the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Objective identification of pressure wave events from networks of 1-Hz, high-precision sensors
Sandra E. Yuter
Matthew Allen Miller
Laura M. Tomkins
Abstract. Mesoscale pressure waves including atmospheric gravity waves, outflow and frontal passages, and wake lows are outputs of and can potentially modify clouds and precipitation. A wavelet-based method for identifying and tracking these types of wave signals in time series data from networks of low-cost, high-precision (0.8-Pa noise floor, 1-Hz recording frequency) pressure sensors is demonstrated. Strong wavelet signals are identified using a wave period-dependent (i.e., frequency-dependent) threshold, then those signals are extracted by inverting the wavelet transform. Wave periods between 1 minute and 120 minutes were analyzed, a range which would include several types of mesoscale disturbances in the troposphere. After extracting the signals from a network of pressure sensors, the cross-correlation function is used to estimate the time difference between the wave passage at each pressure sensor. From those time differences, the wave phase velocity vector is calculated using a least-squares fit. If the fitting error is sufficiently small (thresholds of RMSE < 90 s and NRMSE < 0.1 were used), then a wave event is considered robust and trackable.
- Preprint
(25876 KB) - Metadata XML
- BibTeX
- EndNote
Luke Robert Allen et al.
Status: open (until 12 Oct 2023)
-
RC1: 'Referee comments', Anonymous Referee #3, 15 Sep 2023
reply
Allen et al. present a network of COTS pressure sensors time synced to a central server. They show how these networks can record wavelike oscillations in the gravity wave period range, and track them across individual elements. The tracking system is developed using wavelet and cross correlation methods. Several examples are shown, including the Hunga-Tonga event, gravity wave trains, and wake lows. The authors note that this method could increase the dataset of trackable gravity waves, with implications for winter storms. They note they have a forthcoming publication that will describe this in more detail.
Allen et al. present a fairly convincing case that these COTS sensors can record coherent wavelike phenomena from a variety of sources, including distant volcanic eruptions and gravity waves from diverse meteorological phenomena. They outline an event detection and association method, although it is not clear if this is happening in near real time or is via postprocessing. The manuscript is clearly in AMT's scope. The paper would benefit from some more background on the relevance of the observations conducted, how the sensor network could be replicated or expanded, previous instances of COTS pressure/seismic sensors, and clarity on the detection thresholds used. I recommend it for publication after the authors take the following comments into account.
MAJOR COMMENTS
1.  It is unclear to me how this measurement technique will be used to either augment operational forecast models or conduct atmospheric studies. Even a methods-based paper needs to provide this information in order to motivate the reader to assess the technique in question. In the summary, the authors note that a forthcoming publication will detail the application of this measurement technique, but that is mentioned by afterthought. The introduction gives a good overview of gravity waves and what phenomena they impact, but I would appreciate more discussion of how this ties in with the network concept they have created. I encourage the authors to explicitly describe what measurement gaps they are trying to fill, and the impact such filling will have. I suggest inserting this discussion between the second to last and last paragraph in the introduction.
2.  The low cost, COTS nature of this network is intriguing.  The paper would benefit from mentioning similar low cost network concepts, for example the Raspberry Shake seismometer (see Anthony et al., DOI 10.1785/0220180251 and Lamb et al., DOI 10.3389/fcosc.2020.630967) and the RedVox infrasound cellphone app (Eaton et al, DOI 10.1016/j.apacoust.2022.109015, also see Garces et al, DOI: 10.3390/ signals3020014 and especially the reference list). How does the presented architecture fit in with similar systems? Could it be expanded to a broader network of pressure sensors, analogous to the Raspberry Shake seismic network>
3.  Maybe not a "major comment" but one more reflecting my background in geophysical acoustics: The authors focus on gravity waves, but lower limit of the wave periods specified (1-3 minutes) include acoustic and acosutic gravity modes. The Hunga Tonga event is also mentioned in the examples. I suggest mentioning acoustic and acoustic/gravity waves in the abstract and the introduction because of this. Also, there are instances of very long period acoustic waves reported in the literature that may be meteorologically relevant. These include those from severe covective storms, especially those that generate hail (see Goerke and Woodward 1966 "Infrasonic observation of a severe weather system" and Elbing et al. (2019) DOI: 10.1121/1.5124486) Wind over mountains also generates powerful low frequency infrasound and may correlate with turbulence experienced by airplanes, see Bedard (1978) "Infrasound originating near mountainous regions in Colorado".
