Norwegian Sea net community production estimated from O 2 1 and prototype CO 2 optode measurements on a Seaglider 2

. We report on a pilot study using a CO 2 optode deployed on a Seaglider in the Norwegian Sea from 15 March to October 2014. The optode measurements required drift- and lag-correction and in situ calibration using 16 discrete water samples collected in the vicinity. We found that the optode signal correlated better with the 17 concentration of CO 2 , c (CO 2 ), than with its partial pressure, p (CO 2 ). Using the calibrated c (CO 2 ) and a regional 18 parameterisation of total alkalinity ( A T ) as a function of temperature and salinity, we calculated total dissolved 19 inorganic carbon content, c (DIC), which had a standard deviation of 11 µmol kg -1 compared with in situ 20 measurements. The glider was also equipped with an oxygen (O 2 ) optode. The O 2 optode was drift-corrected and 21 calibrated using a c (O 2 ) climatology for deep samples. The calibrated data enabled the calculation of DIC- and 22 O 2 -based net community production, N (DIC) and N (O 2 ). To derive N , DIC and O 2 inventory changes over time 23 were combined with estimates of air-sea as lower (O 2 ) and c raw (Chl a ) but higher N (DIC). Our results show the potential of glider data to 35 simultaneously capture time and depth-resolved variability in DIC and O 2 concentrations.


37
Climate models project an increase in the atmospheric CO2 mole fraction driven by anthropogenic emissions 38 from a preindustrial value of 280 µmol mol -1 (Neftel et al., 1982) to 538-936 µmol mol -1 by 2100 (Pachauri and 39 Reisinger, 2007). The ocean is known to be a major CO2 sink (Sabine et al., 2004;Le Quéré et al., 2009;Sutton 40 et al., 2014); in fact, it has taken up approximately 25 % of this anthropogenic CO2 with a rate of (2.5±0.6) Gt a -1 41 (in C equivalents) (Friedlingstein et al., 2019). This uptake alters the carbonate system of seawater and is causing 42 a decrease in seawater pH, a process known as ocean acidification (Gattuso and Hansson, 2011). The processes 43 affecting the marine carbonate system include air-sea gas exchange, photosynthesis and respiration, advection 44 and vertical mixing, and CaCO3 formation and dissolution. For that reason, it is important to develop precise, 45 accurate and cost-effective tools to observe CO2 trends, variability and related processes in the ocean. Provided 46 that suitable sensors are available, autonomous ocean glider measurements may help resolve these processes.

47
To quantify the marine carbonate system, four variables are commonly measured: total dissolved inorganic horizontal, but good long-term temporal resolution (Hemsley, 2003). In contrast, ship-based surveys have higher 54 vertical and spatial resolution than moorings but limited repetition frequency because of the expense of ship 55 operations. Ocean gliders have the potential to replace some ship surveys because they are much cheaper to 56 operate and will increase our coastal and regional observational capacity. However, the slow glider speed of 1-2 57 km h -1 only allows a smaller spatial coverage than ship surveys and the sensors require careful calibration to 58 match the quality of data provided by ship-based sampling.       The last oxygen optode calibration before the deployment was performed in 2012 as a two-point calibration at 163 9.91 °C in air-saturated water and at 20.37 °C in anoxic Na2SO3 solution. Oxygen optodes are known to be 164 affected by drift (Bittig et al., 2015), which is even worse for the fast-response foils used in the 4330F optode for 165 glider deployments. It has been suggested that it is necessary to calibrate and drift correct the optode using 166 discrete samples or in-air measurements (Nicholson and Feen, 2017). Unfortunately, no discrete samples were 167 collected at glider deployment or recovery.

168
To overcome this problem, we used archived data to correct for oxygen optode drift. These archived

222
In order to correct for the drift occurring during the glider mission, we selected the CO2 optode measurements in 223 water with σ0 > 1028 kg m -3 (just as for O2; section 2.3). We calculated the median of the raw optode phase shift 224 data ("CalPhase" φcal) for each Seaglider dive. Then, we calculated a drift coefficient (mi) as the ratio between 225 the median φcal for a given dive divided by the median φcal of dive 31. Drift-corrected φcal,d values were 226 calculated by dividing the raw φcal by the specific mi for each dive.

