Monitoring snowpack SWE and temperature using RFID tags as wireless sensors

Le Breton, Mathieu; Larose, Éric; Baillet, Laurent; Lejeune, Yves; van Herwijnen, Alec

doi:https://doi.org/10.5194/egusphere-2022-761

Preprints

https://doi.org/10.5194/egusphere-2022-761

Preprints

22 Aug 2022

| 22 Aug 2022

Monitoring snowpack SWE and temperature using RFID tags as wireless sensors

Mathieu Le Breton, Éric Larose, Laurent Baillet, Yves Lejeune, and Alec van Herwijnen

Abstract. This work shows that passive radio-frequency identification (RFID) tags can be used as low-cost contactless sensors, to measure the variations in snow water equivalent (SWE) of a snowpack. RFID tags are produced massively to remotely identify industrial goods, hence are available commercially off-the-shelf at very low-cost. The introduced measurement system consists of a vertical profile of RFID tags installed before the first snowfall, interrogated continuously by a 865–868 MHz reader that remains above the snowpack. The system deduces the SWE variations from the increase of phase delay induced by the new layers of fresh snow which slows the propagation of the waves. The method is tested both in a controlled laboratory environment, and outdoors on the French national reference center of Col de Porte, to cross-check the results against a solid reference dataset (cosmic rays, precipitation weighting, temperature monitoring, and snow pit surveys). The technical challenges solved concern multipathing interferences, snowmelt acceleration during reheats, measurement discontinuity, and wet snow influence. This non-contact and non-destructive RFID technique can estimate the SWE of dry snow, with the accuracy of ±3−30 kg/m² depending on the number of tags and antennas. In addition, the system can monitor the snow temperature with 1 °C accuracy and spatialization, using dedicated sensors embedded in the tags.

Received: 06 Aug 2022 – Discussion started: 22 Aug 2022

Download & links

Preprint (PDF, 1750 KB)

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 (1750 KB)

Download & links

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Journal article(s) based on this preprint

04 Aug 2023

Monitoring snow water equivalent using the phase of RFID signals

Mathieu Le Breton, Éric Larose, Laurent Baillet, Yves Lejeune, and Alec van Herwijnen

The Cryosphere, 17, 3137–3156, https://doi.org/10.5194/tc-17-3137-2023,https://doi.org/10.5194/tc-17-3137-2023, 2023

Short summary

Mathieu Le Breton et al.

Interactive discussion

Status: closed

RC1:
'Comment on egusphere-2022-761', Christian Mätzler, 08 Sep 2022

See document attached

Citation: https://doi.org/10.5194/egusphere-2022-761-RC1
- AC1:
  'Reply on RC1', Mathieu Le Breton, 19 Oct 2022
  
  Dear Christian Mätzler,
  First, thank you very much for your in-depth comments. We have adressed or are adressing them. I would like to try the interactive possibilities of this open review to adress one question/clarification. It concerns these comments :
  5) No information is given on the scattering and absorption cross sections of the tags used, nor of the supporting structures.
  
  6) No information is given on the method used to discriminate the responses and the backscattered signals from different tags, and how this discrimination may be linked with the phase determination.
  Mainly, I do not see the motivation behind your suggestion to characterize the Radar Cross Section of the tags and supporting structure (comment 5). I think comes from a misunderstanding of the difference between a radar target and an RFID tag (comment 6), that we have clarified in the text. The answer to comment 6 is straightforward: one tag at a time is requested by the interrogator (using standard the communication & anti-collision protocol EPC Gen2) to modulate the signal it backscatters, while all the other tags and the environment backscatter a non-modulated wave (as it would do with a standard radar). The phase difference of arrival is measured between the two modulation states of the received signal, therefore linked only to the singulated tag.
  Indeed in RFID, an important parameter is the DeltaRCS of the tags, which is the difference of the tag's radar cross-section between these two modulation state. We have characterized the DeltaRCS for the model of tag used (depending on frequency and input power). I can add this characterization in the 'instrumentation' section if you find it useful.
  In the end,knowing how the response of each tag is discriminated, we do not see the reason for characterizing the cross-section of the tags and supporting structure (which would make sense however with a radar). Can you confirm me, or otherwise explain the motivation behind the comment 5 ?
  
  Sincerely yours,
  
  Mathieu Le Breton, on behalf of all the coauthors.
  
