the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Development of the adjoint of the GEOS-Chem unified tropospheric-stratospheric chemistry extension (UCX) in GEOS-Chem Adjoint v36
Abstract. Atmospheric sensitivities (gradients), quantifying the atmospheric response to emissions or other perturbations, can provide meaningful insights on the underlying atmospheric chemistry or transport processes. Atmospheric adjoint modelling enables the calculation of receptor-oriented sensitivities of model outputs of interest to input parameters (e.g. emissions), overcoming the numerical cost of conventional (forward) modelling. The adjoint of the GEOS-Chem atmospheric chemistry-transport model is a widely used such model, but prior to v36 has lacked extensive stratospheric capabilities. Here, we present the development and evaluation of the discrete adjoint of the global chemistry transport model (CTM) GEOS-Chem unified chemistry extension (UCX) for stratospheric applications, which extends the existing capabilities of the GEOS-Chem adjoint to enable the calculation of sensitivities that include stratospheric chemistry and interactions. This development adds 37 new tracers, 273 kinetic and photolysis reactions, an updated photolysis scheme, treatment of stratospheric aerosols, and all other features described in the original UCX paper. With this development the GEOS-Chem adjoint model is able to capture the spatial, temporal and speciated variability in stratospheric ozone depletion processes, among other processes. We demonstrate its use by calculating two-week sensitivities of stratospheric ozone to precursor species and show that the adjoint captures the Antarctic ozone depletion potential of active halogen species, including the chlorine activation and deactivation process. The spatial variations in the sensitivity of stratospheric ozone to NOx emissions are also described. This development expands the scope of research questions that can be addressed, by allowing stratospheric interactions and feedbacks to be considered in the tropospheric sensitivity and inversion applications.
- Preprint
(6820 KB) - Metadata XML
-
Supplement
(256 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on gmd-2023-233', Anonymous Referee #1, 25 Feb 2024
Adjoint models are highly effective tools for identifying sensitivities in physical processes, making their development invaluable to the scientific community. The authors have commendably demonstrated how the proposed adjoint model can be utilized to explore sensitivity information, suggesting its potential applicability across a broad spectrum of use cases. However, my primary concern with the study lies in the validation of the developed adjoint model, as detailed below. Addressing the following two points would make me more inclined to recommend this work for publication:
1. The validation of adjoint sensitivities against forward modeling results in this study is conducted qualitatively. To establish the adjoint's accuracy quantitatively, it is essential to validate the model using the relationship <Δx_1, MΔx_2> = <M^TΔx_1, Δx_2>, where x_1 and x_2 represent any state vectors, M symbolizes the tangent linear of the forward model, and M^T denotes the adjoint. The left-hand side (LHS) and right-hand side (RHS) of this equation should match to machine accuracy (e.g., 13 or more digits in double precision) to ensure reliability. Without this level of agreement, the results, while appearing reasonable in certain instances, cannot be universally trusted.
2. Demonstrating the validity of the tangent linear approximation within the context under study is crucial before employing the adjoint model to derive sensitivities. This step is fundamental to ensuring that the adjoint sensitivities are meaningful. A practical approach to validate this would be to compare the outcomes of the tangent linear model with those from nonlinear forward model differences, thereby reinforcing the model's credibility.
Addressing these concerns will significantly strengthen the manuscript, making it a robust contribution to the field.
Citation: https://doi.org/10.5194/gmd-2023-233-RC1 -
RC2: 'Comment on gmd-2023-233', Anonymous Referee #2, 16 Mar 2024
This paper is well in the scope of GMD for its model development of UCX in GEOS-Chem Adjoint.
My main concern for manuscript is that for a model development paper, the methods part - including model description and especially model development - is lacking. As a GEOS-Chem modeler myself, I cannot reproduce the authors’ work based on the two paragraphs currently present in the paper. The development description is too general and vague to be replicable. It could benefit from more details, especially for a model development publication.
Two other minor suggestions:
- The title is repetitive, and I advise a more succinct “Development of the unified tropospheric- stratospheric chemistry extension (UCX) in GEOS-Chem Adjoint v36” would get the point across sufficiently. But this may be a stylistic choice and merely a suggestion.
- Most figures could also benefit from better pixels for publication, for example, Figs 6 and 7’s fonts are small, and once enlarged, they’re quite blurry.
The rest of the paper is scientifically sound and well presented. Once more details for the methods are added, I would support its publication in GMD.
Citation: https://doi.org/10.5194/gmd-2023-233-RC2 - AC1: 'Response to referee comments', Irene Dedoussi, 17 Apr 2024
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
270 | 65 | 22 | 357 | 26 | 16 | 14 |
- HTML: 270
- PDF: 65
- XML: 22
- Total: 357
- Supplement: 26
- BibTeX: 16
- EndNote: 14
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1