2.1 Study site and sampling

This study is based on samples collected on a research cruise (no. LP20190515) on RV Ludwig Prandtl in the Dutch Wadden Sea (Frisian Islands) in May 2019 (Fig. 1). We collected water samples in the Wadden Sea starting at Harlingen and traveling along the following path: through the Vlie Inlet along the islands of Vlieland and Terschelling, through the Ameland Inlet to Ameland Island, via the Frisian Inlet to Lauwersoog, and around Schiermonnikoog Island via the Ems-Dollard Inlet to Emden. In addition, we sampled half a tidal cycle during ebb tide (from high tide to low tide) on 21 May 2019 (Table B2). To set the range of ebb tide data, we also sampled half a tidal cycle during flood tide (from low tide to high tide) on 23 May 2019 for comparison. Both of the half tidal cycles were sampled each day using an anchor station in the waterway on the western side of Ameland in the Ameland Inlet.

Nearly every 30 min, we collected discrete surface (1.2 m depth) water samples with a bypass from the onboard flow-through FerryBox system, which also provided essential physical parameters, such as salinity, with an accuracy of 0.02 PSU (practical salinity units) and temperature with an accuracy of 0.1 °C (Petersen et al., 2011). The FerryBox was cleaned and checked prior to the cruise, and salinity was occasionally checked using discrete samples; this workflow is considered sufficient for gradients in near-shore investigations (Yoana Voynova, personal communication, 2024). We complemented our salinity and temperature data with data from three Rijkswaterstaat stations (Dantziggat, Terschelling 10, and Vliestroom; Table B3), which were close to our stations.

For TA and dissolved inorganic carbon (DIC) measurements, we sampled water with overflow into 300 mL BOD (biological oxygen demand) bottles and preserved it with 300 µL of saturated mercury chloride solution (HgCl₂) to stop biological activity. Each BOD bottle was filled without air bubbles and closed using a ground-glass stopper coated in Apiezon^® M grease and a plastic cap. The samples were stored in a cool, dark environment prior to analysis in the lab.

Water for nutrient samples was filtered through pre-combusted (4 h at 450 °C) GF/F filters, and the filtrate was stored frozen in three 15 mL Falcon^® tubes for triplicate measurements in the lab.

To determine the total carbon (C), organic carbon (C_org) and nitrogen (N) concentrations in SPM, and associated C_org:N ratios, we used pre-combusted (4 h at 450 °C) GF/F filters; the filters were dried after sampling at 50 °C to remove all humidity and then stored frozen until measurement.

2.2 Carbon species analyses

The parallel analyses of TA and DIC were carried out in March 2020 using the MARIANDA (MARIne ANalytics and Data) VINDTA 3C (Versatile INstrument for the Determination of Total dissolved inorganic carbon and Alkalinity) instrument, which measures TA (via potentiometric titration) and DIC (via coulometric titration), both with a measurement precision of <2 µmol kg⁻¹ (Shadwick et al., 2011). Certified reference material (CRM batch no. 187), provided by Andrew G. Dickson (Scripps Institution of Oceanography), was measured before and after the samples and used to ensure a consistent calibration of both measurements.

The calcite and aragonite saturation states (Ω), the pH, and the seawater partial pressure of CO₂ (pCO₂) were computed with the CO2SYS program (Lewis and Wallace, 1998), using the measured parameters TA and DIC, salinity, temperature, silicate, and phosphate as input variables as well as the dissociation constants from Mehrbach et al. (1973), as refitted by Dickson and Millero (1987). Reported calculation uncertainties are ±0.0062 for pH (Millero et al., 1993), ±4.9 % for the aragonite saturation state, and ±3.5 % for pCO₂ (Orr et al., 2018).