4.  There are statements made about the detection threshold of the sensors (e. g. lines 65-70), and what sort of periods and propagation speeds could be resolved. The statements made here would be easy to assess by generating synthetic data with certain background noise levels and features with various periods traveling at various velocities.  These could be fed through the detection and association algorithm to determine which were captured and which were not. I recommend that this be done to more rigorously assess the detection ability of the networks presented here.
MINOR COMMENTS
Line 5: Â Wave periods up to ~300 seconds are acoustic, or acoustic-gravity, modes. Â
Line 25:Â
The signal generated by Hunga Tonga is exceptionally rare and has only been generated twice in my knowledge - Hunga Tonga and the Krakatoa eruption of 1883. It may be worth mentioning that large bolide impacts or thermonuclear explosions can generate similar phenomena as well. See Pierce and Posey (1971) "Theory of the excitation and propagation of Lamb's atmospheric edge mode from nuclear explosions". The authors may have also meant gravity waves generated by large volcanic plumes, which are somewhat more common (though still quite rare, especially in eastern North America!).Lines 65-70 Is this detection threshold based on any simulations or analysis, or is it just a guess?
Line 120: Again, a study using synthetic data would be useful to better assess what K values to use.
Lines 145-150:  This 2 hour threshold seems arbitrary.  Doesn't it depend on the separation distance between the pair of furthest-apart sensors in a given array, divided by the typical propagation speeds of the wave features of interest?
Line 201 - What kind of 'testing'?
Line 215: The phase speed is not precisely the speed of sound - better to say "the phase speed is known".  The waves from Hunga Tonga were not really sound waves in the normal sense of the word.
Line 223 - The ash plume technically reached the mesosphere
Line 225 - It went a lot higher than 30 km (55 km, see Kloss et al., DOI: 10.1029/2022GL099394)
Line 233 - The signals were not 'shock waves' when they reached the sensors. The term 'shock wave' denotes a specific nonlinear process that was not at work here.
Line 244 - Lamb waves, not Lamb shock waves
Figures 1 and 4 are never referred to in the text.
Citation: https://doi.org/10.5194/egusphere-2023-1600-RC1 -
RC2: 'Comment on egusphere-2023-1600', Anonymous Referee #1, 18 Sep 2023
reply
GENERAL COMMENTS
The manuscript "Objective identification of pressure wave events from networks of 1-Hz, high-precision sensors" by L.R. Allen et al. describes the study of pressure wave events (gravity wave like) with small arrays of low-cost pressure sensors. The instrumentation, wavelet and array processing methodology, data processing as well as a number of five exemplary event cases are provided and serve as a kind of proof of concept for the given approach. The study focuses on a link of the gravity wave signatures to meteorological phenomena like clouds and precipitation and an outlook is given on a further systematic climatological investigation in that context.
The study shows good quality in the scientific approach and presentation of methods and results. It provides a clear step-by-step description of the measurement and processing techniques, making it suitable for the AMT journal. Anyhow, neither the pressure measurements of acoustic, infrasonic, acoustic-gravity or buoyancy waves, nor the array processing of distributed, low-cost sensors, the cross correlation as well as travel time difference calculation for velocity estimation is a novel, innovative or substantially new approach. Therefore the general significance of the concept and results to identify high-amplitude events is comparably low. Acceptance of the study for publication therefore depends on certain clarifications and improvements of the manuscript, as specifically commented below, provided by RC1 and probably supported by additional reviewer comments.Â
Â
SPECIFIC COMMENTS
- Introduction and References Sections: the manuscript would benefit from some improved references to strengthen the respective contexts of gravity waves, volcanoes, meteorology and instrumentation. Therefore, in lines 12-13 a comprehensive peer-reviewed review paper on gravity waves and their effects should be added, I recommend Fritts and Alexander 2003 (https://doi.org/10.1029/2001RG000106) for that. The different phenomena in lines 24/25 that are also the basis for the examples later on, should be referenced from literature. I recommend papers connecting gravity wave (like) observations of volcanic and meteorological examples like Vergoz et al, 2022 (https://doi.org/10.1016/j.epsl.2022.117639) and Coleman and Knupp, 2009 (https://doi.org/10.1175/2009WAF2222248.1) or similar. Further phenomena as mentioned by RC1 are also worthwhile. I also second the RC1 opinion to mention other distributed instrumentation approaches like Raspberry or RedVox. Furthermore the infrasound measurements of the USarray especially of gravity waves should be mentioned, see C. de Groot Hedlin et al, 2014 (https://doi.org/10.1016/j.epsl.2013.06.042), around lines 32 to 39.
