Interactive comment on “ Imminent ocean acidification projected with the NCAR global coupled carbon cycle-climate model ”

In this paper the evolution of ocean aragonite saturation levels over the 21st century is examined for two CO2 emission scenarios with the help of a coupled climate-carbon model. The authors present results not only at the global scale but also at the regional and seasonal scales. They offer a detailed analysis of the evolution in the Arctic Ocean. The consequences for ecosystems are also adressed. The model results lead to the conclusion that unless future CO2 levels are drastically controlled a dramatic situation may develop in the oceans.

sons are twofold. First any tracer distribution that would be coherent with the model hydrodynamics most probably differ from that provided by WOA01 (the correlation and relative standard deviation would be very different from 1). Second, considering that SIO 4 keeps its present-day distribution despite the profound changes that could occur throughout the 21st century is not appropriate. One could argue that the impact of the authors method on the actual values of chemical variables is small but this method is nevertheless not scientifically sound.
References for the calculation of K sp and the pressure dependency have been included. Also, the reference for OCMIP routines has been changed. Regarding the treatment of silica, we don't fully agree with the referee. Silica is part of total alkalinity and setting Si(OH) 4 concentrations to a constant value everywhere would lead to a systematic bias as well. Further, the changes in ocean circulation throughout the 21st century are relatively weak. Therefore, it is not clear which method would be better and we believe that using the present-day silica distribution is justifiable. In any case, as shown by Steinacher (2007), the effect of changes in Si(OH) 4 concentrations on the calculation of carbonate ion concentration and saturation state is small. In a sensitivity analysis, silicate concentrations have been either set to zero, increased uniformly by 20 µmol/l (which corresponds to 10-90% do the original concentration), or doubled. These changes lead to local differences in [CO 2− 3 ] and Ω arag of 0.4±0.2%, 0.5±0.1% and 0.6±0.3% respectively; maximum values are around 1%. This is now mentioned in the text. Further, the second part of the paragraph has been reformulated:  . In these routines, based on work by Dickson (2002), Millero (1995), and Mehrbach et al. (1973 (Lueker et al., 2000) is utilized. The apparent solubility product K sp is calculated after Mucci (1983) and the pressure dependency of chemical constants after Millero (1995 and Ω arising from this treatment of Si(OH) 4 have been estimated to be less than 1% (Steinacher, 2007)." and Ω arag in the Taylor diagram ( Fig. 2) is striking. I suppose this is a consequence of the model performances at reproducing the T and DIC (and may be S) fields. It would be worth investigating the reasons for such a difference. If available such an analysis would prove useful for model evaluation as well as provide indications of confidence levels in the predicted changes. For a better understanding similar plots for DIC, T and S should be provided. They would help the reader appreciate the model performances.
The difference between [CO 2− 3 ] and Ω arag in the Taylor diagram is mainly a consequence of the pressure-dependent solubility product K sp . The relationship between data-model differences in Ω and in [CO 2− 3 ] is approximately K sp increases with depth. This implies that data-model differences in Ω are less heavily weighted at depth than at the surface relative to data-model differences . Consequently, the model-data comparison is more favorable for Ω than for [CO 2− 3 ] when data from the entire water column are compared with observations as model-data deviations in [CO 2− 3 ] are larger at depth than at the surface as shown in Fig. 3. Almost no difference is found in the Taylor diagram statistics between Ω and [CO 2− 3 ] when only surface values are considered where variations in K sp are small (dark blue symbols in Fig. 2a). This is now mentioned in the text: "Correlation coefficients are somewhat smaller (0.86 to 0.91) for [CO 2− 3 ] than for Ω arag because the data-model differences in Ω arag decrease with depth relative to the differences in  ]. This results because the increase in the pressure-dependent solubility product K sp with depth decreases the weighting of the larger errors in deep-water [CO 2− 3 ] (Fig. 3)." Further, a similar Taylor diagram has been added for DIC, T and S fields (Fig. 2b).
-Global evolution of pH and Ω arag arag. Previous works (Orr et al., 2005;Cao et al., 2007;McNeil and Matear, 2007) conclude in a weaker effect of climate on pH than on Ω arag by 2100 A.D. This is in contrast with the present study in which the impact on both pH and Ω arag is of the same order of magnitude (page 4363, lines 26-28). On page 4370 the authors suggest that one possible explanation is that McNeil and Matear (2007) used a prescribed CO 2 concentration scenario rather than a CO 2 emission scenario. This is in contradition with the results of Cao et al. (2007). Indeed Cao et al. (2007) performed experiments with both constrained and prognostic atmospheric CO 2 . In both cases the pH relative changes do remain smaller than the relative changes of Ω arag (Table 1 in Cao et al. (2007)). I do not see any reason why an emission scenario rather than a concentration scenario would lead to different relative behavior in pH and Ω. The reason for the differences among the above-mentionned studies must lie in the ocean processes. One exploratory path could be to reproduce Fig. 6 from Mcneil and Matear (2007) with the present model results and look for differences.

