Climate change impacts on solar power generation and its spatial variability in Europe based on CMIP6

Hou, Xinyuan; Wild, Martin; Folini, Doris; Kazadzis, Stelios; Wohland, Jan

doi:https://doi.org/10.5194/esd-12-1099-2021

Articles | Volume 12, issue 4

https://doi.org/10.5194/esd-12-1099-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/esd-12-1099-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 12, issue 4

Research article

|

09 Nov 2021

Research article |

| 09 Nov 2021

Climate change impacts on solar power generation and its spatial variability in Europe based on CMIP6

Xinyuan Hou, Martin Wild, Doris Folini, Stelios Kazadzis, and Jan Wohland

Download

Final revised paper (published on 09 Nov 2021)
Preprint (discussion started on 09 Jul 2021)

Interactive discussion

Status: closed

CC1: 'Comment on esd-2021-57', Richard Rosen, 11 Jul 2021

This is an interesting but problematic article because, in my view, it tries to do too much. I think the most important part of the article for ESD reviewers and commenters to focus on is the first stage of the calculations, namely whether the inputs that can be obtained from the CMIP6 model runs for climate change scenarious out to 2100 can be reasonably and accurately used as inputs to the GSEE model which calcuates the output of PV cells throughout Europe. If this can be done well from the perspective of physics, then the rest of the article and calculaltions are fairly trivial and not very important for climate change mitigation policy since the computed differences between scenarios is small. But, again, the really important science occurs in the first stage of the calculations as noted above.
The basic methodology of the first stage calculations are spelled out in lines 70-75. There the four kinds of output parameters from the 28 CMIP6 models are listed. Without doing any modeling my guess as a physicist is that the change in cloud cover from one scenario to another (from the 2.6 to the 8.5 forcing scenarios) will have the most impact on the difference in the average PV output across Europe, but I may be wrong, of course. Anyway, what is missing in this research article is a clear and simple description of how these four kinds of output parameters from the CMIP6 models can be input to the GSEE model. And while the reader might have some trust in the reasonableness of using CMIP6 results for calcuations of other kinds of parameters because the CMIP process is such a huge multinational multiyear research effort, the reader is not provided with any scientific rationale for believing that the GSEE model is accurate and useful for the purpose the authors are using it for. However, even the CMIP6 results have hugh uncertainty ranges associated with them as described in the relevant CMIP literature, and calculations of cloud cover differ greatly from model to model and are generatlly considered to be highly uncertain from model to model, and, therefore, from climate change scenario to scenario.
Most importantly, though, before an article like this could be considered publishable material in ESD, it would have to clearly demonstate that the GSEE model on which it solely relies can accurately translate the four inputs form the CMIP6 models into PV cell output in terms of performing the correct physical calculations. For example, reviewers of this article need to ask the authors to provide the exact equations in GSEE which take the four inputs and produce electricity from PV cells out. Since the results cited is this paper involve relatively small changes (a few percent) between the two scenarios modeled for PV output, as one would expect, the GSEE model would have to be far more accurate than a few percent in its translation of each of the four input parameters separately into PV electricity output. Can the authors make this case? I don't know but they must try to do so if they want to put this research article on solid ground. The GSEE model is hardly a well established and well validated model, as far as I know.
The are problems with the methodologies used in the remainder of the paper as well, but I think the authors need to first convince the reader that the physics of the first stage calculations are highly accurate for the rest of the paper to matter.

Citation: https://doi.org/10.5194/esd-2021-57-CC1
CC2: 'Comment on esd-2021-57', Jan Wohland, 14 Jul 2021

Dear Richard Rosen,

thank you for your interest in our article and the time you spent evaluating it. Given that your comment was posted very soon after our article went online, we take the opportunity for a brief reply now while looking forward to the posting of the comments from the ESD reviewers, who are certainly highly qualified to assess our submission.

Regarding your main concern, please note that the GSEE model is documented in Pfenninger & Staffell (2016). We cite this study prominently in the Method section. Model equations and validation are given in that paper and we decided not to repeat the formulae in the interest of brevity. GSEE is the model used in the renewables.ninja dataset, which is heavily used, as for example indicated by almost 600 citations of the paper. The combination of GSEE with CMIP data is examined in detail in Müller et al. (2019), which is also cited in our manuscript. Moreover, the GSEE model is openly available on github (https://github.com/renewables-ninja/gsee), so please feel free to inspect it and raise issues there in case you spot anything.

