST gradients caused by upwelling are sharper and steeper compared to measurements in open basins, which leads to perpetual changes of τ . Thus, we decided to refrain from a response time correction to avoid introducing 265 additional bias into the data set. However, the algorithm we used (Bittig et al., 2018) is capable of handling variable τ , allowing more precise response time corrections if τ is sufficiently characterised as function of, e.g. temperature, salinity, and air/water flows, which leaves room for future studies.

Results and discussion
3.1 Upwelling statistics based on wind and modelled SST data

275
To assess the prevalence of upwelling in the data set of SOOP Finnmaid, we identified the main upwelling periods and areas along the transect using the method of combining a ∆SST and a wind criterion (Sect. 2.3). Here, we present a summary of the climatological mean number of days per month and box where the criteria were met (Fig. 5). The ∆SST criterion was calculated based on sub-transects. We provide a full overview further distinguishing by year and selection of SST data (entire box vs. sub-transect) in the appendix (Fig. B1, B2, and B3). The wind criterion is usually met more frequently than the ∆SST 280 criterion calculated along sub-transects. This reflects that not every occurrence of wind strong enough to induce upwelling leads to upwelled water masses actually reaching the track of SOOP Finnmaid. This is illustrated in Fig. 4 (June 2016) and in the Supplement S1. Downwelling may also lead to quickly vanishing signals (Sect. 3.3). In general, the ∆SST criterion is not very sensitive in May due to a less pronounced thermocline compared to summer. Similarly, only small upwelling-induced trace gas signals are observed in May, which become greater in late summer owing to longer decoupling of surface water and underlying only applies to instances of both criteria being met on the exact same day and therefore excludes occasional instances of lag between wind and ∆SST signals. Box openGo is not included here since no upwelling-favourable wind direction can be defined.
sub-thermocline waters (Sect. 3.4). Upwelling in autumn and winter either leads to a general deepening of the mixed-layer depth (discussed in Gülzow et al., 2013) or plays no important role when the physical and biogeochemical differences between surface and upwelled waters have vanished in winter.
The ∆SST criterion based on the entire area is more sensitive than that calculated along sub-transects, but less specific regarding the prediction of upwelling in dynamic areas like :: the : GoF since it is essentially a measure for SST variability > 2 • C 290 within the box (Sect. 2.3). It is triggered more frequently than the wind criterion and due to the high sensitivity of ∆SST, the agreement with the wind criterion and, thus, both criteria being met, is high (Fig. B3).
Boxes Born, Ö-S, Ö-E, Go-SE, and GoF follow similar patterns with respect to both criteria (Fig. 5), which is not surprising given the fact that in all of these cases, upwelling is induced by the same south-westerly to westerly winds and the minimum distances to the coast are rather similar. The upwelling-favourable wind direction is opposite in boxes Go-NW and Hiiu. In box 295 Hiiu, however, both criteria are almost never met at the same time, which indicates that the distance between sub-transect and coast is too large to observe strong upwelling signals (minimum 27 km, median 43 km, Table 1). Admittedly, the sub-transect in box Ö-S is comparable to Hiiu in terms of distance to the coast, but the crucial difference seems to be the upwelling-favourable wind direction since strong westerly winds are more frequent and intense. This is supported by the fact that, even if we calculate ∆SST based on the entire area, the number of days where both criteria are met in box Ö-S is higher than in box Hiiu (Fig. B3), 300 clearly indicating that upwelling is more common in Ö-S. In contrast, the sub-transect in box Go-NW is frequently influenced by upwelling, but yet, it is the only box where the ∆SST is met more often than the wind criterion (Fig. 5e). We attribute this to the small distance to the coast (minimum 4 km, median 10 km, Table 1) leading to higher SST variability and, thus, more similarity to the ∆SST criterion that includes the entire area (Fig. B3e). Based on this statistical analysis, we chose box Ö-S as example area for most of the following discussion since it features 305 prominent upwelling signals concerning both frequency and magnitude. Additionally, data coverage in this area is high as it is crossed by SOOP Finnmaid on either of its routes (Fig. 1).