Printer-friendly Version
Interactive Discussion

Discussion Paper
We agree that our results are not fully consistent with the results of Cao et al. (2007) in terms of magnitude. However, the opposite behavior of pH decrease (enhanced by climate change) and Ω arag decrease (reduced by climate change) is consistent. This has been clarified in the text on page 4370, line 9: "This opposite behavior is consistent with the analysis of Cao et al. (2007) for carbon emission scenarios, although they found a weaker effect on pH than on Ω arag in their study with an Earth System Model of Intermediate Complexity." The difference between using an emission scenario rather than a concentration scenario has already been discussed by Cao et al. (2007, paragraph 13): "McNeil and Matear [2006] reports that the effects of climate change on surface ocean pH are negligible from a coupled climate-carbon cycle simulation driven by the IS92a atmospheric CO 2 concentration pathway. With prescribed CO 2 concentrations we project negligible climatic effects on surface pH (Table 1), consistent with their study. In this case, the indirect DIC effect almost cancels the direct temperature effect (not shown), leading to a negligible net climatic effect on pH. However, with prescribed CO 2 emissions, we find that consideration of climate change has a pronounced effect on surface pH (namely, to cause a greater decrease in pH, as seen from Figure 2c); the direct temperature effect dominates the indirect DIC effect (as explained above) (Figure 3a)." A reference to Cao et al. (2007) has been added on page 4370, line 13: Mcneil and Matear (2007) found a climate-change feedback reducing the change in Ω arag by about 15% but almost no feedback on pH; they applied a scenario with prescribed CO 2 concentration and not a scenario with prescribed carbon emission as done here (see Cao et al. (2007) for a discussion of differences between emission and concentration scenarios).
-Changes in the Arctic (pages 4366 and 4367). The combination of model and data S3313 BGD 5, S3308-S3321, 2009 Interactive Comment Full Screen / Esc

Printer-friendly Version
Interactive Discussion Discussion Paper such as performed here implies that the model bias is and would remain linear. Isn't this assertion at odds with the non-linearity of the carbonate chemistry? I would surely not state as the authors do on page 4367 that "... the emerging undersaturation of the surface Arctic Ocean is a robust feature and independent of these model biases". Since the reasons for the bias are not clearly elucidated there are no reasons to believe that the evolution of the aragonite saturation would be that predicted by the model. The reasons for the model bias could result in non-homogeneous bias to occur with time.
We have compared the effect of adjusting the individual components (T, S, DIC, Alk, PO 4 ) instead of correcting Ω directly and we have found a small effect (<1%). This is mentioned in the text on page 4366, line 21 and rules out problems with the non-linearity of the carbonate chemistry. We agree that a potential nonhomogeneous bias with time can't be entirely eliminated by this method. However, while it is true that we do not fully understand the reason for the model bias, we do well understand the driving forces behind the decreasing trend in saturation. This is the increase in atmospheric CO 2 by anthropogenic emissions and the penetration of this perturbation into the ocean. The statement "...is a robust feature and independent of these model biases" on page 4367 refers to the fact, that the emerging undersaturation of the surface Arctic Ocean is projectd with and without model-data correction.
-Results presentation. A first remark is that the titles of subsubsections 3.3.1 and 3.3.2 contradict that of the parent subsection 3.3. Also subsubsection 3.3.2 is quite important in size and subject. It should deserve to be discussed in a subsection of its own. Further some global aspects discussed on page 4363 are again addressed at the end of section 3.3.2. Lines 7-20 on page 4370 should be gathered with lines 22-28 on page 4363. Some reorganization of section 3. would improve the readability of the manuscript. The discussion about the Arctic ocean should be separated and material from 3.3.1 and 3.3 should be merged with that in 3.2. I suggest something like this: 3.1 Comparison of modeled aragonite saturation and CO3 concentration with S3314 BGD 5, S3308-S3321, 2009 Interactive Comment Full Screen / Esc

Printer-friendly Version
Interactive Discussion Discussion Paper observation-based estimates 3.2 Projected global and regional changes 3.3 Seasonal and interannual variability of Ω arag