  Citation: https://doi.org/10.5194/egusphere-2022-761-AC1
  - RC2: 'Reply on AC1', Christian Mätzler, 19 Oct 2022
    
    Dear Mathieu Le Breton,
    Thank you for the comment and explanation with respect to my Comments 5 and 6. It is certainly useful for Comment 6. But my concern was not related to the radar cross section, but to the scattering properties of the inserted materials that cause multipath effects, and thus speckel noise that you see in your data. What is your explanation for the large data spread?
    Best regards, Christian Mätzler
    
    Citation: https://doi.org/10.5194/egusphere-2022-761-RC2
    
    AC4: 'Reply on RC2', Mathieu Le Breton, 09 Jan 2023
    
    Dear Christian Matzer,
    First, thank you for your comments. We have just received the second referee's comment so I am answering both. We have processed your (see the answer in the attached document), they will help to improve the manuscript.
    Regarding the multipathing, however, we propose an alternative approach.
    We agree that there is a need to better understand the multipathing effect on RFID systems in snow contexts. However, this is far beyond the scope of the present study: multipathing is a general issue in RFID (very present for localization indoors for example). All the reflective elements should be a source of multipathing: The environment (snow surface, layers of snow, soil), and the installation itself (tags, light plastic supporting structure that holds the tag, large metallic structure that holds the reader). Any modification of geometry (snow depth) or dielectric constant (i.e., snow density, moisture content of the snow and soil) should modify the influence of multipathing on phase measurements. In practice, multipathing in RFID is more often mitigated than predicted.
    
    It is mitigated in the study by using multiple antennas and tags (=spatial diversity). To further mitigate it in a future installation, we suggest (as you did) to use an array of tags placed very close to the ground (also necessary to reduce the thermal influence of the system). It should help average the spread due to spatial diversity, and reduce the reflection on the tags and on the supporting material.
    
    We have a work in progress that explores more in depth the role of multipathing when using an RFID system is snow context, for which we will dedicate a communication in itself. The document attached presents (still unpublished) results of two experiments and one model: the amplitude of the phase variation induce by multipathing (from 1.4 to 3 rad) is coherent with the spread in our study (1.2 rad).
    
    I also suggest to improve the link of our study, to existing studies related to multipathing in the snow or with RFID systems:
    —The model used and preliminary results are also in:
    Le Breton, M., 2019. Suivi temporel d’un glissement de terrain à l’aide d’étiquettes RFID passives, couplé à l’observation de pluviométrie et de bruit sismique ambiant (PhD Thesis). Université Grenoble Alpes, ISTerre, Grenoble, France. https://www.theses.fr/2019GREAU013 (page 144 to 156).
    —Some methods exploit multipathing for snow, such as
    Espín-López, P. F., Pasian, M., 2021. Determination of Snow Water Equivalent for Dry Snowpacks Using the Multipath Propagation of Ground-Based Radars. IEEE Geoscience and Remote Sensing Letters 18, 276–280. https://doi.org/10.1109/LGRS.2020.2974546;
    Kulsoom, F., Dell’Acqua, F., Pasian, M., Espín-López, P. F., 2021. Snow Layer Detection by Pattern Matching in a Multipath Radar Interference Scenario. International Journal of Remote Sensing 42, 3193–3218. https://doi.org/10.1080/01431161.2020.1854890).
    —Mitigating multipathing in RFID is a topic in itself. Some studies are dedicated to propose mitigation methods, such as
    DiGiampaolo, E., Martinelli, F., 2020. A Multiple Baseline Approach to Face Multipath. IEEE Journal of Radio Frequency Identification 4, 314–321. https://doi.org/10.1109/JRFID.2020.3022576
    
    In short, we encountered the multipathing issue and demonstrated a simple way to mitigate it. However, describing in detail the multipathing issue and how to further mitigate it, is a topic in itself which, in our opinion, should not be in the present paper. Instead we suggest to (a) propose a future installation that further reduce the multipathing issue, (b) to detail more about the causes of multipathing in the text, and (c) to link our observations with other studies that deal specifically about multipathing with snow or rfid.
    Is this approach suitable to you ?
    Best regards,
    
    Citation: https://doi.org/10.5194/egusphere-2022-761-AC4
RC3:
'Comment on egusphere-2022-761', Anonymous Referee #2, 07 Jan 2023

This work proposes the use of RFID for monitoring the snowpack SWE and temperature. The topic and the application is interesting and the authors provided experimental results which seem to provide good results. However there are some criticism that should be clarified.

1) the authors claim to use standard RFID, that this is not really true a standard RFID do not be equipped with sensors, you must design a customised tag. The same for the battery you can't connect a battery to a standard tag. You should provide more information related to the Considered RFID schema.

2) you perform a measure of phase difference but it is not clear how. If you used inductive coupling rfid it is quite difficult. I suppose that you used a RF tag, and you claim that the reader is able to detect the phase difference. Can you please better explain how? You have to analyse the signals at the demodulator in order to detect the phase difference of did you use another technique? Please explain.

3) did you take into account the attenuation introduced in the RF signal by the snow?

4) I suppose that you modified the reader, don't you? If yes please report the introduced customisations.

in this form the manuscript is not acceptable for publicatIon in this form. I suggest major revisions.