Regional comparison of upwelling events
In this section, we investigate upwelling events that were caused by strong winds across the entire study area in August 2016, leading to temperature and trace gas signals in almost all previously defined upwelling areas ( Fig. B1 and B2), which allows 310 us to compare the observations in these regions and to assess the importance of upwelling for the observed trace gas dynamics.
This case study exemplifies more general findings we gained during the analysis of the entire data set.
The entire month was characterised by strong westerly to south-westerly winds (Fig. 4), leading to upwelling in boxes Born, Ö-S, Ö-E, Go-SE, and GoF, interrupted by a week (15-22 August 2016) of more north-easterly winds, triggering upwelling in boxes Hiiu and Go-NW. The resulting SST drops by up to 16 • C predominantly near all southern and eastern coasts propagated 315 seawards and relaxed within several weeks (Supplement S1). Coverage of data from SOOP Finnmaid in this period is very dense, with the majority of transects along the east side of Gotland (see ratio of black and grey dashes in Fig. 4). We illustrate the temporal and spatial evolution of this event (Fig. 6) taking the example of box Ö-S.
Before the event, SST is at a typical summer value of 21 • C throughout the entire sub-transect. As expected, upwelling leads to SST decreases (displayed as peaks in Fig. 6a) with temperatures down to 9 • C. These minima move over time (see also 320 Supplement S1) and are subject to relaxation, which is further discussed in Sect. 3.3. The pre-upwelling temperature is usually not reached again during this time of the year, most probably due to an increased mixed-layer depth as an additional effect of stronger winds, and weakened solar irradiation at the end of August.
The observed trace gas patterns are similar to the temperature distribution: The first sub-transect can be considered as background conditions with typical late summer values of about 250 µatm for pCO 2 and 3.2 nmol L −1 for cCH 4 in this area. During 325 the upwelling event, we observe elevated pCO 2 and cCH 4 , with trace gas maxima correlated to minimum SST. For CO 2 , this results in a switch from undersaturation to supersaturation. CH 4 is always supersaturated or in equilibrium with the atmosphere in the SOOP Finnmaid data set and strong upwelling further increases this supersaturation and, eventually, CH 4 outgassing.
However, upwelling is not the only factor controlling increased CH 4 supersaturation (Fig. 6d). Warming of upwelled waters increases pCH 4 and, therefore, relative saturation. As with SST, the enhanced trace gas levels relax subsequentially.

