the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 12 Oct 2022
A method for generating a quasi-linear convective system suitable for observing system simulation experiments
Louis J. Wicker
Abstract. To understand the impact of different assimilated observations on convection-allowing model forecast skill, a diverse range of observing system simulation experiment (OSSE) case studies are required (different storm modes and environments). Many previous convection-allowing OSSEs predicted the evolution of an isolated supercell generated via a warm air perturbation in a horizontally homogenous environment. This study introduces a new methodology where a quasi-linear convective system is generated in a highly-sheared and modestly unstable environment. Wind, temperature, and moisture perturbations superimposed on a horizontally homogeneous environment simulate a cold front that initiates an organized storm system that spawns multiple mesovortices. Mature boundary layer turbulence is also superimposed onto the initial environment to account for typical convective scale uncertainties.
Creating an initial forecast ensemble remains a challenge for convection-allowing OSSEs because mesoscale uncertainties are difficult to quantify and represent. The generation of the forecast ensemble is described in detail. 24 hour full-physics simulations (e.g., radiative forcing, surface friction, microphysics) initialize the forecast ensemble. The simulations assume different surface conditions to alter surface moisture and heat fluxes and modify the effects of friction. The subsequent forecast ensemble contains robust non-gaussian errors that persist until corrected by the data assimilation system. An example OSSE suggests a combination of radar and conventional (surface and soundings) observations are required to produce a skilled quasi-linear convective system forecast, which is consistent with real case studies. The OSSE framework introduced in this study will be used to understand the impact of assimilated environmental observations on forecast skill.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(3863 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3863 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Jonathan D. Labriola and Louis J. Wicker
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1033', Anonymous Referee #1, 10 Nov 2022
This paper presents a new setup for an OSSE assimilation system. The main novelties are the nature run (front convection) and the ensemble generation. I like very much the choice of the nature run but I have some questions regarding the initial ensemble. The ensemble it is too far from the nature run (see below). I also think that adding results of the free ensemble (without assimilation) would make the results more interesting. For that reasons I recommend major revision.
1) The initial ensemble seems quite far from the nature run as it is shown in Figure 7, where the nature run profiles are often far from the ensemble spread. I think this is a result of the generation ensemble, ,the ensmbl ewhich has run for over 24 h (with full physics) in comparison with the nature run. These are difficult conditions for an assimilation system, and I do not think that these are quite realistic (the errors in the cold front are ~4K). I wonder how the nature run would be if the same procedure would have been applied (run each part of the front independently for 24 hours).
I guess that to have a bad ensemble is the author’s choice, but this should be clear in the abstract or introduction: not only that the errors are non-gaussian, but also that the nature run profile is often out of the ensemble spread with the consequence of a much weaker convection in all ensemble members.
2) In figures 11, 12,13, 14 and 15 would it be very useful to plot the skill of the initial ensemble without data assimilation. We could then see the skill of the assimilation system. In figure 13 you could also plot the nature run.
3) It would be interesting to see how realistic the prediction of reflectivity is. I suggest plotting the reflectivity of one member (without assimilation, CTRL and ENVI) at assimilation time and after 3h forecast.
4) The generation of turbulence in the nature run is not clear. I do not understand why radiation and surface friction are switched off, when these are the main driver for boundary-layer turbulence. I also do not understand why you want to run the simulation for 12 hours, which is much longer than any boundary layer process
Minor:
1) Equation 1. Set the appropriate limits. The equation is only true if the argument is larger than 0.
2) Equations 1,2,3: the letter f is used for two different functions. This is confusing.
3) Please specify if the simulations to generate the ensemble are done with the nature-run physics or with the forecast physics.
4) Please specify which cloud cover scheme (if any) is used in the nature run and forecast physics. Also specify if the turbulent schemes are 1D or 3D.
5) Line 265: I guess there is a not missing.
6) I cannot see the soundings in Figure 10.
7) Do you see much updraft helicity in the forecast?
8) OSSE assimilation systems can also be used for improving the assimilation system independently of the errors in the model physics (see for example Zeng et al. 249, 105282, 2021,https://doi.org/10.1016/j.atmosres.2020.105282)
9) Figure 11. Innovation is proably not the right label for the y-axis.
Citation: https://doi.org/10.5194/egusphere-2022-1033-RC1 - AC1: 'Reply on RC1', Jonathan Labriola, 13 Feb 2023
-
RC2: 'Comment on egusphere-2022-1033', Anonymous Referee #2, 21 Nov 2022
In this manuscript, the technique used to generate a nature run that is representative of a tornado outbreak in the southeastern United States is introduced. Since past studies conduct OSSEs to simulate evolution of supercecullar convections by generating a “warm bubble” into an unstable and highly sheared environment, it is meaningful to perform idealized OSSE that simulates the evolution of a convective line initiated via a frontal boundary in a highly-sheared and modestly-unstable environment. Creating OSSEs that simulate different storm modes and environments can help better understand how assimilated observations impact the environment and the subsequent evolution of convection. Forecasts that assimilate radar and environmental observations are found to be more skillful than assimilating radar data only. Environmental observations help correct wind profile error and increase convergence. The authors introduce a new method to create initial ensembles, however, it is not addressed very clearly. I recommend accepting this paper after a minor revision.