Citation: https://doi.org/10.5194/egusphere-2022-761-RC3
- AC2: 'Reply on RC3', Mathieu Le Breton, 08 Jan 2023
  
  Dear referee, first, thank you very much for your review, we appreciate your work.
  
  Please find below some answers, and a question (your original text is in bold)
  
  In short, your comments showed that we need to clarify our text, to avoid any misunderstanding (in particular, the fact that we used only commercial off-the-shelf readers and tags, without customisation), which we will do.
  1) the authors claim to use standard RFID, that this is not really true a standard RFID do not be equipped with sensors, you must design a customised tag. The same for the battery you can't connect a battery to a standard tag. You should provide more information related to the Considered RFID schema.
  We will clarify this point. In particular, what is 'standard' is the communication protocol (EPC-Gen2) and the RF channel (ETSI-302-201). Then we used 'commercial off-the-shelf' devices, both tags and reader. We have modified the tag (UHF tag, model Survivor B, from Confidex, that embeds an EM4325 chip from EM electronic, an a small battery) ony superficially: first, we configured its memory bank wirelessely, in order to make temperature measurements, and to increase its power sensitivity. Second, we painted the tag in white to reduce the radiative heat transfer.
  So we bought an industrial tag, but we have not designed a customised tag.
  2) you perform a measure of phase difference but it is not clear how. If you used inductive coupling rfid it is quite difficult. I suppose that you used a RF tag, and you claim that the reader is able to detect the phase difference. Can you please better explain how? You have to analyse the signals at the demodulator in order to detect the phase difference of did you use another technique? Please explain.
  We used an UHF RFID (working around 868 MHz). The reader (Impinj R700) has the capacity to read phase difference of arrival when reading a tag, out-of-the-box. Several other readers have this capacity, such as the ImpinjSR420 or the ThingMagic M6. Measurements of phase difference of arrival is largely used for RFID localization, see for example the introduction of this review :
  
  Le Breton M, Liébault F, Baillet L, Charléty A, Larose É, Tedjini S. Dense and long-term monitoring of earth surface processes with passive RFID — a review. Earth-Science Reviews. 2022 Nov 1;234:104225.
  
  We do not know, nor need to know, how the Impinj R700's electronic is designed in order to read the phase. For clarification, we will add a reference to research papers describing techniques to read the phase from a reader.
  
  3) did you take into account the attenuation introduced in the RF signal by the snow?
  Could you please clarify ? For which aim should we take the attenuation into account ?
  There are several sources of signal loss, such as, indeed, attenuation in the snow, but also detuning of the tag, reflexion at the surface, or destructive multipathing interferences. We have measured the signal strength data for each tag, but the data is messy: using signal strength is often a difficult method to exploit for RFID sensing outdoors because of its many influence factors (see again the review Le Breton et al. 2022). Indirectly, the signal strenght received by the tag can slightly influence the phase (and therefore our SWE measurement), when the tag enters non-linear behavior (for example when it receives very powerful signal). The non-linear behavior depends strongly on model of chip. With our tag, non-linearity occurs only at high power that are not reached in the study. And in any case, that would be a small source of noise.
  In short: The present study exploits only the phase difference of arrival (related to wave slowness), not the signal strength (related to attenuation).
  4) I suppose that you modified the reader, don't you? If yes please report the introduced customisations.
  No, we have used a commercial off-the-shelf reader, not customised. The method works with any reader that is able to read the phase difference of arrival.
  
  Citation: https://doi.org/10.5194/egusphere-2022-761-AC2
AC3: 'Comment on egusphere-2022-761', Mathieu Le Breton, 09 Jan 2023

Dear Christian Matzer,
First, thank you for your comments. We have just received the second referee's comment so I am answering both. We have processed your (see the answer in the attached document), they will help to improve the manuscript.
Regarding the multipathing, however, we propose an alternative approach.
We agree that there is a need to better understand the multipathing effect on RFID systems in snow contexts. However, this is far beyond the scope of the present study: multipathing is a general issue in RFID (very present for localization indoors for example). All the reflective elements should be a source of multipathing: The environment (snow surface, layers of snow, soil), and the installation itself (tags, light plastic supporting structure that holds the tag, large metallic structure that holds the reader). Any modification of geometry (snow depth) or dielectric constant (i.e., snow density, moisture content of the snow and soil) should modify the influence of multipathing on phase measurements. In practice, multipathing in RFID is more often mitigated than predicted.

It is mitigated in the study by using multiple antennas and tags (=spatial diversity). To further mitigate it in a future installation, we suggest (as you did) to use an array of tags placed very close to the ground (also necessary to reduce the thermal influence of the system). It should help average the spread due to spatial diversity, and reduce the reflection on the tags and on the supporting material.