2.3 Nutrient analyses

The nutrients were measured with a continuous-flow automated nutrient analyzer (AA3, SEAL Analytical) and a standard colorimetric technique (Hansen and Koroleff, 2007) for nitrate (${{\mathrm{NO}}_{\mathrm{3}}}^{-}$), nitrite (${{\mathrm{NO}}_{\mathrm{2}}}^{-}$), phosphate (${{\mathrm{PO}}_{\mathrm{4}}}^{\mathrm{3}-}$), and silicate (Si), whereas a fluorometric method (Kérouel and Aminot, 1997) was employed for ammonium (${{\mathrm{NH}}_{\mathrm{4}}}^{+}$) (Grasshoff et al., 2009). The nutrient samples were measured against Eurofins Scientific reference materials VKI SW4.1B (for NO_x, NO₂, and NH₄) and VKI SW4.2B (for Si and PO₄) in July 2019. The maximum standard deviations were 0.322 µmol L⁻¹ for ${{\mathrm{NO}}_{\mathrm{3}}}^{-}$, 0.014 µmol L⁻¹ for ${{\mathrm{NO}}_{\mathrm{2}}}^{-}$, 0.081 µmol L⁻¹ for ${{\mathrm{NH}}_{\mathrm{4}}}^{+}$, 0.014 µmol L⁻¹ for ${{\mathrm{PO}}_{\mathrm{4}}}^{\mathrm{3}-}$, and 0.165 µmol L⁻¹ for Si.

For C_org determination, filters were acidified with 1N HCl and dried overnight to remove all inorganic carbon content. Filters were measured with a CHN (carbon–hydrogen–nitrogen) elemental analyzer (Eurovector EA 3000, HEKAtech GmbH) at the Institute of Geology, University of Hamburg, and calibrated against a certified acetanilide standard (IVA Analysentechnik). The standard deviations were 0.05 % for carbon and 0.005 % for nitrogen.

2.4 Data analyses

The data analyses were performed using RStudio (Version 1.3.1073, © 2009–2020 RStudio, Posit PBC; R Core Team, 2019). The linear regression Model II was performed using the “lmodel2” R package (Legendre, 2018), and the plots were created with the “ggplot2” R package (Wickham, 2016).

3.1 Spatial parameter distribution

To investigate the spatial distribution of TA in the Dutch Wadden Sea and compare its general status with earlier studies (in particular Hoppema, 1990), we observed TA and related parameters in surface water along a transect from the coastal mainland towards the North Sea.

The temperature varied between 12 and 16 °C, with higher temperatures towards the coastal mainland (Fig. 2a). We identified two main subregions based on the salinity values: (1) the Ems-Dollard Inlet, which showed salinity values lower than 28 PSU and had a minimum value of 20.25 PSU at the most upstream station, and (2) the area around Ameland Island and the remainder of our investigated region in the Dutch Wadden Sea, which showed salinity values from 28 to 33 PSU with little variation (Fig. 2b).

Figure 2Spatial distribution of (a) temperature (°C), (b) salinity (PSU), and (c) total alkalinity (TA; µmol kg⁻¹) from surface water samples in May 2019.

The spatial transect of TA concentrations ranged from 2332 to 2517 µmol TA kg⁻¹. We observed lower concentrations on the North Sea side of the Frisian Islands, with somewhat higher concentrations around Ameland (Fig. 2c). In contrast to the North Sea side, the values were higher (>2380 µmol TA kg⁻¹) in the Wadden Sea. In the Ems-Dollard Inlet, the concentrations were even higher, with values of up to 2517 µmol TA kg⁻¹ at the most upstream station.

Silicate (Si) showed higher concentrations in the Wadden Sea and lower values towards the North Sea (Fig. A1a). The highest concentrations were observed at the coastal mainland and in the Ems-Dollard Inlet. Silicate concentrations ranged between 0.3 and 56.3 µmol Si L⁻¹. Both the calcite and aragonite saturation states (Ω) were supersaturated throughout the study region. The Ω values ranged from 2.3 to 4.6 for calcite (Fig. A1b) and from 1.4 to 2.8 for aragonite (Table B3). The highest values were observed on the North Sea side of the barrier islands, whereas the lowest values were found near Harlingen and in the Ems-Dollard Inlet. Like the calcite and aragonite saturation states, the pH values were higher in the North Sea, whereas they were lower in the Wadden Sea and near the coastal mainland (Fig. A1c). The pH values ranged from 7.86 to 8.19, with the lowest readings observed near Harlingen and in the Ems-Dollard Inlet. The nitrate (${{\mathrm{NO}}_{\mathrm{3}}}^{-}$) concentrations fell within in a low range (<3 $\mathrm{\mu }\mathrm{mol}{{\mathrm{NO}}_{\mathrm{3}}}^{-}{\mathrm{L}}^{-\mathrm{1}}$) throughout the study region. Higher concentrations (<6 $\mathrm{\mu }\mathrm{mol}{{\mathrm{NO}}_{\mathrm{3}}}^{-}{\mathrm{L}}^{-\mathrm{1}}$) were only observed at a few stations close to land, and the maximum concentrations (<38 $\mathrm{\mu }\mathrm{mol}{{\mathrm{NO}}_{\mathrm{3}}}^{-}{\mathrm{L}}^{-\mathrm{1}}$) were observed in the Ems-Dollard Inlet (Fig. A1d). DIC concentrations ranged from 2097 to 2430 µmol DIC kg⁻¹ (Fig. A1e). DIC values showed a similar pattern to TA values: higher concentrations near the coastal mainland and in the Ems-Dollard Inlet and decreasing concentrations toward the North Sea, where DIC reached minimum values.

