the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
A new lightning scheme in Canada's Atmospheric Model, CanAM5.1: Implementation, evaluation, and projections of lightning and fire in future climates
Abstract. Lightning is an important atmospheric process for generating reactive nitrogen, resulting in production of tropospheric ozone, as well as igniting wildland fires, which result in potentially large emissions of many pollutants and short-lived climate forcers. Lightning is also expected to change in frequency and location with the changing climate. As such, lightning is an important component of Earth system models. Until now, the Canadian Earth System Model (CanESM) did not contain an interactive lightning parameterization. The fire parameterization in CanESM5.1 was designed to use prescribed monthly climatological lightning. In this study, we have added a logistical regression lightning model that predicts lightning occurrence interactively based on three environmental variables and their interactions into CanESM5.1’s atmospheric model, CanAM5.1, creating the capacity to interactively model lightning, allowing for future projections under different climate scenarios. The modelled lightning and resulting burned area were evaluated against satellite measurements over the historical period and model biases were found to be acceptable. Modelled lightning was within a factor of two of the measurements and had exceptionally accurate land/ocean ratios.
The modified version of CanESM5.1 was used to simulate two future climate scenarios (SSP2-4.5 and SSP5-8.5) to assess how lightning and burned area change in the future. Under the higher emission scenario (SSP5-8.5), CanESM5.1 predicts an increase in northern mid-latitude lightning flashrate of 5 %, but a decrease in tropical lightning of -10 %, resulting in almost no change to the global mean lightning amount by the end-of-the century (2081–2100 vs 2015–2035 average). By century’s end, the change in global total burned area with prescribed climatological lightning was about two times greater than that with interactive lightning (43 % vs 19 % increase, respectively). Conversely, in the northern mid-latitudes the use of interactive lightning resulted in three times more area burned as that with unchanging lightning (36 % vs 13 % increase, respectively). These results show that the future changes to burned area are greatly dependent on a model’s lightning scheme, both spatially and overall.
- Preprint
(2058 KB) - Metadata XML
-
Supplement
(468 KB) - BibTeX
- EndNote
Status: open (until 18 May 2024)
-
RC1: 'Comment on gmd-2024-24', Anonymous Referee #1, 18 Mar 2024
reply
A new lightning scheme in Canada's Atmospheric Model, CanAM5.1: Implementation, evaluation, and projections of lightning and fire in future climates
General comments:
There were no lightning schemes (no lightning calculated) in Canda's Earth System Model 5.1 before this study, which introduced a lightning scheme into CanESM5.1. The Etten-Bohm scheme is based on a logistic relation between lightning occurrence and multiple environmental variables (CAPE, LCL, r), which is unique among existing lightning schemes. The Etten-Bohm scheme is expected to show a relatively acceptable performance in CanESM, but unfortunately, due to the large biases in CAPE, LCL, and r simulations in CanESM, the lightning simulation accuracy cannot be considered satisfactory. I am not sure about the statistical metrics such as R and RMSE (root mean squared error) values between lightning simulations and observations in this study, but previous studies achieved better lightning prediction accuracy in their models (Finney et al., 2014; Lopez, 2016; He et al., 2022).
Is this paper valuable for publication? Yes, this is the first time the authors introduced a new lightning scheme to CanESM. However, I considered the following aspects that may improve the scientific significance of this paper:
- If possible, try to implement more existing lightning schemes (as you listed in L35-L45) into CanAm5.1 and compare those lightning schemes with the Etten-Bohm scheme. The benefit of doing so is that you can evaluate how different lightning schemes influence the prediction accuracy to what degree. By comparing several different lightning schemes in a New ESM, you can provide more information for the scientific community. I also recommend that you consider the uncertainties that existed in observations when evaluating lightning schemes (you can refer to (He et al., 2022) Figure 4). Even you can use a second ESM (such as CESM) to provide additional supporting information (for example, if CESM can better represent CAPE, LCL, how these improvements can influence the prediction accuracy of lightning).