330
To extend these findings to the different regions, we investigated the relationships between trace gas data and temperature over time (Fig. 7). The example from box Ö-S is representative of the majority of strong upwelling events affecting trace gases in the data set of SOOP Finnmaid, which are generally characterised by near-linear relationships between trace gases and SST. Maximum pCO 2 and cCH 4 values would not be reached without upwelling in summer (Fig. B4). We observe the same behaviour in boxes Go-SE, Ö-E, and Go-NW despite their reduced data coverage (Fig. 7). Box Ö-E features the highest 335 pCO 2 in the data set of over 800 µatm. In box Go-NW, the observable upwelling event began only at the end of trace gas data coverage on the western route due to a different favourable wind direction, hence, the more extreme values are missing in this example. However, the resulting pattern resembles the ones of boxes Ö-S, Ö-E, and Go-SE, just with lower maximum pCO 2 and cCH 4 .
Boxes Born and GoF show the same relationships as the previous boxes in their pCO 2 -SST diagram with considerable 340 dynamic range in case of box GoF (Fig. 7a,g). The respective cCH 4 -SST diagrams, however, only contain a small branch of increasing cCH 4 with decreasing SST (Fig. 7i,o). These regions are dominated by temperature-independent CH 4 variability, which indicates that other processes than upwelling might cause higher-than-usual cCH 4 : Box Born is situated between two basins (Arkona and Bornholm basin), which are interlinked via lateral transport, and which both feature gassy sediments (Gülzow et al., 2014;Tóth et al., 2014), from which CH 4 may be released via pressure changes caused by strong winds 345 (Schneider von Deimling et al., 2010;Gülzow et al., 2013). cCH 4 variability in box GoF might be driven by the highly variable physical conditions, e.g. changes of the estuarine circulation up to full reversal (Westerlund et al., 2019) or enhanced vertical transport by boundary wall shear (Schmale et al., 2010). These effects would lead to a less distinct impact of upwelling compared to boxes Ö-S, Ö-E, Go-SE, and Go-NW, where vertical decoupling is more stable.
The discussed phenomena can be contrasted with the behaviour of the sub-transect within box Hiiu: Although we find a clear 350 correlation between decreasing SST and increasing pCO 2 resembling that of the other boxes, we do not observe temperatures is reduced since SOOP Finnmaid uses two different routes around Gotland. Black dashed lines indicate atmospheric equilibrium partial pressure for CO2 and concentration for CH4 (calculated using mean salinity per box in the given time period), respectively. lower than 15 • C and pCO 2 higher than 420 µatm, which is close to atmospheric equilibrium (Fig. 7f). This relationship is similar to that of box openGo, the patterns in which we attribute to mixed-layer deepening and air-sea gas exchange caused by stronger winds instead of upwelling because of its distance to the coast (minimum 40 km, median 64 km, Table 1). Since the route of SOOP Finnmaid within box Hiiu is the furthest away from the coast of all boxes (minimum 27 km, median 355 43 km, Table 1) and the observed pCO 2 -SST relationships are so similar to openGo, we infer that upwelling has only minor influence on the observed values in this region during this time period. This is consistent with maps of modelled SST (Fig. 3 and Supplement S1, most pronounced around 18 August 2016), where no upwelled water masses reach out to the ship track, and confirms the same finding from the statistical identification of main upwelling areas presented in Sect. 3.1.
The cCH 4 -SST diagram of box Hiiu (Fig. 7n) resembles, to some extent, that of box GoF (Fig. 7o) without the upwelling 360 branch and smaller maximum cCH 4 , and indicates considerable cCH 4 variability compared to, e.g. boxes Ö-S, Ö-E, Go-SE, and Go-NW. In late July, for example, CH 4 concentration drops from 4.8 to 3.3 nmolL −1 at more or less constant temperature ( Fig. 7n), equating to a change in relative CH 4 saturation from 1.9 to 1.3. In the adjacent region openGo, no instances of increasing cCH 4 at constant SST (vertical branches in Fig. 7n-p) were observed.
We summarise that upwelling affects observed SST, pCO 2 , and cCH 4 drastically in the defined boxes in late summer of 365 2016. It typically causes near-linear relationships between surface trace gas signals and temperature with varying ranges and slopes between regions. For CO 2 , this can be observed in all regions (with limitations in box Hiiu), while in the case of CH 4 , strong variability caused by other processes may mask the effects of upwelling and closest-to-linear relationships are observed in boxes Ö-S, Ö-E, Go-SE, and Go-NW.

Typical relaxation of upwelling-induced trace gas signals 370
The surface water properties of a region influenced by upwelling change over the course of the upwelling event. This can be seen in Fig. 7, where the evolution of the relationship over time between SST and pCO 2 or cCH 4 is indicated by colour, and in Fig. B4. Before an event, we observe high temperature and low trace gas levels, with low spatial variability within sub-transects. During strong upwelling, a variety ::::: larger ::::::::: variability of SST, pCO 2 , and cCH 4 are : is observed concurrently as SOOP Finnmaid transects the respective region. These values usually form a near-linear relationship :::: Trace ::: gas :::: and :::: SST :::: data 375 :: are ::::::: usually :::::: related :::::: linearly ::::: after ::: the :::::::: upwelling ::::: event. After upwelling-favourable winds cease, the range of signals as well as their intensity is reduced through relaxation in a quasi-linear fashion (Fig. 7). The final state after relaxation, when compared to the initial state, is shifted towards lower temperatures, higher pCO 2 , and slightly elevated cCH 4 , which equals a roughly comparable CH 4 supersaturation at this decreased, final temperature. Depending on time of the year, SST might re-increase due to subsequent warming, or not recover completely (as in late summer).