We have a work in progress that explores more in depth the role of multipathing when using an RFID system is snow context, for which we will dedicate a communication in itself. The document attached presents (still unpublished) results of two experiments and one model: the amplitude of the phase variation induce by multipathing (from 1.4 to 3 rad) is coherent with the spread in our study (1.2 rad).

I also suggest to improve the link of our study, to existing studies related to multipathing in the snow or with RFID systems:
—The model used and preliminary results are also in:
Le Breton, M., 2019. Suivi temporel d’un glissement de terrain à l’aide d’étiquettes RFID passives, couplé à l’observation de pluviométrie et de bruit sismique ambiant (PhD Thesis). Université Grenoble Alpes, ISTerre, Grenoble, France. https://www.theses.fr/2019GREAU013 (page 144 to 156).
—Some methods exploit multipathing for snow, such as
Espín-López, P. F., Pasian, M., 2021. Determination of Snow Water Equivalent for Dry Snowpacks Using the Multipath Propagation of Ground-Based Radars. IEEE Geoscience and Remote Sensing Letters 18, 276–280. https://doi.org/10.1109/LGRS.2020.2974546;
Kulsoom, F., Dell’Acqua, F., Pasian, M., Espín-López, P. F., 2021. Snow Layer Detection by Pattern Matching in a Multipath Radar Interference Scenario. International Journal of Remote Sensing 42, 3193–3218. https://doi.org/10.1080/01431161.2020.1854890).
—Mitigating multipathing in RFID is a topic in itself. Some studies are dedicated to propose mitigation methods, such as
DiGiampaolo, E., Martinelli, F., 2020. A Multiple Baseline Approach to Face Multipath. IEEE Journal of Radio Frequency Identification 4, 314–321. https://doi.org/10.1109/JRFID.2020.3022576

In short, we encountered the multipathing issue and demonstrated a simple way to mitigate it. However, describing in detail the multipathing issue and how to further mitigate it, is a topic in itself which, in our opinion, should not be in the present paper. Instead we suggest to (a) propose a future installation that further reduce the multipathing issue, (b) to detail more about the causes of multipathing in the text, and (c) to link our observations with other studies that deal specifically about multipathing with snow or rfid.
Is this approach suitable to you ?
Best regards,

Citation: https://doi.org/10.5194/egusphere-2022-761-AC3

Interactive discussion

Status: closed

RC1:
'Comment on egusphere-2022-761', Christian Mätzler, 08 Sep 2022

See document attached

Citation: https://doi.org/10.5194/egusphere-2022-761-RC1
- AC1:
  'Reply on RC1', Mathieu Le Breton, 19 Oct 2022
  
  Dear Christian Mätzler,
  First, thank you very much for your in-depth comments. We have adressed or are adressing them. I would like to try the interactive possibilities of this open review to adress one question/clarification. It concerns these comments :
  5) No information is given on the scattering and absorption cross sections of the tags used, nor of the supporting structures.
  
  6) No information is given on the method used to discriminate the responses and the backscattered signals from different tags, and how this discrimination may be linked with the phase determination.
  Mainly, I do not see the motivation behind your suggestion to characterize the Radar Cross Section of the tags and supporting structure (comment 5). I think comes from a misunderstanding of the difference between a radar target and an RFID tag (comment 6), that we have clarified in the text. The answer to comment 6 is straightforward: one tag at a time is requested by the interrogator (using standard the communication & anti-collision protocol EPC Gen2) to modulate the signal it backscatters, while all the other tags and the environment backscatter a non-modulated wave (as it would do with a standard radar). The phase difference of arrival is measured between the two modulation states of the received signal, therefore linked only to the singulated tag.
  Indeed in RFID, an important parameter is the DeltaRCS of the tags, which is the difference of the tag's radar cross-section between these two modulation state. We have characterized the DeltaRCS for the model of tag used (depending on frequency and input power). I can add this characterization in the 'instrumentation' section if you find it useful.
  In the end,knowing how the response of each tag is discriminated, we do not see the reason for characterizing the cross-section of the tags and supporting structure (which would make sense however with a radar). Can you confirm me, or otherwise explain the motivation behind the comment 5 ?
  
  Sincerely yours,
  
  Mathieu Le Breton, on behalf of all the coauthors.
  