- I am not sure whether if you applied meteorological nudging (u, v, T) in your simulations, by the way, you need a detailed experiment setup in your paper (use a chart). You can evaluate how meteorological nudging influences the prediction accuracy of lightning in the Etten-Bohm scheme by turning on/off nudging. For example, if meteorological nudging can significantly improve the simulation accuracy of CAPE et al. as well as lightning, this can prove that improving the simulation of CAPE et al. in CanESM can definitely lead to the improvement of lightning simulations (and the extent of this improvement to which degree).
- What caused the large biases in CAPE, LCL, r? → Please provide some explanations in your paper. You mentioned you will use a new TKE deep convection scheme, if possible, please test this new convection scheme and estimate how it impacts the simulations of lightning and burned area.
- You conducted future projections; however, I recommend that you can firstly evaluate the response of global lightning activities to short-term surface warming (1993-2013) of Etten-Bohm scheme in CanESM. Evidence shows that there was no statistically significant trend in global lightning activities during 1993-2013 (LIS/OTD, (Williams, 1992)).
- You predicted a decreasing lightning trend with global warming (Table2), what is the reason (CAPE, LCL, r decreased?)? He and Sudo (2023) suggested that historical global warming enhanced lightning activities, but increases in aerosol burden exerted an opposite effect (1960-2014). Can you also separate the effects of warming and aerosols?
- "Control lightning" vs. "interactive lightning", what is the implication? The simulated lightning trends can largely impact the simulated burned area, but this is not a new finding.
Anyway, please try to improve the scientific significance.
Specific comments:
- Please show statistical metrics (R, RMSE, MBE) between simulations and observations.
- He et al. (2022) recently developed a new lightning scheme based on Lopez (2016) and McCaul et al. (2009). This paper can provide additional information for your paper's introduction (L32, L35-L45, L48-49).
- L104, Uman (1986) only mentioned a blurry concept, could you please provide the detailed equations (to calculate the fraction) and relevant explanations. Another widely used equation for calculating cloud-to-ground fraction was proposed by Princ and Rind 1993. I am not sure which one is better but for your reference.
- Which cumulus convection scheme is used? You need to add a detailed experiment setup into your paper.
- L159 over some parts of the western …
- L174, But from about 30S to 50N, the zonal pattern is modelled correctly → The lightning is systematically underestimated in the model? Which parameters mostly contributed to this systematic bias?
- L171, what does TKE represent?
- L198, Figure S3 and Figure3, it looks like there is a rectangular dark blue box over polar regions, which is weird. It looks like there are bugs in your computer program, please check and justify this situation.
- Figure 6 and Figure 11. The figures are blurry, please provide figures with at least 300 dpi.
- Figure S2, model largely underestimated CAPE within low-latitude regions, could you please explain it?
- Figure S5, model can partially capture the spatial pattern of r but systematically overestimte r compared to MERRA-2, what is the reason?
Technical corrections
Supplement, Figure S0, in the following sentence:
uppermost level (zt) may be reached by moists onvection.
Change "onvection" to "convection".
Reference
Finney, D. L., Doherty, R. M., Wild, O., Huntrieser, H., Pumphrey, H. C., and Blyth, A. M.: Using cloud ice flux to parametrise large-scale lightning, Atmospheric Chemistry and Physics, 14, 12665–12682, https://doi.org/10.5194/acp-14-12665-2014, 2014.
He, Y. and Sudo, K.: Historical (1960–2014) lightning and LNOx trends and their controlling factors in a chemistry–climate model, Atmospheric Chemistry and Physics, 23, 13061–13085, https://doi.org/10.5194/acp-23-13061-2023, 2023.
He, Y., Hoque, H. M. S., and Sudo, K.: Introducing new lightning schemes into the CHASER (MIROC) chemistry–climate model, Geoscientific Model Development, 15, 5627–5650, https://doi.org/10.5194/GMD-15-5627-2022, 2022.
Lopez, P.: A lightning parameterization for the ECMWF integrated forecasting system, Monthly Weather Review, 144, 3057–3075, https://doi.org/10.1175/MWR-D-16-0026.1, 2016.
McCaul, E. W., Goodman, S. J., LaCasse, K. M., and Cecil, D. J.: Forecasting lightning threat using cloud-resolving model simulations, Weather and Forecasting, 24, 709–729, https://doi.org/10.1175/2008WAF2222152.1, 2009.