380
In order to discuss the processes that are involved in the relaxation of upwelling signals, we calculated theoretical relaxation curves in trace gas -temperature diagrams (Fig. 8). Assumed endmember characteristics, physical driving parameters, and process descriptions are summarised in Sect. B1. We focus on air-sea gas exchange, air-sea heat exchange, and mixing with a typical water mass with pre-upwelling conditions. CH 4 oxidation in the upper, oxic water column should not play a major role on the short time scales considered here (Jakobs et al., 2013). Primary production (e.g. by nitrogen fixation) has the potential 385 to decrease pCO 2 distinctly, but is difficult to constrain since it depends on meteorological conditions and nutrient availability with possible time lags of several weeks (Vahtera et al., 2005;Wasmund et al., 2012).
The relaxation of SST is mainly driven by mixing. We estimated a total surface heat flux of ca. 300 J m −2 s −1 , which translates to a daily SST change of ca. 0.4 K d −1 assuming a mixed-layer depth of 15 m, which is rather typical for windy conditions in summer (derived from model data, not shown). Therefore, air-sea heat exchange contributes only little to the 390 observed warming of upwelled water masses : in ::: the ::::: order :: of ::::: 5-10 : K, leaving mixing as the dominant process. Despite the excess of the surrounding water masses, mixing does not necessarily lead to pre-upwelling conditions since the endmember Mixing also shapes the typically observed cCH 4 -SST relationships (Fig. 7i-o and 8a), leading to near-linear mixing curves 395 since concentration is a conservative parameter with respect to temperature changes (we neglect the influence on water volume :::::: density :::: here). The upwelled water mass releases ca. 7700 nmol m −2 d −1 of CH 4 into the atmosphere. This results in a daily cCH 4 loss of 0.51 nmol L −1 d −1 in a 15 m mixed layer, which is an efficient sink considering the magnitude of observed concentrations. Therefore, air-sea gas exchange alters the slope of the cCH 4 -SST relationship. Note, however, that gas flux is highly dependent on wind speeds, which are biased in cCH 4 -SST diagrams presented here: Pre-upwelling conditions 400 involve low wind speeds, while the upwelling event is caused by stronger winds, which eventually weaken. Heat exchange has no influence on cCH 4 , but the relative CH 4 saturation is determined by its partial pressure pCH 4 , which increases by the order of 2 % K −1 (Wiesenburg and Guinasso, 1979). This effect should not play a major role concerning relaxation given the low surface heat flux. However, as outlined above, SST might re-increase in the following weeks, thereby increasing pCH 4 and, thus, potentially lead to enhanced fluxes into the atmosphere :::: over : a :::::: longer :::: time ::::: period ::: of ::::: weeks :::::::: following ::: the ::::::::: upwelling :::: event.

405
Likewise, mixing leads to elevated pCH 4 and relative saturation compared to linear behaviour (Fig. 8b).
Similarly, the relaxation of pCO 2 cannot be considered independently from SST relaxation due to its temperature dependence. Warming by air-sea heat exchange causes a pCO 2 increase in the order of 4 % K −1 (Takahashi et al., 1993), which should not play a major role concerning relaxation given the low surface heat flux. As with pCH 4 , however, this effect could lead to increasing pCO 2 and enhanced CO 2 fluxes into the atmosphere (or reduced fluxes into the sea) in the following weeks.

410
The relaxation of upwelling-induced pCO 2 signals ( Fig. 7a-g) cannot be explained solely by mixing because the theoretical pCO 2 -SST mixing curve obtained from CO 2 system calculations features a distinct curvature with lower pCO 2 compared to linear behaviour (Fig. 8d). The observed near-linear relationship is likely caused by air-sea CO 2 exchange: At a wind speed of 10 m s −1 , the upwelled water mass in this example (Fig. 8d) releases ca. 0.074 mol m −2 d −1 of CO 2 into the atmosphere, which translates into a daily C T (total dissolved inorganic carbon) loss of 4.8 µmol kg −1 d −1 in a 15 m mixed layer. This 415 C T decrease in CO 2 -oversaturated waters explains the deviation from the expected linear (conservative) mixing curve in C T estimated from pCO 2 observations (Fig. B5 vs. 8c). Primary production triggered by upwelling has a similar (potentially even greater) influence on pCO 2 , but with different kinetics. One could argue that air-sea CO 2 exchange should similarly increase C T in CO 2 -undersaturated waters, which is not observed (Fig. B5). This can be explained with the aforementioned wind speed bias, resulting in very low fluxes under pre-upwelling conditions (see also Sect. 3.5). The bent C T -SST curve translates into 420 a near-linear pCO 2 -SST curve, which, in conclusion, can be interpreted as the combined result of mixing and decrease of the highest pCO 2 values due to gas exchange and possibly primary production.

