| 10 Feb 2026
EMMA-Tracker v1.0: A lifecycle-based algorithm for identifying and tracking mesoscale convective systems in observations and climate models
Abstract. Understanding the long-term climatology and physical drivers of Mesoscale Convective Systems (MCSs) in Europe is hindered by the lack of multi-decadal datasets and the difficulties distinguishing precipitation resulting from MCSs from synoptic scale frontal precipitation. Reference datasets for model evaluation that mix these physically distinct phenomena can cause misinterpreting climate model skill in representation of mesoscale convective processes. To address this, we introduce the EMMA-Tracker (Evolution-based MCS Model Assessment), a novel algorithm designed to identify and track MCSs using only standard model output variables. This intentional design choice enables a physically consistent comparison between observations and climate model ensembles, providing a pathway to investigate how MCS characteristics may evolve in a warming climate through the analysis of future projections. We apply this tracker to IMERG precipitation and ERA5-derived atmospheric instability to generate a 27-year (1998–2024) warm-season climatology— the longest reference dataset of European MCSs to date. The algorithm’s core innovation is a series of physics-based post-processing filters that utilize the system’s full spatiotemporal lifecycle to isolate coherently propagating MCSs from stationary thunderstorms and frontal rainbands. Our results show that these coherent MCSs are the dominant driver of extreme hourly precipitation. Their contribution systematically increases with hourly precipitation intensity, exceeding 60 % of heavy precipitation (P99.9) across most of continental Europe and 80 % over parts of the Mediterranean. The EMMA-Tracker provides both an observational reference for climatological studies and for targeted, process-oriented evaluation of regional and convection-permitting climate models.
This paper describes a new MCS tracking algorithm, the Evaluation-Based MCS Model Assessment (EMMA-Tracker v1.0) specifically designed to track MCSs using variables that are standard to model output, precipitation and a lifted index. The detection algorithm is relatively standard to existing tracking methods, but in the post-processing filter step, the authors use criteria on track straightness, environmental instability, and area volatility to constrain the dataset, and importantly, filter out any frontally-driven MCSs. The authors run this algorithm using IMERG v7 precipitation and a lifted index calculated from ERA5 data to generate a 27 year long warm-season MCS climatology record over Europe with the goal of creating a dataset useful for climatology studies and the process-based evaluation of models.
Through its proposal of a new MCS tracker and new MCS definition specific to a region and season and designed to benefit model evaluation, this work to some extent falls outside of the current paradigm of MCS trackers, which are typically designed to be globally generalizable, and has the potential to be an important contribution to the field. However, in the comments below I describe areas in which the argument of the authors should be strengthened, specifically highlighting where additional context and justification for their tracking algorithm choices are needed, and a few areas where their analysis does not appear to be aligned with the results in their figures.
Specific Comments:
Line 11: I understand what coherent means in the abstract after reading the full manuscript, but it was unclear to me on first read. I might suggest including specifics on what the authors use to assess MCS ‘coherence’ here.
Line 23: I think that the authors’ statement about MCS simulation being an important benchmark for climate model skill is not well-justified here. Please include additional literature (i.e., Chen et al. 2020) that better explains why this is the case.
Lines 49-60: The association of some extreme precipitation events with MCSs and fronts is not inherently contradictory. While I appreciate the framework the authors are arguing here, I would urge them to also incorporate the context of recent literature on co-occurring precipitation phenomena, such as Tsai et al. 2025. Their MCS definition also does not include the description in Schumacher and Rasmussen (2020) that describes externally-driven MCSs that are sustained by lifting associated with large-scale forcing of ascent, such as that along a frontal boundary. I agree with the overarching point argued that an evaluation dataset making a clear distinction between these two systems could be useful in evaluating certain aspects of model performance, but given that the central argument of the manuscript rests on a delineation between two phenomena that traditionally can be co-located, I would recommend the overall argument of this section be strengthened and would be aided by additional literature.
Line 73-74: It is true that infrared brightness temperature is not a standard model output, but outgoing longwave radiation, which can easily be converted, is, and as the authors note, most previous MCS trackers use brightness temperature only (i.e., Feng et al. 2025). Additionally, the co-location of precipitation in brightness temperature in MCS tracking has been shown to eliminate the cited issues with the incorporation of non-precipitating cirrus (i.e. Feng et al. 2021). I am a little bit confused about the choice to use precipitation only here. Could further justification be provided?
Line 110-111: This argument could be strengthened by referencing the high-resolution radar dataset used in the supplemental figure evaluation directly in the text.
Line 143 - 144: Interpolating temperature and specific humidity variables from 25 km to a higher resolution has the potential to introduce biases in the final data product for the lifted index, especially given that these fields are not always smoothly varying. Could the authors please explain/justify this choice further?
