| 29 Sep 2025
Ocean dynamics amplify remote warming effects of reforestation
Abstract. Forestation, including reforestation, afforestation, and forest restoration, is prevalent in net-zero climate strategies due to the large carbon sequestration potential of forests. In addition to capturing carbon, forestation has biogeophysical effects that can influence surface temperatures locally (local effects), and at distant locations (non-local effects). Biogeophysical effects may offset the cooling benefits of carbon sequestration, hence requiring a robust understanding of their mechanisms to adequately integrate forestation into climate mitigation strategies. Yet, the role of ocean dynamics, such as ocean circulation, ocean-atmosphere interactions, and ocean-sea ice interactions in mediating the non-local effects of forestation remains underexplored. In this study, we investigate the impact of ocean dynamics on the magnitude and geographic patterns of the non-local biogeophysical effects of large-scale reforestation, with the exclusion of cloud feedbacks, over a multi-century timescale using the University of Victoria Earth System Climate Model. We conduct multi-century paired global reforestation simulations, with the first set of simulations using a dynamic ocean and the second set using prescribed sea surface temperatures. We separate local from non-local effects using the checkerboard approach. Our results show that non-local warming effects are of much greater magnitude and encompass a greater geographic area, particularly at high latitudes, when ocean dynamics are considered. Moreover, this study shows that ocean dynamics introduce a lag in the non-local effects, leading to a continued increase in non-local warming even after the local effects have stabilized. This committed non-local warming is driven by the thermal inertia of the ocean, which sustains a gradual long-term increase in sea surface temperatures, combined with amplifying climate feedbacks. Decision-making frameworks must therefore consider the complete Earth system response to forestation over a sufficiently long timeframe to account for the committed non-local warming.
Competing interests: Kirsten Zickfeld is an editor for this journal.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
The study analyses long-term forestation effects on temperature, with a focus on non-local influences and the role of slower components of the Earth system such as the ocean. I find this work provides novel insights and is well structured, clearly highlighting the most important findings. Overall, my comments are minor and mainly aimed at clarifying the implications of the specific model used (UcVic) and how it differs from or compares with models used in previous literature. I believe that the use of UcVic model simulations performed here and the performed analysis with a focus on the role of the ocean in the long term is relevant and provides new insights that can inform future research. Hopefully these comments can further help improve the manuscript.Â
A first comment concerns the final land-cover map after regrowth shown in Figure 1. The model includes only two forest types, shrubs and two grass types as the penultimate vegetation category. As I understand it, the land model has dynamic vegetation, meaning that land cover is established during the simulation according to the local climate. Although the global patterns appear reasonable, the distribution seems somewhat biased toward needleleaf forest over Europe and appears unable to represent the Siberian boreal forests. Could the authors add some discussion on whether the vegetation module has been evaluated, how it compares to present day land cover, and what the potential implications of these differences might be?
Secondly, I would like to commend the authors for the clear discussion of the potential role of cloud feedbacks, which are not represented in the model. However, several studies (a.o Portmann etal., 2022) also highlight the importance of atmospheric dynamics, which, as I understand from the methods section, are not included in the current modelling setup. I believe it would be useful to further discuss potential implications of this as has been done for the cloud feedbacks, given that atmospheric dynamics have been shown to be particularly important in the context of non-local effects.
In the discussion, the authors note that some effects of vegetation greening are found here that are typically not present in previous research (line 380-381). One reason given is that greening is slow to establish. Could the authors clarify what is meant by this? Only a few previous studies employ dynamic vegetation; most rely on prescribed land-cover maps for forestation scenarios. Is that what is meant here, i.e. that differences arise (partially) due to the model setup? The underlying point is the same, but being more explicit would help the reader follow the argument.
Lastly, the authors acknowledge that their simulations differ significantly from results presented by Boysen et al. (2020) in the deforest-glob simulations, particularly in terms of latent heat flux. I follow the argument that, because of a high bias in sensible heat flux, the total non-radiative flux is comparable, implying that the results may still be reasonably comparable. However, I wonder whether the authors have any insight into why this bias occurs. Including some explanation of what happens within the land model, or at least a hypothesis regarding the possible cause, would help readers understand and contextualize the discrepancy relative to other models.