  Citation: https://doi.org/10.5194/egusphere-2022-761-AC1
  - RC2: 'Reply on AC1', Christian Mätzler, 19 Oct 2022
    
    Dear Mathieu Le Breton,
    Thank you for the comment and explanation with respect to my Comments 5 and 6. It is certainly useful for Comment 6. But my concern was not related to the radar cross section, but to the scattering properties of the inserted materials that cause multipath effects, and thus speckel noise that you see in your data. What is your explanation for the large data spread?
    Best regards, Christian Mätzler
    
    Citation: https://doi.org/10.5194/egusphere-2022-761-RC2
    
    AC4: 'Reply on RC2', Mathieu Le Breton, 09 Jan 2023
    
    Dear Christian Matzer,
    First, thank you for your comments. We have just received the second referee's comment so I am answering both. We have processed your (see the answer in the attached document), they will help to improve the manuscript.
    Regarding the multipathing, however, we propose an alternative approach.
    We agree that there is a need to better understand the multipathing effect on RFID systems in snow contexts. However, this is far beyond the scope of the present study: multipathing is a general issue in RFID (very present for localization indoors for example). All the reflective elements should be a source of multipathing: The environment (snow surface, layers of snow, soil), and the installation itself (tags, light plastic supporting structure that holds the tag, large metallic structure that holds the reader). Any modification of geometry (snow depth) or dielectric constant (i.e., snow density, moisture content of the snow and soil) should modify the influence of multipathing on phase measurements. In practice, multipathing in RFID is more often mitigated than predicted.
    
    It is mitigated in the study by using multiple antennas and tags (=spatial diversity). To further mitigate it in a future installation, we suggest (as you did) to use an array of tags placed very close to the ground (also necessary to reduce the thermal influence of the system). It should help average the spread due to spatial diversity, and reduce the reflection on the tags and on the supporting material.
    
    We have a work in progress that explores more in depth the role of multipathing when using an RFID system is snow context, for which we will dedicate a communication in itself. The document attached presents (still unpublished) results of two experiments and one model: the amplitude of the phase variation induce by multipathing (from 1.4 to 3 rad) is coherent with the spread in our study (1.2 rad).
    
    I also suggest to improve the link of our study, to existing studies related to multipathing in the snow or with RFID systems:
    —The model used and preliminary results are also in:
    Le Breton, M., 2019. Suivi temporel d’un glissement de terrain à l’aide d’étiquettes RFID passives, couplé à l’observation de pluviométrie et de bruit sismique ambiant (PhD Thesis). Université Grenoble Alpes, ISTerre, Grenoble, France. https://www.theses.fr/2019GREAU013 (page 144 to 156).
    —Some methods exploit multipathing for snow, such as
    Espín-López, P. F., Pasian, M., 2021. Determination of Snow Water Equivalent for Dry Snowpacks Using the Multipath Propagation of Ground-Based Radars. IEEE Geoscience and Remote Sensing Letters 18, 276–280. https://doi.org/10.1109/LGRS.2020.2974546;
    Kulsoom, F., Dell’Acqua, F., Pasian, M., Espín-López, P. F., 2021. Snow Layer Detection by Pattern Matching in a Multipath Radar Interference Scenario. International Journal of Remote Sensing 42, 3193–3218. https://doi.org/10.1080/01431161.2020.1854890).
    —Mitigating multipathing in RFID is a topic in itself. Some studies are dedicated to propose mitigation methods, such as
    DiGiampaolo, E., Martinelli, F., 2020. A Multiple Baseline Approach to Face Multipath. IEEE Journal of Radio Frequency Identification 4, 314–321. https://doi.org/10.1109/JRFID.2020.3022576
    
    In short, we encountered the multipathing issue and demonstrated a simple way to mitigate it. However, describing in detail the multipathing issue and how to further mitigate it, is a topic in itself which, in our opinion, should not be in the present paper. Instead we suggest to (a) propose a future installation that further reduce the multipathing issue, (b) to detail more about the causes of multipathing in the text, and (c) to link our observations with other studies that deal specifically about multipathing with snow or rfid.
    Is this approach suitable to you ?
    Best regards,
    
    Citation: https://doi.org/10.5194/egusphere-2022-761-AC4
RC3:
'Comment on egusphere-2022-761', Anonymous Referee #2, 07 Jan 2023

This work proposes the use of RFID for monitoring the snowpack SWE and temperature. The topic and the application is interesting and the authors provided experimental results which seem to provide good results. However there are some criticism that should be clarified.

1) the authors claim to use standard RFID, that this is not really true a standard RFID do not be equipped with sensors, you must design a customised tag. The same for the battery you can't connect a battery to a standard tag. You should provide more information related to the Considered RFID schema.

2) you perform a measure of phase difference but it is not clear how. If you used inductive coupling rfid it is quite difficult. I suppose that you used a RF tag, and you claim that the reader is able to detect the phase difference. Can you please better explain how? You have to analyse the signals at the demodulator in order to detect the phase difference of did you use another technique? Please explain.

3) did you take into account the attenuation introduced in the RF signal by the snow?

4) I suppose that you modified the reader, don't you? If yes please report the introduced customisations.

in this form the manuscript is not acceptable for publicatIon in this form. I suggest major revisions.