Williams, E. R.: The schumann resonance: A global tropical thermometer, Science, 256, 1184–1187, https://doi.org/10.1126/science.256.5060.1184, 1992.
Citation: https://doi.org/10.5194/gmd-2024-24-RC1 -
AC1: 'Initial reply on RC1', Cynthia Whaley, 12 Apr 2024
reply
Thank you for your thorough review of the paper. We plan to respond fully point by point and provide a revised manuscript once we hear from a second reviewer. However, while we wait for a second review, we would like to respond to a few issues raised by the first reviewer now:
1. Regarding the lightning biases due to biases in CanESM's CAPE and r (LCL bias was minimal) impacting the lightning results:
While the Etten-Bohm et al (2021) lightning parameterization isn’t perfect (none are), we feel that the lightning results using the scheme sufficiently capture the latitudinal and seasonal variations necessary to assess wildfire burned area in present day and future climate scenarios simulated by CanESM. The results further show an excellent land/ocean discrimination of lightning occurrence. It is true that CanESM has some differences in basic state parameters compared to MERRA-2, but since the CAPE, LCL, and r inputs are standardised (e.g., (CAPE - meanCAPE)/standard_dev(CAPE)) before being input to the lightning calculation, the influence of their systematic bias is minimised, which is why the lightning results have a small error compared to ISS LIS results, despite a systematic offset. This was mentioned in Section 3.2 of the original manuscript, but we will have clarified that approach in the revised manuscript in the methods section (Sec 2.1). We note that all General Circulation Models will experience basic state and cloud variable biases compared to the real atmosphere, but that the standardisation minimises these issues, motivating the use of this parameterization.
In terms of the accuracy of lightning prediction compared to other parameterizations, Fig. 1 from Clark et al. 2017 (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2017GL073017) shows that either other lightning parameterizations implemented in CAM5 experience a wide range of predictions, with a notable struggle in representing a realistic land/ocean contrast. Only the Price & Rind parameterization had realistic land-ocean contrast. However the Price & Rind cloud-top-height lightning parameterization is not good for future lightning projections, as shown in Finney et al (2018). Furthermore, Finney et al. (2014) further show a wide range in statistical accuracy when a suite of lightning parameterizations were applied to ERA-I data. In the revision, we plan to add more statistical metrics to the evaluation so that the reader can better evaluate the performance of the parameterization.
Refs:
Clark, Ward, and Mahowald: Parameterization-based uncertainty in future lightning flash density, GRL, 2017, https://doi.org/10.1002/2017GL073017Finney, Doherty, Wild et al: A projected decrease in lightning under climate change, Nature Climate Change, 2018, https://doi.org/10.1038/s41558-018-0072-6
Finney, D. L., Doherty, R. M., Wild, O., Huntrieser, H., Pumphrey, H. C., and Blyth, A. M.: Using cloud ice flux to parametrise large-scale lightning, Atmos. Chem. Phys., 14, 12665–12682, 2014, https://doi.org/10.5194/acp-14-12665-2014
2. Regarding the suggestion to implement multiple lightning parameterizations in CanESM to provide information for the scientific community:
While it would be a valuable exercise, it was not the purpose or in the scope of this paper to test several lightning schemes in CanESM. The main goal of our study was to test the implementation of this new parameterization and show how including interactive lightning affects burned area compared to just assuming climatological lightning. We note that the Clark et al (2017) study mentioned above included evaluation of multiple lightning parameterizations in CAM5, therefore that information is there for the scientific community already.
3. Regarding the suggestion to do nudged simulations to possibly improve lightning results:
The simulations were not nudged (and this will be clarified in the revision). In CanESM5, nudging does not directly target the variables that are of most interest (i.e., we do not nudge CAPE, LCL, or r), and we typically find that nudging degrades cloud properties and precipitation in the model, even though temperatures, winds, and humidity are improved. This is because cloud and convection parameterizations have been developed and tuned using observational constraints using un-nudged simulations. Typically, the skill of convection parameterizations in global models is assessed through comparisons of precipitation patterns.
