the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Forward Modelling of SAR Backscatter during Lake Ice Melt Conditions using the Snow Microwave Radiative Transfer (SMRT) Model
Justin Murfitt
Claude Duguay
Ghislain Picard
Juha Lemmetyinen
Abstract. Monitoring of lake ice is important to maintain transportation routes but in recent decades the number of in situ observations have declined. Remote sensing has worked to fill this gap in observations, with active microwave, particularly synthetic aperture radar (SAR), being a crucial technology. However, the impact of wet conditions on radar and how interactions change under these conditions has been largely ignored. It is important to understand these interactions as warming conditions are likely to lead to an increase in the occurrence of slush layers. This study works to address this gap using the snow microwave radiative transfer (SMRT) model to conduct forward modelling experiments of backscatter for Lake Oulujärvi in Finland. Experiments were conducted under dry conditions, under moderate wet conditions, and under saturated conditions. These experiments reflected field observations during the 2020–2021 ice season. Results of the dry snow experiments support the dominance of surface scattering from the ice-water interface. However, conditions where layers of wet snow are introduced show that the primary scattering interface changes depending on the location of the wet layer. The addition of a saturated layer at the ice surface results in the highest backscatter values due to the larger dielectric contrast created between the overlying dry snow and the slush layer. Improving the representation of these conditions in SMRT can also aid in more accurate retrievals of lake ice properties such as roughness, which is key for inversion modelling of other properties such as ice thickness.
- Preprint
(2315 KB) - Metadata XML
- BibTeX
- EndNote
Justin Murfitt et al.
Status: final response (author comments only)
-
RC1: 'Comment on tc-2023-60', Anonymous Referee #1, 23 Jun 2023
This paper investigates the forward modeling of backscattering variation during lake ice melt conditions using the Snow Microwave Radiative Transfer (SMRT) model. In-situ measurements were obtained during three Intensive Observation Periods (IOPs), comprising one period with dry snow conditions and two periods with wet snow conditions, to aid in establishing input parameters for the SMRT model. The study aimed to identify the dominant factors influencing each period through modeling efforts. However, since the SMRT model integrated volume scattering and surface scattering separately without considering the interaction. It need to be applied carefully into the sensitivity analysis. Several concerns regarding the model's application are outlined.
- In Figure 2, the time series plot of backscattering reveals higher HH values compared to VV values for IOP I and IOP III. This indicates that volume scattering dominates in these periods. If surface scattering were dominant, higher VV values would be expected, as demonstrated in the modeling work shown in Figure 5. Another approach to verify this is by examining the cross-polarization (cross-pol) of the Sentinel data. In the case of volume scattering, cross-pol values are significantly higher than those of surface scattering. Therefore, during the dry snow period, solely considering the surface effect is insufficient; additional factors such as volume scattering need to be taken into account.
- On page 4, line 123, it is mentioned that the current SMRT model does not implement cross-polarization calculation. However, despite this limitation, including the measurement Sentinel data of cross-pol would still be valuable in enhancing our understanding of the scattering mechanism. Therefore, I recommend adding cross-pol information to Figure 2. Additionally, to provide a comprehensive reference for backscattering levels, it would be beneficial to include data from the entire season, including the ice-off period, pre-season, and post-season. This would offer a more complete perspective.
- On page 4, line 128, the paragraph is somewhat confusing. The initial sentences address the dry snow condition, where the snow is considered a two-mixture random medium consisting of ice grains and air. To provide clarity, the sentences need to be rephrased. Subsequently, the paragraph transitions to discussing the wet snow condition. However, it then reintroduces the SHS model and exponential model, which are specifically applied to the dry snow condition. I recommend reorganizing this paragraph to separate the discussion of the dry and wet snow conditions and clearly specify the models chosen for each layer. Additionally, it should be noted that all the assumptions and models described in the original Picard's paper for the wet snow condition pertain to the passive microwave remote sensing regime, which calculates brightness temperature. Applying them directly to backscattering may not yield accurate results.
- On page 9, line 243, it should be noted that the remote sensing data utilized in this paper is backscattering. The temperature of each layer has minimal impact on backscattering and should be clarified from the outset. Consequently, it is not necessary to consider temperature as a factor in the tables and in-situ measurements. Furthermore, I suggest using the term "snow media correlation length" instead of "Pex" in the sentence to maintain consistency and clarity.
- Figure 3 illustrates the modeling approach for a multi-layer structure during three IOP periods. To enhance clarity, it would be beneficial to consolidate the information regarding the selected model and input parameters for each layer in a single location. Currently, this information is dispersed across sections 2.1 and 2.5, making it challenging to piece together. It would be helpful to provide a clear explanation of the distinctions between dry snow, snow ice, and pure ice. It seems that all three conditions involve a mixture of ice and snow, differentiated by varying volume fractions. Please provide further elaboration on this matter.
- As previously mentioned, the temperature of each layer has a minimal effect on the backscattering calculation compared to the brightness temperature. Therefore, the introduction of the CLIMo model on page 9, line 250 does not appear necessary for this purpose.
