Review Reports - Ensemble-Based Flow Field Estimation Using the Dynamic Wind Farm Model FLORIDyn

Round 1

Reviewer 1 Report

This article explores the possibility of estimating the wind farm flow field by means of implementing an Ensemble Kalman Filter framework to FLORIDYN.

When reviewing the manuscript, I realized that its main topic is not a discussion of wind turbine wakes but wind farm flow modeling, which is not my area of expertise, and I apologize to the authors that I won't be able to give a lot of valuable comments.

The manuscript is very well written and explains the implementation of the Ensemble Kalman Filter framework well as far as I can assess this. A detailed discussion of the application of the model to an example wind farm with multiple well-done plots shows the capabilities and weaknesses of the framework.

I suggest major revisions with some comments I have that could help non-expert readers to

1. I suggest a change of the title to include the new adaptations (from the title, I did not expect this work to be heavily on the modeling side)

2. In the introduction, ll. 35, could you indicate a range where the free-vortex method is appropriate? From what I understand, it seems to be useful only for up to 4D downstream?

3. When describing different approaches in the introduction, please emphasize where you are talking about flow field estimation within the wind farm: You motivate it by wind farm control (WFC), then mention methods already used in WFC for steady state applications and switch to dynamic effects but it does not become clear that the flow field is modeled and that wake models are used for this. This applies also for the next paragraphs.

4. L. 307: should it be power converted to wind speed?

5. L. 380: no comma behind “both [the reference system…]”

6. L. 502: no comma behind “both [the measurement and …]”

7. Ll. 546: I can see an effect for T4 with the crossing of T2 at OP 40 or T3 with the wake of T2 in OP 70, but not “the OPs of T3 just passed T2 prior to the snapshot, which leads to a lower variance around OP 70 in the wake of T3.” ?

8. Figure 11:

a. The position of the axis labels in the middle is confusing

b. I am not sure what I am seeing in the top left plot (i.e., how the model works): Do I see how the wind speed at the turbine is propagated downstream but without the wake model? (Since there is not a lot of change of the velocity) and is the wake model “folded” into this by means of the weighting functions in the middle plot? Please indicate this in the main text.

9. You could comment somewhere that P ~u^3 which makes a precise estimation of u even more crucial since the error has a big impact on the power estimation

10. Figure 13/ comment in ll. 582: I don’t understand what you mean by “Note also, that the time window has been reduced to the second half of the simulation to allow for a clearer picture”: Should this be marked by the vertical lines? How can I see this in figure 13?

11. The journals are missing in the list of references

12. Reference 31: information missing

Author Response

Dear reviewer,

thank you very much for taking the time to read through the manuscript and to give your opinion and feedback. Attached you will find our responses to your comments, but also to the ones of the three other reviewers.

We hope to have addressed all points adequately.

With best regards,

Marcus Becker

Author Response File: Author Response.pdf

Reviewer 2 Report

Review of article "Ensemble based flow field estimation using the dynamic wake model FLORIDyn" by M. Becker and co-workers.

The paper is well written, scientifically sound and relevant to the wind farm wake and control community. It is an interesting article since the dynamic version can potentially capture large scale dynamic flow variation (aka topological changes, atmospheric larg- scale variation not Turbulence) which the static models cannot.

A few concerns and checks suggested for the authors:

1) Check typo in Eq. 6. I think the subscripts should be cw and dw unlike dw for both.

2) I do not fully capture, the reason to perform validation for Section 3.2. Is EnKF used only on Floridyn itself? And would this mean, that the measurements are Floridyn itself and the predictive step in EnKF is Floridyn OP-updates, which means that EnKF is just behaving as a low-pass filter here. If I understand this correctly, -- then I am a bit surprised to keep the EnKF in the validation study. Can the author's explain this. Additionally, With EnKF from Floridyn, there wiill be modification of the states, so the difference between EnKF and Reference is understandable -- but can it be claimed that the EnKF in this case is of higher fidelity that the Floridyn only update. Perhaps this case can be called a EnKf framework sanity check. Sec 3.3 with SOWFA would serve as a more reasonable test of EnKF filter with the measurements: really the SOWFA "LES" Simulations.

3. In the validation section 3.2, without yaw misalignment I assume, the turbine yaw control is such that the turbine yaw changes at the same rate as the change in wind turbine direction?

4. I am curious, How does EnKF impact the TI% turbines in Floridyn simulation. The TI% in SOWFA is 6%, but when EnKF assimilated, does the fluctuating velocity field in ENKF - Floridyn reflect the TI at similar level ~ 6%. This would be an interesting thing to observe, if the Floridyn estimates are used for loads modelling in future.

5. Has any validation been done with existing dynamic meandering models for transient simulations? I appreciate the author's claims that Floridyn is fast algorithm, since it is using updation of the OP states, like conventional DWM, but did they actually comare with existing DWM models (point to previous papers, if any?).

Author Response

Dear reviewer,

We hope to have addressed all points adequately.

With best regards,

Marcus Becker

Author Response File: Author Response.pdf

Reviewer 3 Report

In this manuscript, the authors have presented Ensemble based flow field estimation using the dynamic wake model FLORIDyn. Although the authors have put a lot of effort and presented the results and discussion professionally. However, this article seriously lacks a problem formulation section. There are too many variables in most places authors refer the reader to different references. I think each variable must be explained in a proper way and the manuscript should give information as a whole.

Author Response

Dear reviewer,

We hope to have addressed all points adequately.

With best regards,

Marcus Becker

Author Response File: Author Response.pdf

Reviewer 4 Report

1. The introductory statement of the abstract need to be rephrased "wind turbines are positioned in a wind farm in such a way as to minimize...." or something like that

2. the statement "More recent publications iterate on the approach", cite the references here

3. Which is the reference or baseline method against which your method was compared?

4. In the conclusion section, it talks about control, is this approach used for the planning stage/ installation of wind farms or during the operational stage/control stage?

Author Response

Dear reviewer,

We hope to have addressed all points adequately.

With best regards,

Marcus Becker

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

My suggestions have been taken into account, I recomment publication.

Reviewer 3 Report

The authors have incorporated the required changes. The manuscript is recommended for the acceptance.