Citation: https://doi.org/10.5194/egusphere-2022-761-RC3
- AC2: 'Reply on RC3', Mathieu Le Breton, 08 Jan 2023
  
  Dear referee, first, thank you very much for your review, we appreciate your work.
  
  Please find below some answers, and a question (your original text is in bold)
  
  In short, your comments showed that we need to clarify our text, to avoid any misunderstanding (in particular, the fact that we used only commercial off-the-shelf readers and tags, without customisation), which we will do.
  1) the authors claim to use standard RFID, that this is not really true a standard RFID do not be equipped with sensors, you must design a customised tag. The same for the battery you can't connect a battery to a standard tag. You should provide more information related to the Considered RFID schema.
  We will clarify this point. In particular, what is 'standard' is the communication protocol (EPC-Gen2) and the RF channel (ETSI-302-201). Then we used 'commercial off-the-shelf' devices, both tags and reader. We have modified the tag (UHF tag, model Survivor B, from Confidex, that embeds an EM4325 chip from EM electronic, an a small battery) ony superficially: first, we configured its memory bank wirelessely, in order to make temperature measurements, and to increase its power sensitivity. Second, we painted the tag in white to reduce the radiative heat transfer.
  So we bought an industrial tag, but we have not designed a customised tag.
  2) you perform a measure of phase difference but it is not clear how. If you used inductive coupling rfid it is quite difficult. I suppose that you used a RF tag, and you claim that the reader is able to detect the phase difference. Can you please better explain how? You have to analyse the signals at the demodulator in order to detect the phase difference of did you use another technique? Please explain.
  We used an UHF RFID (working around 868 MHz). The reader (Impinj R700) has the capacity to read phase difference of arrival when reading a tag, out-of-the-box. Several other readers have this capacity, such as the ImpinjSR420 or the ThingMagic M6. Measurements of phase difference of arrival is largely used for RFID localization, see for example the introduction of this review :
  
  Le Breton M, Liébault F, Baillet L, Charléty A, Larose É, Tedjini S. Dense and long-term monitoring of earth surface processes with passive RFID — a review. Earth-Science Reviews. 2022 Nov 1;234:104225.
  
  We do not know, nor need to know, how the Impinj R700's electronic is designed in order to read the phase. For clarification, we will add a reference to research papers describing techniques to read the phase from a reader.
  
  3) did you take into account the attenuation introduced in the RF signal by the snow?
  Could you please clarify ? For which aim should we take the attenuation into account ?
  There are several sources of signal loss, such as, indeed, attenuation in the snow, but also detuning of the tag, reflexion at the surface, or destructive multipathing interferences. We have measured the signal strength data for each tag, but the data is messy: using signal strength is often a difficult method to exploit for RFID sensing outdoors because of its many influence factors (see again the review Le Breton et al. 2022). Indirectly, the signal strenght received by the tag can slightly influence the phase (and therefore our SWE measurement), when the tag enters non-linear behavior (for example when it receives very powerful signal). The non-linear behavior depends strongly on model of chip. With our tag, non-linearity occurs only at high power that are not reached in the study. And in any case, that would be a small source of noise.
  In short: The present study exploits only the phase difference of arrival (related to wave slowness), not the signal strength (related to attenuation).
  4) I suppose that you modified the reader, don't you? If yes please report the introduced customisations.
  No, we have used a commercial off-the-shelf reader, not customised. The method works with any reader that is able to read the phase difference of arrival.
  
  Citation: https://doi.org/10.5194/egusphere-2022-761-AC2
AC3: 'Comment on egusphere-2022-761', Mathieu Le Breton, 09 Jan 2023

Dear Christian Matzer,
First, thank you for your comments. We have just received the second referee's comment so I am answering both. We have processed your (see the answer in the attached document), they will help to improve the manuscript.
Regarding the multipathing, however, we propose an alternative approach.
We agree that there is a need to better understand the multipathing effect on RFID systems in snow contexts. However, this is far beyond the scope of the present study: multipathing is a general issue in RFID (very present for localization indoors for example). All the reflective elements should be a source of multipathing: The environment (snow surface, layers of snow, soil), and the installation itself (tags, light plastic supporting structure that holds the tag, large metallic structure that holds the reader). Any modification of geometry (snow depth) or dielectric constant (i.e., snow density, moisture content of the snow and soil) should modify the influence of multipathing on phase measurements. In practice, multipathing in RFID is more often mitigated than predicted.

It is mitigated in the study by using multiple antennas and tags (=spatial diversity). To further mitigate it in a future installation, we suggest (as you did) to use an array of tags placed very close to the ground (also necessary to reduce the thermal influence of the system). It should help average the spread due to spatial diversity, and reduce the reflection on the tags and on the supporting material.

We have a work in progress that explores more in depth the role of multipathing when using an RFID system is snow context, for which we will dedicate a communication in itself. The document attached presents (still unpublished) results of two experiments and one model: the amplitude of the phase variation induce by multipathing (from 1.4 to 3 rad) is coherent with the spread in our study (1.2 rad).