Compared with the other transects of this study region, the strong influence of the inner Ems Estuary is visible at the most upstream station in the Ems-Dollard Inlet, showing the lowest salinity values; the lowest pH and calcite saturation state values; and the highest TA, DIC, nitrate, silicate, and phosphate values. The outer side of the Vlie Inlet reflects the North Sea conditions: lower temperatures and higher salinities. The North Sea impact is also visible in the mixing plot between TA and salinity (Fig. 3). Statistically significant linear mixing behavior was observed in the transect through the Ems-Dollard Inlet (R²=0.81) and through the Vlie Inlet (R²=0.77), where TA concentrations decreased with increasing salinities from the mainland towards the North Sea (Fig. 3). Whereas in the Ems-Dollard Inlet mixing is dominated by riverine water with high TA concentrations, the mixing in the Vlie Inlet showed a more prominent mixing of Wadden Sea and North Sea water. The TA concentrations in the Vlie Inlet and around Ameland, both on the North Sea side (Ameland NS) and the Wadden Sea side (Ameland WS), were higher than the TA concentration computed for the salinity end-member in the Ems-Dollard Inlet, suggesting the Dutch Wadden Sea as a source of TA (Fig. 3). Both Ameland NS and Ameland WS data clearly indicated nonconservative behavior with a range of TA concentrations at near-constant salinities.

Figure 3Mixing plot of total alkalinity (TA) and salinity (PSU) on the North Sea side of Ameland and the Frisian Inlet (Ameland NS), on the Wadden Sea side of Ameland (Ameland WS), around Schiermonnikoog and in the Ems-Dollard Inlet (Ems-Dollard), and in the Vlie Inlet (Vlie).

Download

3.2 Tidal cycle

In order to (1) identify potential TA sources and (2) to quantify potential TA export to the North Sea, we observed half a tidal cycle at an anchor station in the Ameland Inlet during ebb tide. We identified patterns in several biogeochemical parameters in water leaving the tidal flats (Fig. 4, Table B1). Temperature increased from 13.25 to 14.7 °C (Fig. 4a). Salinity was constant at around 32.5 PSU (Fig. 4b), which is within the range of coastal southern North Sea water excluding admixture of local freshwater sources.

Figure 4Half a tidal cycle (from high tide to low tide) showing the temporal distribution of (a) temperature, (b) salinity, (c) total alkalinity (TA), (d) dissolved inorganic carbon (DIC), (e) nitrate, (f) silicate, (g) ammonium, (h) calcite (upper points) and aragonite (lower points) saturation states, (i) pCO₂, (j) pH, (k) the C_org:N ratio of SPM, and (l) the DIC : TA ratio. Note the different y axes and the +2 h time difference between the local summer time and UTC time.

Download

During ebb tide, TA ranged from 2387 µmol TA kg⁻¹ during high tide to 2438 µmol TA kg⁻¹ during low tide (Fig. 4c). We observed an increase of 51.6 µmol TA kg⁻¹ (ΔTA) during ebb tide (6.8 h), resulting in a TA increase of 7.6 $\mathrm{\mu }\mathrm{mol}\mathrm{TA}{\mathrm{kg}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$ at the sampling location.

DIC concentrations behaved similarly to TA, with minimum values at high tide (2172 µmol DIC kg⁻¹) and maximum values (2273 µmol DIC kg⁻¹) at low tide, resulting in an increase of 101.3 µmol DIC kg⁻¹ (ΔDIC), or 14.9 $\mathrm{\mu }\mathrm{mol}\mathrm{DIC}{\mathrm{kg}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$ (Fig. 4d). DIC increased almost twice as much as TA.