4. Regarding the historical trends in lightning and the suggestion that including future projections was not warranted:
Thank you for the suggestion to provide an evaluation of the 1993-2013 time series. This will be assessed for the revised paper. Note that Qie et al. (2020, Atmos. Res.) use LIS observations to show statistically significant decreasing trends from 1996-2013 in parts of the tropics (Fig. AR1a attached), and other regional studies, like Chakraborty et al (2021) corroborate those lightning trends (e.g. increase in India for 1998-2014), and those spatial patterns in lightning trends are consistent with what we simulate for 2015-2100 (Fig. AR1b attached). Our overall future lightning trends also agree well with those from the sophisticated ice flux scheme in Finney et al. (2018), and that in Etten-Bohm et al. (in review) using the same lightning scheme in CAM5 (Fig. AR1b attached). Therefore, in the context of a climate model, this type of future trend evaluation is important for the goals of this paper.
Refs
Qi et al: Regional trends of lightning activity in the tropics and subtropics, Atmos. Res. 242, 11, 104960, 2020.
Chakraborty et al: Lightning occurrences and intensity over the Indian region: long-term trends and future projections, ACP, 21, 11161-11177, 2021.
5. Regarding improving the scientific significance:
The scientific significance of our study is that we used a new lightning scheme that doesn’t depend on highly uncertain cloud or precipitation variables in our climate model and got good results, including excellent ocean-land lightning gradients, and a similar response to climate warming as a process-based ice flux lightning scheme. We also quantify and emphasize the significant importance that lightning has on burned area now and in the future, which is not intuitive, given that fire depends on several other factors, such as fire weather (temperature, RH, winds), which is more prominent in the literature. The significance of our study has been emphasised in our responses above, in the original manuscript, and will be further highlighted in the revised manuscript.
-
CC1: 'Reply on AC1', Yanfeng He, 23 Apr 2024
reply
Thank you very much for your initial reply.
The main issue I was concerned about is that your lightning scheme does not capture the observed latitudinal lightning distribution (please see Figure 2a). There should be a higher occurrence of lightning in tropical regions compared to middle- to high-latitude regions. Although you mentioned that you standardized CAPE, etc., to minimize model biases, you should acknowledge that there are still large model biases in the simulated lightning occurrence (Figure 2a). In Figure 5a, you should simulate a peak within low-latitude regions; it is unrealistic for the simulated lightning density to be higher in high-latitude regions than in low-latitude regions. This unrealistic latitudinal lightning distribution can also affect the burned area calculations, which is a major problem in your current manuscript.
Please attempt to improve the simulated latitudinal lightning distribution. Although you mentioned that large biases in CAPE, etc., have caused this unrealistic distribution, try to identify the specific causes of the problem and suggest ways to address it in future work. Please provide details about how you applied the standardization to CAPE, etc., including the time resolution or time step used in the lightning calculations.
Citation: https://doi.org/10.5194/gmd-2024-24-CC1
-
CC1: 'Reply on AC1', Yanfeng He, 23 Apr 2024
reply
-
RC2: 'Comment on gmd-2024-24', Anonymous Referee #2, 23 Apr 2024
reply
The manuscript presents some results of the CanAM5.1 model with a new lightning scheme. In general, the manuscript is interesting, but I think it does not fit the scope of GMD. Judging from the title, a new lightning scheme is the main subject of the manuscript, but it is taken from previously published paper and described in just less than 80 lines. The rest of the manuscript is devoted to the analysis and evaluation of the CanAM5.1 performance in the modeling of the cloud related quantities, which fully define rather low (see figure 1, 2, 4, 5) accuracy of the simulated lightning frequency. Analysis of the burning areas and future projections looks interesting but irrelevant to the main aim of the manuscript. I understand that the accuracy of the proposed parameterization is comparable to other available parameterizations, but it is not shown in the manuscript. I think that the manuscript should be transferred to the more relevant journal (I think ACP is a good alternative).
Citation: https://doi.org/10.5194/gmd-2024-24-RC2
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
320 | 71 | 18 | 409 | 28 | 11 | 17 |
- HTML: 320
- PDF: 71
- XML: 18
- Total: 409
- Supplement: 28
- BibTeX: 11
- EndNote: 17
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1