- Kindly provide further elaboration on the snow ice porosity. Does a 10% snow ice porosity mean that 10% of the volume consists of air and 90% consists of pure ice? To enhance clarity, I recommend color coding each number to clearly indicate which parameters are derived from in-situ measurements and which ones are based on ad-hoc best fit parameters. This would help differentiate between the two sources of data and improve the overall understanding of the parameter selection process.
- Figure 8 requires additional description to improve clarity. Is (a)(c ) referring to IOPIIa, and (b)(d) referring to IOPIIb? It would be helpful to provide clarification regarding the color blocks. Does red represent VV, while blue represents HH? Furthermore, when comparing Figure 8 (c) and (d): despite varying VWC values from 0 to 1%, the backscattering remains almost identical and shows no sensitivity to RMSH or correlation length. This suggests that surface scattering is not the dominant factor in the overall backscattering for that particular case.
Minor comments:
- Please keep the color consistent through out the figures. Eg. Figure 2/5/10 using blue for HH and orange for VV. But figure 4 use orange for HH and blue for VV
- For figure 5,6,8,9, the color block of the observed HH/VV. Is the max-min range of the HH/VV or the standard
Citation: https://doi.org/10.5194/tc-2023-60-RC1 - AC1: 'Reply on RC1', Justin Murfitt, 22 Aug 2023
-
RC2: 'Comment on tc-2023-60', Anonymous Referee #2, 11 Jul 2023
In this study, the radar backscatter response due to liquid water on lake ice is evaluated using in situ snow and ice measurements, satellite radar data and a radiative transfer model. This works builds on decades of radar remote sensing of lake ice including recent findings on the dominant scattering mechanisms from ice and snow under dry conditions, and expands these findings to include the impact of liquid water. To assess the impact of liquid water they collected snow and ice measurements on a lake in Finland throughout the winter and used those observations to initialize the SMRT model which estimates a backscatter signal through a snow and ice medium given several input parameters. They tested the sensitivity of the model to these parameter values and compared results to Sentinel-1 C-band SAR data. The results helped confirm that the ice-water interface is the dominant scattering signal in dry snow/ice conditions, but when liquid water is present it becomes the dominant impact on the signal.
The study is well designed and conducted and provides important scientific findings on the impact of liquid water on radar signals, which is critical for interpreting spaceborne signals. I believe the is a good contribution to the literature, though I do have some minor concerns and some suggestions to improve clarity.
Because this work is building off of several recent studies, it feels like some of the details and background are missing which makes it difficult for someone to step into without having read all of the previous literature. For example, there is quite a bit in the introduction on recent work showing the dominant scattering mechanism is due to the ice-water surface scattering. That could be reduced to a sentence or two with the relevant references. A higher-level summary of radar remote sensing of lake ice – i.e. all of the potential scattering mechanisms and their contribution, which frequencies have worked best, what is the state of the art in terms of detection of lake ice thickness and other properties – would be useful. Other specific examples are given below.
There are so many parameters affecting the signal that are being tested in this sensitivity study that it is at times difficult to follow. This is with in situ measurements providing a lot of model input. In the discussion it says it describes the "importance of properly parameterizing all aspects of roughness for the different interfaces" (line 485) and that “accurate information on the VWC throughout the snowpack is crucial” (line 576), but it's not clear what the relative impact of these properties (and others) have on the results. How would these parameters be constrained in a larger scale application of spaceborne SAR data for lake ice?
Comments:
Lines 75, 79, 495: There’s no Murfitt 2023 citation in the reference list.
Line 92-93: Sentence starting with “However, these experiments…” is vague. Can you explain in a little more detail what the limitations in snow cover representation were?
Figure 1: What is the green dot on the inset map?
Lines 147-148: RMSH seems to be a key parameter, used extensively to throughout the analysis and results, but the description here is fairly minimal. Later (line 391) RMSH is described as a key property influencing backscatter as demonstrated by previous studies. It would be good to provide more of that background up front.
Line 176: I assume the 82 EW and 69 IW SAR images make up the 151 Sentinel-1 images acquired, but right now it reads like a list. I would suggest editing it to read “151 Sentinel-1 (C-band, 5.405 GHz), comprised of 82 Extra Wide (EW) swath HH-pol and 69 Interferometric Wide (IW) swath VV-pol, SAR images…”
Line 221: Maybe I missed this, but how was water content measured?
Line 409: Why is one of the values negative (-11.8) and the other not (9.19)? Since you say “decrease” maybe the negative sign isn’t needed?
Conclusion: There is not a clear take-home message. I recommend that conclusions and relevant findings from this work be clearly stated in this section, which I think would help strengthen the paper.
Citation: https://doi.org/10.5194/tc-2023-60-RC2 - AC2: 'Reply on RC2', Justin Murfitt, 22 Aug 2023
- AC3: 'Reply on RC2', Justin Murfitt, 22 Aug 2023
Justin Murfitt et al.
Justin Murfitt et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
291 | 96 | 21 | 408 | 6 | 6 |
- HTML: 291
- PDF: 96
- XML: 21
- Total: 408
- BibTeX: 6
- EndNote: 6
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1