Nitrate increased by 0.92 $\mathrm{\mu }\mathrm{mol}{{\mathrm{NO}}_{\mathrm{3}}}^{-}{\mathrm{L}}^{-\mathrm{1}}$ (Δ${{\mathrm{NO}}_{\mathrm{3}}}^{-}$) during ebb tide, from a minimum of 1.26 $\mathrm{\mu }\mathrm{mol}{{\mathrm{NO}}_{\mathrm{3}}}^{-}{\mathrm{L}}^{-\mathrm{1}}$ to a maximum of 2.17 $\mathrm{\mu }\mathrm{mol}{{\mathrm{NO}}_{\mathrm{3}}}^{-}{\mathrm{L}}^{-\mathrm{1}}$ (Fig. 4e), resulting in a nitrate increase of 0.13 $\mathrm{\mu }\mathrm{mol}{{\mathrm{NO}}_{\mathrm{3}}}^{-}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$.

Silicate showed a similar pattern, with low values (1.8 µmol Si L⁻¹) at high tide that increased during ebb tide to a maximum of 11.2 µmol Si L⁻¹, resulting in a silicate increase (ΔSi) of 9.4 µmol Si L⁻¹, or 1.4 $\mathrm{\mu }\mathrm{mol}\mathrm{Si}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$, during ebb tide (Fig. 4f).

Ammonium increased from 3.47 to 6.22 $\mathrm{\mu }\mathrm{mol}{{\mathrm{NH}}_{\mathrm{4}}}^{+}{\mathrm{L}}^{-\mathrm{1}}$ during ebb tide (Fig. 4g), resulting in an ammonium increase (Δ${{\mathrm{NH}}_{\mathrm{4}}}^{+}$) of 2.74 $\mathrm{\mu }\mathrm{mol}{{\mathrm{NH}}_{\mathrm{4}}}^{+}{\mathrm{L}}^{-\mathrm{1}}$, or 0.4 $\mathrm{\mu }\mathrm{mol}{{\mathrm{NH}}_{\mathrm{4}}}^{+}{\mathrm{L}}^{-\mathrm{1}}{\mathrm{h}}^{-\mathrm{1}}$.

The calcite and aragonite saturation states had maximum values (Ω_Ca=3.8, Ω_Ar=2.4) at high tide and decreased to their minimum (Ω_Ca=3.1, Ω_Ar=2.0) during ebb tide (Fig. 4h). The influence of the North Sea is indicated by the observed maximum at high tide, which decreased during the ebb.

The seawater pCO₂ had minimum values at high tide (385.1 µatm) and increased up to 576.6 µatm during low tide (Fig. 4i).

Like Ω, the maximum pH was 8.07 at high tide and decreased to a minimum (7.93) during ebb tide (Fig. 4j).

The C_org:N ratios of SPM increased during ebb tide (Fig. 4k). A minimum C_org:N ratio of 5.6 was observed around high tide and increased to a maximum of 13.0 during ebb tide. Simultaneously, the SPM concentration increased during ebb tide, from 12.8 mg SPM L⁻¹ to a maximum of 82.4 mg SPM L⁻¹ at the second-last station (Table B1).

3.3 TA generation

Tidal forcing leads to an exchange between Wadden Sea and North Sea water almost twice a day. The tidal forcing also induces a strong benthic–pelagic coupling (Huettel et al., 2003; Røy et al., 2008). Many studies support the fact that outflowing water exports material from the sediment, including remineralization products from organic matter (e.g., Billerbeck et al., 2006; Røy et al., 2008). Here, we focus on the hypothesis that the sediments are a significant source of TA.

For an initial rough estimate of maximum TA export during ebb tide, we used the mean observed TA increase ($\mathrm{\Delta }\text{TA}/\mathrm{2}$) of 25.8 µmol TA kg⁻¹ during ebb tide (in the Ameland Inlet, part of the Borndiep tidal basin), a tidal prism of 478×10⁶ m³ of the Borndiep tidal basin, and a share of intertidal flats of 53 % (Louters and Gerritsen, 1994). Assuming that only the intertidal sediments exchange TA, we estimated a TA export of 6.6 Mmol TA per tide to the North Sea. Assuming two ebb tides and a lunar cycle of 24.8 h, this would result in a daily export of 12.7 Mmol TA.