Minor comments:
- The first question is about the nature run. It is not very clear what the final setup of the nature run is. It looks to me the nature run is initialized from the environmental sounding shown in Fig. 1a. A frontal boundary is added to provide a mechanical forcing for convection initialization. A turbulence simulation is conducted to help introduce more realistic eddies. Then the perturbations of the u,v,w, qv, and theta fields from the 12-h forecasts of the turbulence simulation are added back to the initial condition of the nature run. I think it is better to specify more clearly what the final setup of nature run is before section 2.4, similar to the setup descriptions (step 1 to 4) in section 3.
- It is unclear to me how the 40 initial ensembles are generated.
It is mentioned in 200, cold and warm sector simulations for each ensemble member are assigned a land surface type. How many warm and cold sector simulations are conducted? Based on the captions in Fig. 7, it looks to me there is only one warm sector simulation and one cold sector simulation (the ones that use the unperturbed sounding). Are the 24-h forecasts from cold and warm sector simulations blended together in different ways (with different times and locations) to form different initial cold front boundary for 40 different ensemble members? In line 360, it is mentioned that each forecast member is initialized from the same sounding. Is it the same sounding as the nature run? What are the perturbed soundings in Fig. 7 used for? Are there 40 perturbed soundings in warm sectors and another 40 ones in cold sectors? It is better to summarize the setup descriptions of the initial ensembles before section 3.3 instead at the beginning of the section 3. Summarize what are the difference in the 40 initial ensembles (e.g., do they use the same or different sounding, cold front boundary, land surface type, and potential temperature perturbations, etc., for simulation?). The timeline is also not very clear to me. What is the time setup for nature and the runs to generate initial ensembles? Fig. 9 only shows the time setup after generation of the initial ensembles.
- Line 235: should be “Due to the idealized nature”.
- Usually for OSSEs, model forecasts are verified against the true state instead of the observations due to the errors of the observations. Did you conduct verification using observations or the “true” state from nature run? If using the latter, RMSE instead of RMSI should be used.
- It seems assimilating convectional observations together with radar data produces much stronger updraft relative to assimilating radar data alone. Environmental observations help correct wind profile error and increase convergence. Which one do you think have more influence on the analysis of the updraft, sounding, or surface observations?
Citation: https://doi.org/10.5194/egusphere-2022-1033-RC2 - AC2: 'Reply on RC2', Jonathan Labriola, 13 Feb 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1033', Anonymous Referee #1, 10 Nov 2022
This paper presents a new setup for an OSSE assimilation system. The main novelties are the nature run (front convection) and the ensemble generation. I like very much the choice of the nature run but I have some questions regarding the initial ensemble. The ensemble it is too far from the nature run (see below). I also think that adding results of the free ensemble (without assimilation) would make the results more interesting. For that reasons I recommend major revision.
1) The initial ensemble seems quite far from the nature run as it is shown in Figure 7, where the nature run profiles are often far from the ensemble spread. I think this is a result of the generation ensemble, ,the ensmbl ewhich has run for over 24 h (with full physics) in comparison with the nature run. These are difficult conditions for an assimilation system, and I do not think that these are quite realistic (the errors in the cold front are ~4K). I wonder how the nature run would be if the same procedure would have been applied (run each part of the front independently for 24 hours).
I guess that to have a bad ensemble is the author’s choice, but this should be clear in the abstract or introduction: not only that the errors are non-gaussian, but also that the nature run profile is often out of the ensemble spread with the consequence of a much weaker convection in all ensemble members.
2) In figures 11, 12,13, 14 and 15 would it be very useful to plot the skill of the initial ensemble without data assimilation. We could then see the skill of the assimilation system. In figure 13 you could also plot the nature run.
3) It would be interesting to see how realistic the prediction of reflectivity is. I suggest plotting the reflectivity of one member (without assimilation, CTRL and ENVI) at assimilation time and after 3h forecast.
4) The generation of turbulence in the nature run is not clear. I do not understand why radiation and surface friction are switched off, when these are the main driver for boundary-layer turbulence. I also do not understand why you want to run the simulation for 12 hours, which is much longer than any boundary layer process
Minor:
1) Equation 1. Set the appropriate limits. The equation is only true if the argument is larger than 0.
2) Equations 1,2,3: the letter f is used for two different functions. This is confusing.
3) Please specify if the simulations to generate the ensemble are done with the nature-run physics or with the forecast physics.