- Data and Methods Sections: Data and methodology are clearly described with only some small questions open: e.g. why is an 2 hour time window chosen in line 147 and what is the specific, measurable impact of using four sensors in line 203 and not three, or let's say five. More generally it would be good, as also stated by RC1 to provide a sensitivity analysis on the impact of the described threshold, signal-to-noise ratio, wavelet and error limits chosen, and so on.
- Examples Section: to be even more specific, the question should be discussed if the given examples are only the raisins picked or if they represent typical events. This might shed light on the accuracy of detecting notable gravity wave events, on the true/false positive/negative rate of detections of events, and on the reliability of the approach in general. A receiver operation characteristic (ROC) analysis is a possible approach to do so. This might be a topic for open discussion during the review process, and not an obligation to be done, since the manuscripts summary already briefly discusses that there are several prerequisites and issues that are necessary and might allow or prohibit a gravity wave detection at the given surface instrumentation. That discussion should be enhanced taking into account influence factors for detection or non-detection of events and their significance. I would generally expect a more elaborated discussion section within the manuscript, taking into account the given and upcoming open discussion remarks. Â Â Â
Â
TECHNICAL COMMENTS:
the language as well as representation of formulae and figures is fine for me, I'm also in favor of the available data (and statement) and additional visual material. I have only minor technical remarks:
- line 215: doubling of "where"
- line 224/225: the plume reached around 57 km and thus the mesosphere. The above given reference Vergoz et al could also be used around line 227 and following to reference observations of the lamb wave. According to Vergoz et al, it was detected during even five circles around the globe.Â
- line 242: "phase speed of 261.9", according to table 1 this is the azimuth and the phase speed is 292.1
- line 267: name the number of detecting sensors for this case (five?)
- Caption of figure 6: "Hunga Ha'apai" instead of " Hunga Ta'apai"
Â
Â
Â
Citation: https://doi.org/10.5194/egusphere-2023-1600-RC2 -
RC3: 'Comment on egusphere-2023-1600', Anonymous Referee #4, 20 Sep 2023
reply
This manuscript presents a method to extract signals from small networks of low-cost, high-precision pressure sensors. This topic fits the scope of the journal in terms of data processing techniques that can be applied to atmospheric measurements. The manuscript has fairly clear motivations, is logically organized, written well, and includes appropriate figures and supplementary information. The methods are clearly stated and enough detail is provided so that these may be replicated in other studies. The examples are concise descriptions of an assortment of phenomena that drive the observed pressure signals, and sufficiently highlight the method’s ability to detect the associated signal for these different cases.
There is only one additional minor suggestion. The first sentence of the abstract could be split into two with one stating the causes of pressure waves, and the second on why they are important, such as their impact on clouds and precipitation. Similarly in the introduction, it might make more sense to state what they are, how they are generated, and how to distinguish them before discussing their potential impact on cloud and precipitation processes.
Citation: https://doi.org/10.5194/egusphere-2023-1600-RC3 -
RC4: 'Comment on egusphere-2023-1600', Anonymous Referee #2, 24 Sep 2023
reply
This manuscript presents a method for extracting signals from a small network of low-cost, high-precision pressure sensors. The topic of the manuscript is clear and logically organized. The manuscript makes a compelling case that these COTS sensors can record coherent wave-like phenomena from a variety of sources, including distant volcanic eruptions and gravity waves from a variety of meteorological phenomena.
It is recommended that gravity waves and their effects on clouds and precipitation be described in more detail again in the first part, i.e., explaining what gravity waves are, how they are generated, and what kinds there are.
The manuscript may need to describe how this measurement technique will be used to enhance operational forecasting models or for atmospheric research.
Citation: https://doi.org/10.5194/egusphere-2023-1600-RC4
Luke Robert Allen et al.
Data sets
Data for Objective identification of pressure wave events from networks of 1-Hz, high-precision sensors Matthew A. Miller and Luke R. Allen https://doi.org/10.5281/zenodo.8136536
Video supplement
Supplemental videos of the paper "Objective identification of pressure wave events from networks of 1-Hz, high-precision sensors" Luke R. Allen, Laura M. Tomkins, Sandra E. Yuter https://doi.org/10.5446/s_1476
Luke Robert Allen et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 126 | 30 | 14 | 170 | 4 | 4 |
- HTML: 126
- PDF: 30
- XML: 14
- Total: 170
- BibTeX: 4
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1