The significant correlation of TA and silicate (R²=0.93) as well as the insignificant relation between TA and salinity (R²=0.32) and between silicate and salinity (R²=0.21) suggest that TA originates from the tidal flats in this part of the Dutch Wadden Sea, not from admixture carried by river runoff. The significant correlation between TA and silicate during ebb tide point to the same source (Fig. 5b).

Figure 5Correlations of TA with (a) dissolved inorganic carbon (DIC), (b) silicate, (c) nitrate, and (d) ammonium during ebb tide in the Ameland Inlet.

Download

To further elucidate potential TA sources in the Dutch Wadden Sea, we correlated TA with DIC, silicate, nitrate, and ammonium, respectively, in half a tidal cycle, from high tide to low tide (Fig. 5).

The correlation between TA and DIC is a measure between anaerobic and aerobic processes. Our data show a strong positive correlation between DIC and TA (R²=0.93), with TA concentrations being higher than DIC concentrations (Fig. 5a). We observed a release of excess DIC compared with TA, as indicated by the slope of 1.89 and by an increase in DIC (ΔDIC=101.3 µmol kg⁻¹) that is almost twice as high as TA (ΔTA=51.6 µmol kg⁻¹) (Fig. 5a). This excess DIC may be caused by strong CO₂ production due to high aerobic OM degradation, which can be supported by the seawater pCO₂ being supersaturated with respect to the atmosphere (Fig. 4i). The TA increase can be fueled by various processes, which we will discuss below. We detected a positive linear correlation of increasing TA and silicate (R²=0.93) during ebb tide, supporting pore water outflow (Fig. 5b), as pore water is the major Si source during summer (van Bennekom et al., 1974). A stronger influence of pore water with the ongoing ebb tide is indicated by increasing values. The positive correlation between nitrate and TA (R²=0.67) (Fig. 5c) was less strong than the correlations between TA and DIC and between TA and Si, which could be traced back to an effect of the first four sampling points that were probably at the tipping point from high tide to low tide. In the remaining samples, the increasing nitrate and TA concentrations suggest stronger TA generation than nitrate production, balancing TA that may be consumed by nitrification (i.e., nitrate production).

4.1 TA spatial variability

Hoppema (1990) reported TA distributions in the westernmost part of the Dutch Wadden Sea, around the barrier islands of Texel, Vlieland, and Terschelling. The aforementioned work focused on the tidal basins drained by the tidal inlets of Marsdiep and Vlie, located more to the west than our sampling stations (not visible on the map). Hoppema (1990) did not observe a continuous increase in salinity in the Wadden Sea from the freshwater source towards the North Sea and associated this with the influence of tidal differences and an arbitrary sampling scheme. The presence (dominance) of North Sea water in the Dutch Wadden Sea and on the tidal flats is supported by our transect data, which show relatively high salinities at the coastal North Sea level. Brackish salinity values were only detected in the Ems-Dollard Inlet, which receives freshwater from the river Ems, and close to Harlingen and Lauwersoog, which have direct freshwater inflows from smaller rivers and streams. The absence of clear salinity gradients in the more eastern part of the Dutch Wadden Sea investigated in our study suggests that most of the IJsselmeer discharge was exchanged with the North Sea through the Marsdiep Inlet (e.g., Duran-Matute et al., 2014).

The spatial TA data from Hoppema (1990) show lower TA concentrations at stations with more freshwater influence and higher TA concentrations in the tidal inlets. The data from this study also show high TA concentrations in the tidal inlets, suggesting TA generation in sediments, which is possibly fueled by high nutrient and OM imports (van Beusekom and De Jonge, 2002). The even higher TA concentrations at stations with lower salinities close to the mainland observed in this study also show the influence of the catchment area on the coast and, possibly, TA generation in the shallow sediments near the coast. In May 1986, Hoppema (1990) found TA concentrations ranging between 2319 and 2444 µmol TA kg⁻¹ at salinities between 18.62 and 29.17 PSU. Our lowest observed TA concentration was 2332 µmol TA kg⁻¹ at a salinity of 32.14 PSU, whereas our highest TA concentration was 2517 µmol TA kg⁻¹ at a salinity of 20.25 PSU close to the coastal mainland. A comparison of both studies shows that the general TA levels are in a similar range but that the spatial gradients are opposite.