Interannual variability of upwelling-induced trace gas signals
Since upwelling in the Baltic Sea is an episodic phenomenon based on wind conditions, it is subject to considerable interannual variability. In Fig. 9, we present seasonal plots (from May to September) of pCO 2 and cCH 4 versus SST in box Ö-S. The coloured date scale allows to follow the temporal evolution of signals throughout the season and also highlights larger data 445 gaps (e.g. in 2012 and 2013).
Most years feature consistent patterns with respect to pCO 2 (Fig. 9), reflecting its yearly cycle (see Introduction and Schneider and Müller, 2018): CO 2 is already undersaturated with respect to the atmosphere in May due to primary production during the spring bloom. Over the following weeks, the change in pCO 2 is usually rather small, but SST increases as a result of solar irradiation and often weaker winds (see also Fig. B4). The resulting stabilisation of the surface thermocline and the accumu-450 lation of remineralised CO 2 below combined with decreasing air-sea CO 2 exchange and ongoing primary production lead to increasing pCO 2 gradients between surface and sub-thermocline water (which are the cause of upwelling-induced pCO 2 signals, Fig. 2) and a permanent undersaturation of the surface water with respect to the atmosphere. The characteristic cCH 4 -SST conditions follow the CH 4 saturation ::::::::: equilibrium curve towards lower concentrations at higher temperatures most probably due to air-sea CH 4 exchange, maintaining a persistent supersaturation. In most years, most notably in 2010, 2012, 2014, and 2016, 455 strong upwelling around August overrides these typical summer conditions, resulting in characteristic pCO 2 -SST and cCH 4 -SST patterns. For these years, the ranges of SST, pCO 2 , and, to a certain extent, cCH 4 are similar (but still not equal), with the notable exception of very dynamic cCH 4 in 2010. For the other years, we observe a high degree of variability from these typical conditions: Strong upwelling-favourable winds in June 2011 led to an early increase of pCO 2 and lower SST overall. Later, at the end of July 2011, a pronounced, sharp increase of cCH 4 at rather constant temperature was observed, which clearly is not 460 related to upwelling.
The year 2015 is particularly interesting because it demonstrates the influence of quasi-continuous upwelling-favourable winds over the course of several months, which overrides the typical summer trace gas situation. The year was dominated by upwelling-favourable, westerly winds until the beginning of August (Fig. A4), effectively prohibiting strong thermal stratification of the surface water ( Fig. 10a and low maximum temperature in Fig. 9f,n). This special case is problematic for the 465 detection method because the observable ∆SST gradients become too small for every day to be counted as an "upwelling day" (Fig. A4). As a result of weakened stratification, surface CO 2 undersaturation is unusually weak compared to the typical summer situation (Fig. 10b and high minimum pCO 2 in Fig. 9f). Furthermore, we observe reduced cCH 4 variability as a result of continuous mixing and intensified air-sea exchange through increased turbulence, so that cCH 4 follows the saturation ::::::::: equilibrium : curve more closely than during most years (Fig. 9n). Elevated cCH 4 (Fig. 10c) does not necessarily translate into 470 elevated saturation (Fig. 10d) depending on SST -however, as pointed out in Sect. 3.3, the water mass will become supersaturated as a consequence of subsequent warming. In July 2015 (turquoise hues in Fig. 9f,n), near-linear trace gas -temperature curves are characteristic for strong upwelling (see Sect. 3.2). Compared to, e.g. August 2014 and 2016, however, where the upwelling SST, pCO 2 , and cCH 4 signals stand out prominently from the rest of the values, their range concerning all three parameters is reduced in 2015 since decoupling of surface and underlying water was partly impeded.