I also suggest to improve the link of our study, to existing studies related to multipathing in the snow or with RFID systems:
—The model used and preliminary results are also in:
Le Breton, M., 2019. Suivi temporel d’un glissement de terrain à l’aide d’étiquettes RFID passives, couplé à l’observation de pluviométrie et de bruit sismique ambiant (PhD Thesis). Université Grenoble Alpes, ISTerre, Grenoble, France. https://www.theses.fr/2019GREAU013 (page 144 to 156).
—Some methods exploit multipathing for snow, such as
Espín-López, P. F., Pasian, M., 2021. Determination of Snow Water Equivalent for Dry Snowpacks Using the Multipath Propagation of Ground-Based Radars. IEEE Geoscience and Remote Sensing Letters 18, 276–280. https://doi.org/10.1109/LGRS.2020.2974546;
Kulsoom, F., Dell’Acqua, F., Pasian, M., Espín-López, P. F., 2021. Snow Layer Detection by Pattern Matching in a Multipath Radar Interference Scenario. International Journal of Remote Sensing 42, 3193–3218. https://doi.org/10.1080/01431161.2020.1854890).
—Mitigating multipathing in RFID is a topic in itself. Some studies are dedicated to propose mitigation methods, such as
DiGiampaolo, E., Martinelli, F., 2020. A Multiple Baseline Approach to Face Multipath. IEEE Journal of Radio Frequency Identification 4, 314–321. https://doi.org/10.1109/JRFID.2020.3022576

In short, we encountered the multipathing issue and demonstrated a simple way to mitigate it. However, describing in detail the multipathing issue and how to further mitigate it, is a topic in itself which, in our opinion, should not be in the present paper. Instead we suggest to (a) propose a future installation that further reduce the multipathing issue, (b) to detail more about the causes of multipathing in the text, and (c) to link our observations with other studies that deal specifically about multipathing with snow or rfid.
Is this approach suitable to you ?
Best regards,

Citation: https://doi.org/10.5194/egusphere-2022-761-AC3

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

ED: Reconsider after major revisions (further review by editor and referees) (15 Jan 2023) by Franziska Koch

Dear Dr. Mathieu Le Breton,

we have received two valuable reviews for this manuscript. While the reviewers find the work in general interesting, they raise important issues, especially to explain several points better. The reviewers recommend major revision, which goes in line with my own reading of the manuscript.
The feedback provided in the reviewer assessments is important and should be taken carefully into account. In addition to the good points made by the reviewers, I have for now some points to be addressed carefully as well (see comments below).
As a next step, I would like to invite you to revise the manuscript.
Best regards,
Franziska Koch

General comments:
- Please use for snow depth units metres or centimetres (not mm as you did for example in Figure 4 or Figure 5).
- Any comments on the maximum SWE to be monitored with your method?
- Regarding language issues, a thorough copy-editing by a (professional) native speaker is recommended.
- In the current version of your RFID measurement technique, you need some corrections to depict / process the ‘right’ SWE (Section 3.3). Can you please discuss / outline in more detail, which could be a way to avoid such corrections (e.g. fixing to values of manual snow pits (l. 249), picking carefully the wet/dry periods…) in future to have a fully independent technique?
- Section 3.3. (especially to get to your final result in Figure 8) is difficult to follow and understand. Please revise in terms of explaining all steps for a better understanding.

