Control Simulation Experiments of Extreme Events with the Lorenz-96 Model
- 1Data Assimilation Research Team, RIKEN Center for Computational Science (R-CCS), Kobe, 650-0047, Japan
- 2Graduate School of Mathematics, Nagoya University, Nagoya, 464-8601, Japan
- 3Prediction Science Laboratory, RIKEN Cluster for Pioneering Research, Kobe, 650-0047, Japan
- 4RIKEN Interdisciplinary Theoretical and Mathematical Sciences Program (iTHEMS), Kobe, 650-0047, Japan
- 1Data Assimilation Research Team, RIKEN Center for Computational Science (R-CCS), Kobe, 650-0047, Japan
- 2Graduate School of Mathematics, Nagoya University, Nagoya, 464-8601, Japan
- 3Prediction Science Laboratory, RIKEN Cluster for Pioneering Research, Kobe, 650-0047, Japan
- 4RIKEN Interdisciplinary Theoretical and Mathematical Sciences Program (iTHEMS), Kobe, 650-0047, Japan
Abstract. Control Simulation Experiment (CSE) is a recently developed approach to investigate the controllability of dynamical systems, extending the well-known Observing Systems Simulation Experiment (OSSE) in meteorology. For effective control of chaotic dynamical systems, it is essential to exploit the high sensitivity to initial conditions for dragging a system away from an undesired regime by applying minimal perturbations. In this study, we design a CSE for reducing the number of extreme events in the Lorenz-96 model. The 40 variables of this model represent idealized meteorological quantities evenly distributed on a latitude circle. The reduction of occurrence of extreme events over 100 years runs of the model is discussed as a function of the parameters of the CSE: the ensemble forecast length for detecting extreme events in advance, the magnitude and localization of the perturbations, and the quality and coverage of the observations. The design of the CSE is aimed at reducing weather extremes when applied to more realistic weather prediction models.
Qiwen Sun et al.
Status: final response (author comments only)
-
RC1: 'Comment on npg-2022-12', Anonymous Referee #1, 25 Sep 2022
The manuscript is nicely exploring the possibility to perturb a simulation (that would represent a real system) to reduce the risk of extreme events. This is explored in the context of the Lorenz-96 model, with perturbations built on the difference between members of an ensemble forecast. The authors have carefully design experiments in doing so and they show that a considerable reduction of the occurrence of extremes is notable by using such an approach. They further explore the impact of different parameters like the lead time of the forecast and the amplitude of the perturbations, as well as the possibility to perturb locally the system and the number of observations. This is a well written and very clear thought experiment that is worth publishing in Nonlinear Processes in Geophysics. Here a few points that are worth addressing at the time of revision.
One key aspect of control is the energy to be introduced in the system. The authors have here computed such an energy in the context of the Lorenz system, but it would be important to give a first clue to what quantity of energy would be necessary in a more realistic setting. As the Lorenz-96 model provides a toy model of the large scale variables at a specific latitude, it would be very interesting to convert the energy needed in an energy that the meteorological community could apprehend (power, work…) and discuss that in the conclusions.
When perturbing a system (as done for instance with the increase of CO2), there are extremes that become less frequent like for instance a reduction of cold waves in certain regions with the increase of the global temperature. But this has other effects with an increase of heat waves. If one transposes this to the current setting, some extremes are suppressed, but some others might be arising. Did you see such type of situations in the context of the Lorenz model? In any case it is necessary to elaborate on this somewhere in the manuscript.
In Figure 9, the authors show a saturation of the number of actions as a function of the localization scale. I am wondering whether it is related to the spatial correlation of the perturbations needed. Furthermore I am wondering what is the nature of the global perturbations. Do they look like bred modes? It would be really interesting to elaborate on that aspect.
Minor aspect
- The last paragraph of the introduction should be placed in the conclusions
- Line 126. There is no Appendix in the document
-
RC2: 'Review of npg-2022-12: Control Simulation Experiments of Extreme Events with the Lorenz-96 Model by Sun et al.', Anonymous Referee #2, 13 Jan 2023
The authors propose a method (Control Simulation Experiment) to control the number of extreme events in the Lorenz-96 model. The sensitivity of this technique to the choice of various parameters, such as the forecast lead time in which extreme events are detected or the amplitude of the perturbations. The authors find that the method is effective to reduce extremes and the sensitivity tests show how the perturbations can be tuned in order to reduce the high number of extreme events with a minimized action over the system.
General comment:
The manuscript is clear, the technical details of the experiments performed are well described. The authors provide several references to contextualize their research. This manuscript can be of interest for NPG readers, however my main concern is about the implications of this study in a more realistic context. The Control Simulation Experiment is aimed at reducing the extremes, but the challenge is that the models are able to simulate these extremes and in case of ensemble forecasting how the ensemble should be designed to include extremes. Therefore, I do not see the benefit of reducing the extremes in a simulation, when those states actually take place in the system that the model represents. I suggest including some clarifications in the introduction in this regard as well as in the conclusions to better express the general objective of this research.
Specific comments:
L11: “of the first two authors” can be removed.
L34-35: In line with my general comment, what is the benefit of reducing simulated weather extremes that occur in reality?
L149: (j,j) -> (i,j)
L170: Why only the maximum value is used to define extremes and not the minimum?
L199: When the procedure to generate the perturbation vectors is described and the selection of the ensemble member B is explained, the alternative computation in case an ensemble member B is not found (L208-209) should be indicated here.
L281: It is more correct saying similar or approximately equal instead of equal.
L292-293: The meaning of this sentence is not clear.
Fig.10: Please, indicate the meaning of the triangle in the caption.
-
EC1: 'Comment on npg-2022-12', Natale Alberto Carrassi, 17 Jan 2023
Dear Authors
we have now received two very useful and positive reports about your works. I think they both address aspects of your work that can be improved, particularly in relation to the applicabilty.
I encourage you to take them into account while preparing the revised manuscript.
Please send first detailed responses to the Reviewers' quests.
Best Regards
Qiwen Sun et al.
Qiwen Sun et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
495 | 76 | 10 | 581 | 6 | 4 |
- HTML: 495
- PDF: 76
- XML: 10
- Total: 581
- BibTeX: 6
- EndNote: 4
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1