Please further note that we directly address the issues of model uncertainty in CMIP6 by evaluating ensemble agreement and that we have also analyzed the relative role of cloud cover versus changes in clear-sky components. We‘ll gladly comment on these aspects in greater detail again when providing the formal response to the reviewers.

Thanks again for your opinion.

Jan Wohland
(on behalf of all authors)

REFERENCES

Pfenninger, S., Staffell, I., 2016. Long-term patterns of European PV output using 30 years of validated hourly reanalysis and satellite data. Energy 114, 1251–1265. https://doi.org/10.1016/j.energy.2016.08.060

Müller, J., Folini, D., Wild, M., Pfenninger, S., 2019. CMIP-5 models project photovoltaics are a no-regrets investment in Europe irrespective of climate change. Energy 171, 135–148. https://doi.org/10.1016/j.energy.2018.12.139

Citation: https://doi.org/10.5194/esd-2021-57-CC2
RC1:
'Comment on esd-2021-57', Anonymous Referee #1, 10 Aug 2021

Review of the paper “Climate change impacts on solar power generation and its spatial variability in Europe based on CMIP6”, by Xinyuan Hou et al., submitted for publication in Earth System Dynamics.

General comments

This paper assesses the availability of solar power for electricity production in Europe, under future emission scenarios. The authors first analyse the future change in climate variables relevant for solar power availability, as described in state-of-the-art CMIP6 climate projections. Then they use the Global Solar Energy Estimator (GSEE) to predict and analyse the change in the PV potential under a mitigation (SSP2.6) and a fossil-fuel dependent (SSP8.5) scenario.

The paper is well written, in a clear and concise manner. Data and methods are appropriate for the purpose of the study, and conclusions are supported by evidence. The paper does not add much to the existing knowledge on the expected changes of the PV potential in Europe in relation of climate change. However, it has the merit of updating the existing picture by using the last state-of-the-art climate projections, also adding the assessment of the PV power complementarity at the continental scale. I believe that this is a valuable contribution to the knowledge in the field and I recommend publication.

I have no major comments to address to the authors. However, some minor issues should be fixed before final publication. Please see my comments below.

Specific comments

L14: Please add few words to clarify how “a decrease in clear-sky radiation is overcompensated by a decrease in cloud cover”.

L98: Is the role of wind speed negligible in the simulation of the PV potential in GSEE? Several authors take it into account in their computations (see e.g. the CLIMIX model by Jerez et al. 2015).

L115: Computing the daily seasonal cycle by using daily climatology usually produces a very noisy seasonal cycle. Why not to add a filter (e.g. moving average) to smooth it?

L125: I believe that also assessing the significance of correlation would add valuable information. I would also highlight in the discussion that the optimal correlation in this case is not zero, but negative correlation.

L150-152 and L154-155: I believe that this lines should be at the end of the section. Also not to disrupt the narrative of the paper.

Section 3.1: Concerning the changes in cloud cover, which are the key element for explaining the changes in PV potential, can you also discuss possible links with changes in atmospheric circulation? E.g. NAO, jetstream, stormtrack.

L226: What do you mean with "climatic boundary conditions"? Please explain.

L236: This is just a comment, not a request. I feel that the “unfortunately” is not appropriate in the context of a scientific/technical paper, which in my view should be as neutral as possible. However, I don’t want to question your style, nor ask for modifications, just stimulate reflection.

L248-250: You could also refer to the possibility of complementary PV-wind power production -> https://journals.ametsoc.org/view/journals/apme/56/1/jamc-d-16-0031.1.xml

L254: I'm wondering why central Europe also includes the Alpine region, which is supposedly not suitable for PV power production.

Figure 6: I find that this figure is poorly discussed. Especially panel (c) showing interesting results should be further commented. Moreover, you could give an indication on how to discern between “good and bad” correlations.

L268: I see mean correlation is 0.022 and 0.001 on average and the PDF is between ±0.1 circa. I don't understand why you mention 0.05 and 0.4 as reference values.

L315: Small correlations indicate that complementarity is possible. However, ideal complementarity is associated with negative correlations. You should further discuss this aspect when commenting on Fig. 6 and 7.

Conclusions: In my understanding, in a context of decreasing aerosol content, the decrease in clear-sky radiation should be associated with increasing water vapour. Am I correct? If so, could you show that?

Citation: https://doi.org/10.5194/esd-2021-57-RC1
- AC1: 'Reply on RC1', Xinyuan Hou, 17 Sep 2021
  
  Please see the attached PDF.
  