Changes in the Arctic Ocean and climate feedbacks
There was a mistake in the subsection numbering. We apologize for the confusion. The numbering is now as follows:

Comparison of modeled aragonite saturation and CO 2− 3 concentration with observation-based estimates 3.2 Projected global mean changes 3.3 Regional changes in saturation at the surface 3.4 Changes at depth and in the volume of supersaturated waters 3.5 Changes in the Arctic Ocean and climate feedbacks 3.6 Seasonal and interannual variability of Ω arag
As suggested by the referee, the global aspects on page 4363 (last two paragraphs) have been merged with the text in section 3.2 (second paragraph) and section 3.3 (last paragraph).
-Abstract, lines 11-12. I am in favour of reporting pH changes in pH units rather than in hydrogen ion concentration changes. pH units usually carry more meaning for the reader.
-Abstract, lines 15-16. I do not understand the sentence "Aragonite undersaturation in Arctic surface waters is projected to occur locally soon and to become more widespread as atmospheric CO 2 continues to grow." The sentence has been changed to include a more precise statement on the time "Aragonite undersaturation in Arctic surface waters is projected to occur locally within a decade and to become more widespread as atmospheric CO 2 continues to grow." In addition, we have changed the manuscript title to: "Imminent ocean acidification in the Arctic projected with the NCAR global coupled carbon cycle-climate model" -Use of adjective "alkaline" (p. 4354, line 23 and p. 4356, line 17): in the everyday language the word alkaline is often used as a synonymous for base. I would recommend not to use this word in the present context mainly because it may confuse the reader and let her believe the authors refer to an alkalinity change rather than to a pH modification but also because of the fact that not all bases are alkali.

The two sentences have been changed to:
"The continued emissions of CO 2 by human activities cause atmospheric CO 2 to rise, climate to warm, and the ocean pH to decrease." and "The hydrolysis of CO 2 in seawater lowers ocean pH, making the oceans less basic." -p. 4356, line 9. Wouldn't fertility be more appropriate than fertilization?
Yes. This has been changed. -p. 4359, line 15. Is the too extensive ice cover in NP and NA a cause or a consequence of the model shortcoming? As formulated the sentence implies the first! This text passage has been reformulated (see answer to Referee 3).
-p. 4361, 1st paragraph. Couldn't this paragraph be reformulated in a more concise way?
This paragraph has been reformulated (see answer to specific comment above). Averaged over the entire Arctic, surface annual mean Ω arag becomes less than unity in the model when atmospheric CO 2 exceeds 490 ppm (2040 A.D. in A2 and 2050 A.D. in B1). Since the figure shows Ω arag as a function of atmospheric CO 2 and there is virtually no difference between A2 and B1 at the same CO 2 level, we see no reason to add a second plot for scenario B1. Instead, the corresponding years in B1 have been added at the top of the plots. The following sentence refering to Fig. 6b has been added to the text on page 4364: "Depending on the seasonal amplitude, short-time undersaturation during at least one month is reached several years earlier in many regions (Fig. 6b)." Further, seasonality and annual minumum values of Ω arag are discussed later in the text, therefore Fig. 6b could be interesting as a reference.
- Figure 6. An explanation of the meaning of the dotted lines is missing in the caption.

The caption has been modified to:
"Projected evolution of the (a) annual-mean and of the (b) lowest monthly mean zonally-averaged aragonite saturation Ω arag for the SRES A2 scenario (model only). The evolution is plotted as a function of the annual-mean atmospheric CO 2 mixing ratio at the ocean surface. The corresponding years in the A2 and B1 scenarios are given at the top. The dotted line indicates the transition from supersaturation to undersaturation in zonal average Ω arag at 77 • N by 2032 and 2016, respectively." - Figure 7. The global average is dominated by the Pacific Ocean. I would suggest that four pannels be drawn : Atlantic, Pacific, Arctic and Southern Ocean.
We agree. The global average has been replaced with plots of the Atlantic and Pacific.

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper which is mostly discussed in the text rather than ∆CO 2− 3 ? It is important to state that we refer to aragonite saturation, because ∆CO 2− 3 is different for the several mineral phases. The caption has been modified to: "... annual mean ∆[CO 2− 3 ] (µmol/l) with respect to aragonite ..." We intentionally show ∆CO 2− 3 in all figures with meridional sections to provide an alternative to Ω for those who are more interested in (excess) concentrations.
- Figure 12. Wouldn't it be possible to re-organize the figure into 2 rows (and 3 columns) rather than 3 for better readability?
We will re-arrange this figure during the production process (and possibly others too) to fit the final layout of the paper as well as possible.