227
The CO2 optode was also affected by lag (

234
To apply the lag correction, the sampling interval (Δt) needs to be sufficiently small compared to the sensor 235 response time (τ) and the ambient variability (Miloshevich, 2004). Before the lag correction, φcal,d was smoothed 236 to remove any outliers and "kinks" in the profile using the Matlab function rLOWESS. The smoothing function 237 applies a local regression every 9 points using a weighted robust linear least-squares fit. Subsequently, τ was 238 determined such that the following lag-correction equation (Miloshevich, 2004)

252
The CO2 optode output was calibrated using the discrete samples collected throughout the mission. Using the 253 discrete sample time and potential density σ0, we selected the closest CO2 optode output. A linear regression 254 between optode output and c(CO2) from the discrete samples (cWS(CO2) was used to calibrate the optode output 255 pc(CO2) in terms of c(CO2). c(CO2) had a better correlation than p(CO2) (R 2 = 0.77 vs. R 2 = 0.02).

256
Plotting the regression residuals (cr(CO2), calculated as the difference between cWS(CO2) and the value predicted 257 by the regression) revealed a quadratic relation between the regression residuals and water temperature (θ). We

327
The inventory changes were calculated as the difference between two transects of the integrated oxygen

332
The inventory changes for every latitude bin were calculated using the following equation:

361
The effect of diapycnal eddy diffusion (Fv) was calculated at zmix when it was deeper than zlim and at zlim when 362 zmix was shallower than zlim, using the following equation:

379
The seawater c(CO2) at the surface was calculated using the median in the top 10 meters between the glider 380 ascent and descent of the following dive c(CO2). From this, Φ(CO2) was calculated: 381     Table 3; we repeated the analysis 1000 times. The total uncertainty in N was calculated as 407 the standard deviation of the 1000 Monte-Carlo simulations.

474
However, prior to the increased spring Chl a inventory, Φ(O2) showed a few days of influx into seawater caused 475 21 by a decrease of θ from 7.6 °C to 5.9 °C that increased Csat(O2). The influx at the beginning of the deployment is

485
The flux from sea to air is positive while that from air to sea is negative.

487
The CO2 flux from March to July was always from the air to the sea (Figure 12), with a median of -5.2 mmol m -2

502
The two N values were calculated as the difference in inventory changes between two transects when the glider 503 moved in the same direction.

504
During the deployment, we sampled two periods of increased Chl a inventory, the first one in May and a second 505 one in June. The chlorophyll a inventory ( raw, lim (Chl )) was calculated integrating craw(Chl a) to zlim. To 506 remove outliers we used a five-point moving mean of raw, lim (Chl ).

522
Integrating N(O2) from March to October gives a flux of 4.9 mol m -2 a -1 (Table 4;

541
Integrating N(DIC) from March to July gives a flux of 3.3 mol m -2 a -1 (Table 4; discussed in section 4.2). 542 543 Table 4. Net community production (N) estimates in the Norwegian Sea (with integration depth zlim).

580
The calibrated optode output captured the c(DIC) changes in space and time with a standard deviation of 11 581 µmol kg -1 compared with the discrete samples. c(DIC) decreased from 2130 µmol kg -1 to 2000 µmol kg -1 and 582 increased with depth to 2170 µmol kg -1 . This shows the potential of the sensor for future studies that aim to 583 analyse the carbon cycle using a high-resolution dataset.

585
These values are comparable with a previous study when the CO2 optode was tested for 65 days on a wave-  603 Table 4 shows estimates of net community production (N) in the Norwegian Sea. All other studies used ships to 604 gather observations. The estimated N in of the four other studies varied from 2.6 to 11.1 mol m -2 a -1 for N(O2)

616
However, the latter estimate was for the entire year, whereas our estimate only covers the months from March to

638
To the best of our knowledge, this study represents the first glider deployment of a CO2 optode. The CO2 optode

639
together with a O2 optode shows the potential of using these sensors on autonomous observing platforms like 640 Seagliders to quantify the interactions between biogeochemical processes and the marine carbonate system at 641 high spatiotemporal resolution. The deployment helped to uncover NCP and air-sea flux variability over a period 642 of 8 months.

643
Despite all the problems (drift, lag and poor calibration), the CO2 optode data could be used to quantify 644 dissolved inorganic carbon concentration variations. The temporal resolution sampling resolution was 106 s in 645 the top 100 m (increasing to 381 s from 500 to 1000 m). This could be improved to less than 10 s, but this would 646 reduce the length of the deployment due to the limited glider battery capacity. With better calibration and 647 stability improvements, the CO2 optode could be routinely used to measure the carbonate system on gliders,