Potential to estimate upwelling-induced air-sea trace gas fluxes
The observed near-linear relationships between pCO 2 or cCH 4 and SST can be used to spatially extrapolate trace gas obser-490 vations from sub-transects based on modelled SST fields, assuming that these relationships are consistent for entire upwelling areas. This enables us to estimate air-sea CO 2 and CH 4 fluxes resulting from upwelling events (Fig. 11, ::::: Sect. ::: B1). Since we used linear regression (Fig. 11f,g), SST minima near the coast translate into pCO 2 and cCH 4 maxima, retaining the overall pattern and fine structure of the SST field (Fig. 11a-c).
The CO 2 flux (FCO 2 , Fig. 11d) depends on the difference in pCO 2 between sea and air, which determines the flux direction, 495 and the transfer coefficient, which is parametrised mostly by wind speed. Therefore, pCO 2 and FCO 2 share the same spatial pattern with positive and negative fluxes being present in the box at the same time. Both CO 2 outgassing (0.13 mol m −2 d −1 ) and uptake (−0.046 mol m −2 d −1 ) peak on 9 August, when wind speeds are highest. CH 4 fluxes (FCH 4 , Fig. 11e) into the atmosphere reach their maximum (5730 nmol m −2 d −1 ) on the same day. However, the spatial distribution of FCH 4 differs from that of cCH 4 due to the temperature dependence of the CH 4 saturation :::::::::: equilibrium concentration: For instance, the low-500 est fluxes on 10 August occur close to the coast despite high concentrations in this area because supersaturation decreases with lower SST, indicating that the upwelled water mass has a lower CH 4 supersaturation than the surrounding waters in this example. However, relative CH 4 saturation is highly sensitive to changes in slope of the applied regression curve, which underestimates the observed maximum supersaturation in this example (Fig. 11g) and likely leads to this special spatial distribution.
Even careful tweaking of the regression curve would result in a pattern much more similar to that of FCO 2 and this similarity 505 increases with increasing supersaturation of the upwelled water mass compared to pre-upwelling conditions. In other examples, the two flux patterns are more similar than here (data not shown). The discussed "tweaking" of the cCH 4 -SST regression, though impacting the derived pattern considerably, would have only a small impact on the areal flux. In any case, daily FCH 4 peak under intermediate conditions where the flux-increasing effects of rising cCH 4 and rising SST combine. On many days in this example, including 10 August (Fig. 11e), this area of maximum daily FCH 4 is in close proximity to the transect of SOOP 510 Finnmaid, where it can be observed by in situ measurements, whereas areas of maximum daily FCO 2 tend to be closer to the coast. It should be noted that the spatial variability of FCO 2 is much higher than that of FCH 4 (see maximum and minimum values in Fig. 11i,j).
The importance of wind is also reflected by the evolution of air-sea gas fluxes over time (Fig. 11i,j). Fluxes are weak under pre-upwelling conditions due to low wind speeds (Fig. 11h), increase distinctly with rising wind speeds, and reach a minimum 515 (considering absolute values for FCO 2 ) in the relaxation period after the upwelling event, when the wind calms down again.
The sea is a permanent source of atmospheric CH 4 with varying strength based mostly on wind speed in this example, which can be generalised to the entire data set. In contrast, FCO 2 is negative (CO 2 uptake from the atmosphere) under pre-upwelling conditions. CO 2 outgassing starts with the onset of upwelling, but at this point, the area is still dominated by increasing CO 2 uptake due to rising wind speeds. Both positive and negative CO 2 fluxes intensify over the following days, but CO 2 outgassing 520 reaches higher absolute values as a result of high pCO 2 due to upwelling. The area is a net source of CO 2 for the atmosphere (mean FCO 2 > 0) from 9 August, when wind speeds are maximal, to 19 August, when pCO 2 gradients have sufficiently diminished due to relaxation. Note, however, that mean FCO 2 depends on the (arbitrary) choice of box boundaries.
The presented flux estimates depend largely on the applicability of the observed near-linear trace gas -SST relationships for the entire area and period, including extrapolation to temperatures lower than the minimum temperature of the trace gas -

Conclusions
Upwelling in the Baltic Sea can be observed using autonomous measurements aboard SOOP, which, compared to dedicated research cruises, provide higher spatial and temporal coverage at the cost of being restricted to surface data and, depending 540 on route, a higher ::::: larger distance to the coast. They enable studies on seasonality, comparison of regions, and observation of processes over long time periods. Combining SOOP-based trace gas measurements with other high-resolution data sets like model or remote sensing data further allows us to a) assess their spatial and temporal representativity by adding information perpendicular to :::::: beyond the ferry track, b) assess the prevalence of upwelling even during SOOP data gaps caused by ship schedule and (rare) outages, c) compare events by size, duration, and signal intensity, and d) estimate upwelling-induced 545 air-sea fluxes.
29 Figure A3. As Fig. 4, but based on model-SST data from the entire box instead of along the sub-transect of SOOP Finnmaid. Please refer to Sect. 2.3 for a comparison to Fig. 4. Both approaches are further discussed in Sect. 3.1. Figure A4. As Fig. 4, but within box Ö-S in 2015 (here, both routes go through the box). Quasi-persistent upwelling-favourable wind conditions until the beginning of August effectively prohibited strong thermal stratification of the surface water. This results in lower possible SST gradients from upwelling and, therefore, a rather unreliable ∆SST criterion, which is only triggered during the most intense periods.
Please refer to Sect. 3.4 for a detailed discussion of this event.