Conservative mixing between TA and salinity is only visible in the Ems-Dollard Inlet and the Vlie Inlet (Fig. 3). While conservative mixing in the Ems-Dollard Inlet is more dominated by the freshwater discharge from the river Ems, the conservative mixing in the Vlie Inlet is more dominated by North Sea water passing through this deep inlet and allowing more North Sea water to be transported towards the coast. After the Marsdiep Inlet, the Vlie Inlet has the highest average tidal prism and is the second-largest inlet in the Dutch Wadden Sea (Elias et al., 2012). Similar to our findings, Hoppema (1990) noted linear mixing of TA and salinity in the Vlie Inlet and also suspected a lower freshwater contribution in the inlet, which is in accordance with model data (Duran-Matute et al., 2014).

In the Ems-Dollard Inlet, conservative mixing was observed, indicating minor contributions from other sources. In a previous study, Norbisrath et al. (2023) observed very high TA concentrations and TA generation in the upper tidal river of the highly turbid Ems Estuary, which may explain the high levels of TA in the Ems-Dollard Inlet (at low salinities) observed in this study.

Hoppema (1990) also observed a range of TA concentrations in the Dutch Wadden Sea and related these to different sinks and sources. TA sinks can be calcium carbonate (CaCO₃) precipitation or the extraction of seawater carbonate by mollusks (e.g., Chen and Wang, 1999; Hoppema, 1990). Variable freshwater inflows can either serve as a sink or a source (e.g., Chen and Wang, 1999; Hoppema, 1990). Other TA sources can be CaCO₃ dissolution, anaerobic metabolic processes in the sediment, or the erosion of sediments that enhance TA (e.g., Hoppema, 1990; Chen and Wang, 1999).

Except for the Ems-Dollard Inlet and close to Harlingen, we mainly observed marine salinities >30 PSU but found higher TA values in the Dutch Wadden Sea than in the North Sea. Therefore, we exclude possible TA sinks and focus only on TA sources. According to Hoppema (1990), the main causes of TA variation in the Dutch Wadden Sea were freshwater inflows and sources in the sediment. In our study, freshwater inflows with high TA concentrations were only observed in the Ems-Dollard Inlet (but not around the islands and the tidal flats). For a further TA source identification in the Dutch Wadden Sea, we investigated the TA variability during ebb tide in a tidal channel close to Ameland.

4.2 Determination of TA generation

Burt et al. (2016) and Schwichtenberg et al. (2020) have indicated that TA generation in the Wadden Sea is an important source with respect to the North Sea's carbon storage capacity. Here, we want to further identify TA generation and potential TA sources.

In a study from the late 1980s, Hoppema (1993) observed a tidal cycle in the Marsdiep Inlet in May and September. Focusing on TA, DIC, and oxygen, this aforementioned work also observed increasing TA values during ebb tide and assumed that the tidal flats and discharging rivers and canals were TA sources. Comparing our present TA data to the historical TA data, there is not a large difference in the range of values observed during a tidal cycle. However, further in-depth interpretation and comparison of both TA data sets is limited by the low number of data, leading us to focus on TA generation during our cruise.

We made a very rough first estimate of the daily TA export. Using a 3D ecosystem model, Schwichtenberg et al. (2020) estimated an annual export of 10–14 Gmol TA yr⁻¹ for the entire Dutch Wadden Sea. Given that the Borndiep tidal basin covers about 14 % of the Dutch Wadden Sea and assuming no seasonal dynamics, our estimate of 12.7 Mmol TA d⁻¹ compares well with the annual average model estimate of 4.6 Mmol TA d⁻¹, but the overestimation suggests that seasonal dynamics may be involved. As our TA export was based on only a half-tidal-cycle observation, the inclusion of it into the model used by Schwichtenberg et al. (2020) would be unreliable (Johannes Pätsch, personal communication, 2022). To test whether the observed TA generation matches their suggested TA export, observational data of at least each season are required to run the model and gain a representative result (Johannes Pätsch, personal communication, 2022).

4.3 TA source attribution