the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Sep 2023
On the transition from strong to weak constraint 4DVar using asimple one-dimensional advection equation for a passive tracer
Abstract. In contrast to strong constraint 4DVar, the weak constraint takes into account the model imperfection in the minimisation process. Relaying on a simple one-dimensional advection equation for a passive tracer, this study shows that the transition from strong to weak constraint, accounting for both observations and model biases, reduces the analysis bias.
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(478 KB)
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1765', zezhong zhang, 27 Oct 2023
In the paper titled "On the transition from strong to weak constraint 4DVar using a simple one-dimensional advection equation for a passive tracer," the author conducts a comparison between strong and weak constraint 4DVar using a straightforward example of a 1D advection equation. The paper concludes that weak constraint 4DVar performs better than strong constraint 4DVar in cases of model inconsistency, aligning with theoretical expectations. However, this paper is limited in its contribution to the community due to its lack of innovative methodology and insufficient numerical experiments. As a result, I recommend rejecting this paper for publication. Here are detailed comments:
The paper only includes one reference, which suggests a lack of investigation into existing research on the comparison between weak and strong constraint 4DVar. It is essential to provide a more comprehensive review of the related literature in this field.
As a paper that compares two existing methods using a very simple problem, the numerical example should be rigorous and comprehensive to offer meaningful insights into these two methods when applied to the given problem. The current numerical results are too simplistic to provide substantial insights. Here are some suggestions for improving the numerical experiments:
-
In the 1D advection equation, consider both introducing diffusion and modifying the true velocity 'u'. This modification would help distinguish which type of model error is more sensitive to the assimilation quality.
-
Explore a range of values for each variable 'u' and 'k' so that there is a continuous change in assimilation quality as model error increases in two directions. This will allow for a more robust analysis of the methods' performance.
-
Set up repeated experiments to calculate bias statistics from a distribution rather than relying on a single data point. This will provide a more reliable assessment of the assimilation quality.
Citation: https://doi.org/10.5194/egusphere-2023-1765-RC1 -
- RC2: 'Comment on egusphere-2023-1765', Anonymous Referee #2, 08 Dec 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1765', zezhong zhang, 27 Oct 2023
In the paper titled "On the transition from strong to weak constraint 4DVar using a simple one-dimensional advection equation for a passive tracer," the author conducts a comparison between strong and weak constraint 4DVar using a straightforward example of a 1D advection equation. The paper concludes that weak constraint 4DVar performs better than strong constraint 4DVar in cases of model inconsistency, aligning with theoretical expectations. However, this paper is limited in its contribution to the community due to its lack of innovative methodology and insufficient numerical experiments. As a result, I recommend rejecting this paper for publication. Here are detailed comments:
The paper only includes one reference, which suggests a lack of investigation into existing research on the comparison between weak and strong constraint 4DVar. It is essential to provide a more comprehensive review of the related literature in this field.
As a paper that compares two existing methods using a very simple problem, the numerical example should be rigorous and comprehensive to offer meaningful insights into these two methods when applied to the given problem. The current numerical results are too simplistic to provide substantial insights. Here are some suggestions for improving the numerical experiments:
-
In the 1D advection equation, consider both introducing diffusion and modifying the true velocity 'u'. This modification would help distinguish which type of model error is more sensitive to the assimilation quality.
-
Explore a range of values for each variable 'u' and 'k' so that there is a continuous change in assimilation quality as model error increases in two directions. This will allow for a more robust analysis of the methods' performance.
-
Set up repeated experiments to calculate bias statistics from a distribution rather than relying on a single data point. This will provide a more reliable assessment of the assimilation quality.
Citation: https://doi.org/10.5194/egusphere-2023-1765-RC1 -
- RC2: 'Comment on egusphere-2023-1765', Anonymous Referee #2, 08 Dec 2023
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 292 | 93 | 24 | 409 | 21 | 25 |
- HTML: 292
- PDF: 93
- XML: 24
- Total: 409
- BibTeX: 21
- EndNote: 25
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1