Specific comments:
- l. 27: SWE ‘expressed as surface density’ is misleading. Please reformulate.
- l. 40ff and further locations in the manuscript: Regarding GPS / GNSS  please be consistent, use either or – I would recommend to use GNSS.
- l.45: It is true for upGPR setups, that they are expensive; however, this is not true for low-cost GNSS setups, especially when you just consider the low-cost equipment – which would be in the range or even in a lower than the RFID tags and antenna setup (power supply, management unit, processing of data etc. not included…). So please just mention upGPR in the context of being expensive.
- l. 46: It should be upward-looking GPRs
- Table 1: Regarding the row GNSS please change the following in the column ‘area’: m² (larger areas would correspond to snow depth recordings via GNSS reflectometry).
- l. 103: The permittivity range corresponds to dry snow. Please add ‘dry’. It can be certainly higher for wet snow.
- l. 141f: What is the size of the RFID tags? Please mention also the temporal resolution of your final SWE measurements (1 min? 6 hours? daily? – was not so clear to me in the text).
- l. 145: Which depth did the snow layers have in your experiment? This is also unclear in l. 175f.
- l. 150f: Is there a reason why you chose these densities? Why did you not choose for example fresh snow with approx. 100 kg/m³ in your experiment?
- l. 154: Please give some more information on the two vertical arrays (diameter of poles, length of arms where RFID tags were installed on etc.). Please discuss at a later point in the manuscript also more in depth potential uncertainties regarding the installation on the vertical arrays.
- l. 167: I guess you rather mean the monitoring of the melt out of individual RFID tags? Please clarify.
- l. 200ff: Please explain better how the final (median) SWE value was derived. Not sure, if I understood correctly, but did you calculate the median of the tag readings at different installed height. If so, is it than valid to use a median calculation? Please clarify.
- Figure 5: Why are the readings of the RFID tag at 13 and 28 cm installation height missing? The connection with the colours regarding antenna 1 and antenna 2 is not easy to grasp. I suggest to use colours, e.g. in the shades of blue to green for connected to reader antenna 1 and e.g. yellow to red for reader antenna 2.
- Figure 6: Why don’t you include RFID readings from installation heights up to 43 cm in this figure as according to the measured snow height during the season these tags should also be covered by snow most of the time. Regarding the precipitation panel: How did you separate into rain and snow?
- L. 280ff: Please relate the reported errors with the entire snow depth / SWE of the experiment or outdoor test.
- L. 288f: Why did you expect that the temperature within the snowpack correlates with the air temperature?
- Figure 10: According to Fig. 8, the snow depth is over a long period higher than approx. 33 or even 43 cm. However, looking at Figure 10, the RFID tags above an installation height of 18 cm seem not to be within the snowpack as they react similar to the air temperature. Please clarify. The 0°C line in these plots is too weak and should be represented better in this figure. Moreover, please add units to the installation heights in the graph.

Hide

AR by Mathieu Le Breton on behalf of the Authors (26 May 2023) Author's response Author's tracked changes

EF by Vitaly Muravyev (01 Jun 2023) Manuscript

ED: Referee Nomination & Report Request started (04 Jun 2023) by Franziska Koch

RR by Christian Mätzler (17 Jun 2023)

ED: Publish subject to technical corrections (18 Jun 2023) by Franziska Koch

AR by Mathieu Le Breton on behalf of the Authors (19 Jun 2023) Manuscript

Journal article(s) based on this preprint

04 Aug 2023

Monitoring snow water equivalent using the phase of RFID signals

Mathieu Le Breton, Éric Larose, Laurent Baillet, Yves Lejeune, and Alec van Herwijnen

The Cryosphere, 17, 3137–3156, https://doi.org/10.5194/tc-17-3137-2023,https://doi.org/10.5194/tc-17-3137-2023, 2023

Short summary

Mathieu Le Breton et al.

Viewed

Total article views: 680 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
454	204	22	680	4	7

HTML: 454
PDF: 204
XML: 22
Total: 680
BibTeX: 4
EndNote: 7

Views and downloads (calculated since 22 Aug 2022)

Month	HTML	PDF	XML	Total
Aug 2022	99	44	6	149
Sep 2022	57	29	2	88
Oct 2022	67	19	4	90
Nov 2022	32	11	1	44
Dec 2022	16	5	0	21
Jan 2023	61	13	4	78
Feb 2023	38	23	0	61
Mar 2023	13	6	0	19
Apr 2023	16	11	0	27
May 2023	15	6	0	21
Jun 2023	19	10	1	30
Jul 2023	21	27	4	52
Aug 2023	0

Cumulative views and downloads (calculated since 22 Aug 2022)

Month	HTML	PDF	XML	Total
Aug 2022	99	44	6	149
Sep 2022	57	29	2	88
Oct 2022	67	19	4	90
Nov 2022	32	11	1	44
Dec 2022	16	5	0	21
Jan 2023	61	13	4	78
Feb 2023	38	23	0	61
Mar 2023	13	6	0	19
Apr 2023	16	11	0	27
May 2023	15	6	0	21
Jun 2023	19	10	1	30
Jul 2023	21	27	4	52
Aug 2023	0

Viewed (geographical distribution)

Total article views: 679 (including HTML, PDF, and XML) Thereof 679 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 04 Aug 2023

Download

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Preprint (1750 KB)
Metadata XML

Short summary

We monitor the quantity of snow on the ground with passive radiofrequency identification (RFID) tags. Tags are produced in billions every year, they are cheap and small. They are interrogated wirelessly across a snowpack by a fixed reader above. The changes in the radiofrequency phase delay reflects the changes in quantity of snow – the snow water equivalent. Besides, a temperature sensor in each tags monitored the snow temperature.


Total:	0
HTML:	0
PDF:	0
XML:	0

Monitoring snowpack SWE and temperature using RFID tags as wireless sensors

Journal article(s) based on this preprint

Interactive discussion

Interactive discussion

Peer review completion

Suggestions for revision or reasons for rejection

Journal article(s) based on this preprint

Viewed

Viewed (geographical distribution)