Line 148-149: Are the authors concerned that this smoothing would reduce the reliability of tracking on fine-scale features? Could they please explain this choice further, especially given that small-scale grid noise is not widely considered a problem in IMERG?
Line 155: I understand the rationale for using a threshold-based algorithm here to build thresholds that work for datasets of different resolutions, but I am not sure why having a threshold that adapts to a systematic bias of a model dataset is a good thing if the goal is to use this dataset for a process-oriented evaluation of models. Could this choice please be explained/justified further?
Line 225: Figure A1b could also be interpreted as, rather than the centroid shifting erratically the whole time, two distinct systems, and an erratic jump in detection between the two. What is the rationale for implementing this track straightness criteria rather than just implementing a criteria that would identify these two different components of the frontal system as two unique areas of convection?
Line 250: Similar to above, Figure A1d appears to show an MCS track that looks relatively reasonable if it were stopped a few time steps earlier, and I have seen valid MCSs masks that are more elongated in the horizontal. Could the authors please explain/justify further why the emphasis here is on eliminating this track in its entirety?
Line 259: I do not see a prominent maximum over the North Atlantic, Great Britain, and the North Sea in Figure 3a. Could the authors please clarify?
Line 278-280: A visual comparison with Morel and Sensei (2002) makes me wonder about the missing peak over the southern Mediterranean that is seen in that climatology; given that the tracking domain includes 30 degrees and northward. Could the authors please explain this discrepancy?
Figure 15: Why is masking only conducted in panel (b) and not (d)?
Line 302: I don’t necessarily see a secondary maximum of rejected frontal systems over the Balkans and Croatia in Figure 4c (where is the primary maximum?); rather, I see one cohesive maximum on the eastern portion of the Alps. Could this description please be clarified?
Line 322: I don’t think that the general pattern over eastern France in Figure 5a can be considered a maxima, given that there is a broad region showing an average of one MCS per year. Could this characterization be please explained/ clarified?
Line 401 - 402: I understand that MCSs are much less common over Europe in winter months, but if this tracking algorithm was specifically designed to filter out synoptically driven precipitation, why is it necessarily more unreliable under these conditions? Could further justification please be provided?
Line 426-428: Stronger justification is needed for why this dataset is useful for process-based evolution of models like EURO-CORDEX given that it has been stated above that it does not have the appropriate hourly resolution for direct comparison.
Technical Comments:
Line 1 and 17: Mesoscale convective system (MCS) (and future text that is defined as an acronym, such as wet hour frequency (WHF)) should always be lowercase.
Line 13: This statement is unclear. I suggest replacing ‘exceeding’ with ‘accounting for’ or another phrase to make it clear what 60% refers to here.
Line 63: It is not clear what a ‘mixed dataset’ refers to here.
Line 82: It is not clear what is meant by ‘genuine’ MCSs here. Internally driven / non-frontal associated MCSs?
Line 120: Starting here and for the remainder of this section, the authors refer to IMERG where I believe they are intending to refer to ERA5.
Line 355: The manuscript transitions from discussing MCS characteristics back to geospatial climatology here with no explanation, and left me as the reader confused.
Line 410 - 414: These sentences have multiple grammatical and punctuation errors that could be corrected for readability.
Line 419: I think this should read ‘IMERG precipitation and ERA5 data.’
Line 435: I believe the sentence should end with ‘...for evaluating organized convection.’
References:
Chen, X., O. M. Pauluis, L. R. Leung, and F. Zhang, 2020: Significant Contribution of Mesoscale Overturning to Tropical Mass and Energy Transport Revealed by the ERA5 Reanalysis. Geophysical Research Letters, 47, e2019GL085333, https://doi.org/10.1029/2019GL085333.
Feng, Z., and Coauthors, 2021: A Global High‐Resolution Mesoscale Convective System Database Using Satellite‐Derived Cloud Tops, Surface Precipitation, and Tracking. JGR Atmospheres, 126, e2020JD034202, https://doi.org/10.1029/2020JD034202.
Song, F., Z. Feng, L. R. Leung, R. A. Houze Jr., J. Wang, J. Hardin, and C. R. Homeyer, 2019: Contrasting Spring and Summer Large-Scale Environments Associated with Mesoscale Convective Systems over the U.S. Great Plains. Journal of Climate, 32, 6749–6767, https://doi.org/10.1175/JCLI-D-18-0839.1.
Tsai, M., and Coauthors, 2025: Co‐Occurring Atmospheric Features and Their Contributions to Precipitation Extremes. Journal of Geophysical Research, 130, https://doi.org/10.1029/2024JD041687.