4) Please specify which cloud cover scheme (if any) is used in the nature run and forecast physics. Also specify if the turbulent schemes are 1D or 3D.
5) Line 265: I guess there is a not missing.
6) I cannot see the soundings in Figure 10.
7) Do you see much updraft helicity in the forecast?
8) OSSE assimilation systems can also be used for improving the assimilation system independently of the errors in the model physics (see for example Zeng et al. 249, 105282, 2021,https://doi.org/10.1016/j.atmosres.2020.105282)
9) Figure 11. Innovation is proably not the right label for the y-axis.
Citation: https://doi.org/10.5194/egusphere-2022-1033-RC1 - AC1: 'Reply on RC1', Jonathan Labriola, 13 Feb 2023
-
RC2: 'Comment on egusphere-2022-1033', Anonymous Referee #2, 21 Nov 2022
In this manuscript, the technique used to generate a nature run that is representative of a tornado outbreak in the southeastern United States is introduced. Since past studies conduct OSSEs to simulate evolution of supercecullar convections by generating a “warm bubble” into an unstable and highly sheared environment, it is meaningful to perform idealized OSSE that simulates the evolution of a convective line initiated via a frontal boundary in a highly-sheared and modestly-unstable environment. Creating OSSEs that simulate different storm modes and environments can help better understand how assimilated observations impact the environment and the subsequent evolution of convection. Forecasts that assimilate radar and environmental observations are found to be more skillful than assimilating radar data only. Environmental observations help correct wind profile error and increase convergence. The authors introduce a new method to create initial ensembles, however, it is not addressed very clearly. I recommend accepting this paper after a minor revision.
Minor comments:
- The first question is about the nature run. It is not very clear what the final setup of the nature run is. It looks to me the nature run is initialized from the environmental sounding shown in Fig. 1a. A frontal boundary is added to provide a mechanical forcing for convection initialization. A turbulence simulation is conducted to help introduce more realistic eddies. Then the perturbations of the u,v,w, qv, and theta fields from the 12-h forecasts of the turbulence simulation are added back to the initial condition of the nature run. I think it is better to specify more clearly what the final setup of nature run is before section 2.4, similar to the setup descriptions (step 1 to 4) in section 3.
- It is unclear to me how the 40 initial ensembles are generated.
It is mentioned in 200, cold and warm sector simulations for each ensemble member are assigned a land surface type. How many warm and cold sector simulations are conducted? Based on the captions in Fig. 7, it looks to me there is only one warm sector simulation and one cold sector simulation (the ones that use the unperturbed sounding). Are the 24-h forecasts from cold and warm sector simulations blended together in different ways (with different times and locations) to form different initial cold front boundary for 40 different ensemble members? In line 360, it is mentioned that each forecast member is initialized from the same sounding. Is it the same sounding as the nature run? What are the perturbed soundings in Fig. 7 used for? Are there 40 perturbed soundings in warm sectors and another 40 ones in cold sectors? It is better to summarize the setup descriptions of the initial ensembles before section 3.3 instead at the beginning of the section 3. Summarize what are the difference in the 40 initial ensembles (e.g., do they use the same or different sounding, cold front boundary, land surface type, and potential temperature perturbations, etc., for simulation?). The timeline is also not very clear to me. What is the time setup for nature and the runs to generate initial ensembles? Fig. 9 only shows the time setup after generation of the initial ensembles.
- Line 235: should be “Due to the idealized nature”.
- Usually for OSSEs, model forecasts are verified against the true state instead of the observations due to the errors of the observations. Did you conduct verification using observations or the “true” state from nature run? If using the latter, RMSE instead of RMSI should be used.
- It seems assimilating convectional observations together with radar data produces much stronger updraft relative to assimilating radar data alone. Environmental observations help correct wind profile error and increase convergence. Which one do you think have more influence on the analysis of the updraft, sounding, or surface observations?
Citation: https://doi.org/10.5194/egusphere-2022-1033-RC2 - AC2: 'Reply on RC2', Jonathan Labriola, 13 Feb 2023
Peer review completion
Journal article(s) based on this preprint
Jonathan D. Labriola and Louis J. Wicker
Data sets
A Method for Generating a Quasi-Linear Convective System Suitable for Observing System Simulation Experiments: Dataset Jonathan Labriola and Louis Wicker https://zenodo.org/record/7126769#.Yz8RwuzMI88
Model code and software
A Method for Generating a Quasi-Linear Convective System Suitable for Observing System Simulation Experiments Jonathan Labriola and Louis Wicker https://zenodo.org/record/7109050#.Yz8ReezMI88
Jonathan D. Labriola and Louis J. Wicker
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 366 | 80 | 16 | 462 | 4 | 4 |
- HTML: 366
- PDF: 80
- XML: 16
- Total: 462
- BibTeX: 4
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3863 KB) - Metadata XML