  Citation: https://doi.org/10.5194/esd-2021-57-AC1
RC2:
'Comment on esd-2021-57', Anonymous Referee #2, 11 Aug 2021

General comment:

The manuscript presents an interesting topic that has already been studied using projections from CMIP-5 and also regional climate models. However it is important to estimate projected changes of solar radiation and photovoltaic production with the new generation of climate projections from CMIP-6, in order to know the agreement or disagreement with previous results. I recommend the article for publication, although I have some comments to be addressed first.

Specific comments:

Methodology:

It could be good to remark that the PV model is using the inclination of the PV panels. This was not very common in other studies like Jerez et al. 2015.

Results:

The results on clear sky radiation are very interesting, and show differences between the two scenarios. Although the authors speculate with the reasons behind that differences, the variables that are involved in the explanation are not shown: water vapor content and aerosols. You should include them in the analysis. Is the clear-sky radiation of the different climate models provided by each institute? If not, how it is calculated?.

In line 170, it would be nice to indicate that the uncertainty in the rsdscs is widespread for the 8.5 scenario.

Regarding the seasonal results: there is an increase of PV production in summer and a decrease in winter months, it should be further commented about this fact in terms of the implications of these changes, considering that PV peaks in summer.

Results are based on changes at the end of the century and there is the issue of uncertainty. Can they be useful for policymakers, investors or utilities? A comment on that would enrich the paper.

The PV correlation changes shown are positive for most of the domain, and very low. However, it is not clear how is the correlation in the historical period (Fig 5), it would be good to see them. In addition, it should be included that a negative correlation close to -1 is the most ideal case, isn’t it?.

From my point of view, it is not very interesting to analyse the correlation between cells of the domain or regions that include different countries. In terms of energy, countries are operating as single units, if I am not wrong. Correlation between single countries is more appropriate to do this balancing study.

Line 31: Please, add more references to this sentence.

Line 43: more up to date paper of EURO-CORDEX: “Regional climate downscaling over Europe: perspectives from the EURO-CORDEX community” Jacob et al. 2020. Please state more clear the differences between global and regional climate models.

Line 45: not all the models keep the aerosols constant, although most of them. Please correct.

Line 160: “All-sky radiation evolves differently…” I would say that they evolve similarly: in both scenarios there is an increase, although it is of different magnitude. The other variables evolve differently.

Line 297: some other references should be included when it is said “due to a too simple treatment of aerosols…” Boé et al. 2020: Boé, J., Somot, S., Corre, L., & Nabat, P. (2020). Large discrepancies in summer climate change over Europe as projected by global and regional climate models: causes and consequences. Climate Dynamics, 54(5), 2981-3002. Gutiérrez et al. 2020: Gutiérrez, C., Somot, S., Nabat, P., Mallet, M., Corre, L., van Meijgaard, E., Perpiñán, O. & Gaertner, M. Á. (2020). Future evolution of surface solar radiation and photovoltaic potential in Europe: investigating the role of aerosols. Environmental Research Letters, 15(3), 034035.

Line 309: “Jerez et al. (2015) also confirmed a slightly positive trend in southern Europe based on regional climate models” I think there is also a difference between the analysed periods.

In order to help the reader with the results and the comparison, it would be very useful to have the same scale for the maps of the two scenarios.

Line 319: “In the mitigation SSP1-2.6…” it sounds repetitive with previous sentence, try another formula.

Citation: https://doi.org/10.5194/esd-2021-57-RC2
- AC2: 'Reply on RC2', Xinyuan Hou, 17 Sep 2021
  
  Please see the PDF that we uploaded as an answer to RC1, where we jointly address all issues raised by both reviewers.
  
  Citation: https://doi.org/10.5194/esd-2021-57-AC2

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

ED: Publish subject to minor revisions (review by editor) (19 Sep 2021) by Somnath Baidya Roy

AR by Xinyuan Hou on behalf of the Authors (27 Sep 2021) Author's response Author's tracked changes Manuscript

ED: Publish as is (29 Sep 2021) by Somnath Baidya Roy

AR by Xinyuan Hou on behalf of the Authors (30 Sep 2021)

Short summary

Solar photovoltaics (PV) matters for the carbon neutrality goal. We use climate scenarios to quantify climate risk for PV in Europe and find higher PV potential. The seasonal cycle of PV generation changes in most places. We find an increase in the spatial correlations of daily PV production, implying that PV power balancing through redistribution will be more difficult in the future. Thus, changes in the spatiotemporal structure of PV generation should be included in power system design.