the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Evaluating the Utility of Active Microwave Observations as a Snow Mission Concept Using Observing System Simulation Experiments
Eunsang Cho
Carrie M. Vuyovich
Sujay V. Kumar
Melissa L. Wrzesien
Rhae Sung Kim
Abstract. As a future satellite mission concept, active microwave sensors have the potential to measure snow water equivalent (SWE) with advantages including finer spatial resolution and improved capabilities in deeper snowpack and forest-covered areas as compared to existing missions (e.g., passive microwave sensors). In mountainous regions, however, the potential utility of spaceborne active microwave sensors for SWE retrievals particularly under deep snow and forest cover has not been evaluated yet. In this study, we develop an observing system simulation experiment (OSSE) that includes the characterization of expected error levels of the active microwave-based volume-scattering SWE retrievals and realistic orbital configurations over a western Colorado domain. We found that active microwave sensors can improve a root mean square error (RMSE) of SWE by about 20 % in the mountainous environment if the active microwave signals with a mature retrieval algorithm can estimate SWE up to 600 mm of deep SWE and up to 40 % of tree cover fraction (TCF). Results also demonstrated that the potential SWE retrievals have larger improvements in tundra (43 %) snow class, followed by boreal forest (22 %) and montane forest (17 %). Even though active microwave sensors are known to be limited by liquid water in the snowpack, they still reduced errors by up to 6–16 % of domain-average SWE in the melting period, suggesting that the SWE retrievals can add value to meltwater estimations and hydrological applications. Overall, this work provides a quantitative benchmark of the utility of a potential snow mission concept in a mountainous domain, helping prioritize future algorithm development and field validation activities.
- Preprint
(3625 KB) -
Supplement
(1210 KB) - BibTeX
- EndNote
Eunsang Cho et al.
Status: final response (author comments only)
-
RC1: 'Comment on tc-2022-220', Melody Sandells, 22 Dec 2022
This paper demonstrates (mostly) an improvement in simulated SWE from assimilation of synthetic observations derived from an active microwave sensor through an OSSE. The analysis is based on the difference between a calibrated nature run and uncalibrated open loop run, with an assessment of the effects of SWE retrieval and forest cover fraction limits for a fixed satellite orbit and error budget configuration (with extended error budget in the supplementary material). The analysis is performed over a region in Colorado over a range of elevations, vegetation densities and snow types. The paper is very well written and easy to follow, with some excellent visual representations of the results. As the authors highlight, the results depend on the assimilation system used and this complements the Garnaud et al. (2019) study well. The results are useful to demonstrate the necessary retrieval performances for the design of retrieval algorithms and as such is worthwhile for publication.
It would be great to elaborate on these discussion points as part of this paper:
‘We found that active microwave sensors can improve a root mean 20 square error (RMSE) of SWE by about 20% in the mountainous environment if the active microwave signals with a mature retrieval algorithm can estimate SWE up to 600 mm of deep SWE and up to 40% of tree cover fraction (TCF).’ What would be really good for this paper is a small discussion of to what extent the retrieval algorithms can actually fulfil the different criteria e.g. in Figure 10, it might be possible to colour-code the boxes behind the circles according to whether current retrieval methods can achieve this or not. This will then provide impetus to improve retrieval methods for other conditions.
Following on from this point, on line 100-101: ‘to set priorities related algorithm developments.’ Also line 361: ‘thus different priorities for the algorithm development may be required by seasonal snow characteristics’. It would be really useful to have some concrete priorities identified from this study or at least a discussion of the factors. Fig 10 shows relative level of improvements but some may be out of reach so it’s better to target the middle range improvements, or concentrate on montane forest because this is the largest % cover – at least for this site, but what about from a global perspective? The largest improvements are for tundra high forest fraction limits, but what does this mean in reality? How much of the tundra has dense forests according to the updated Sturm classification?
Figure 3 seems to suggest there is more to be gained from improving the physics of the model than in developing assimilation techniques. Is this fair to say? For example, the OL in the mid elevation ranges overestimates melt by a lot in March. Why is the assimilation unable to recover this lost mass if the microwave data has SWE information? Do the OL and NR agree on the timing of the melt? If not, would you get better performance by detecting melt from the NR and forcing the DA model to be cold / use the observations when the NR is not melting but OL is.
Figure 4 – Is it fair to say there’s no point in assimilating over bare ground? There seems to be negligible difference from the OL - why is this, especially as there are no SWE limits here?
Line 297 / Fig 3. Why does the assimilation degrade the performance for shallow SWE?
In terms of specific comments for the paper, it would be good to address the following:
- Lines 40-45. Cut all these: certainly the acronyms (these aren’t used again) but listing the satellites doesn’t add anything to the paper so these can be removed. The fact that there are numerous instruments plus references are enough.
- Remove TCF from the abstract and place the acronym definition at line 93
- Line 93. Explain where the 40% comes from. Why is this considered achievable? It seems to be based on a LiDAR study, but this isn’t referenced.
- Line 100-101 - related -> related to
- Line 119. Would be useful to state under what sensor assumptions here
- Figure 1. This needs a larger map to locate the region. Is this part of a larger modelling domain? If so, this should also be highlighted.
- Line 145. Clarify NR and OL are at same spatial resolution
- Lines 215-222. A supplementary figure demonstrating the flow of TAT-C derivation of the swaths would be helpful
- Line 225. Why was the EnKF chosen for this work? (Is the EnKF acronym used after its definition? If not, it’s not needed).
- Equation 1 gives temporal RMSE – please could you include the equation for spatial RMSE in Fig 9 or adapt this one?
- Fig 3 – would be nice to have box plots of SWE to see how much of the study area is affected by these limits. Also, please put this figure (+others) through a colour-blind checker. Perhaps make the OL a thicker line too.
- Line 298-300. Please could you rephrase this? I think this is saying that the improvements come from better retrieval algorithms rather than the better spatial resolution with active, but I’m not sure.
- Line 306-307. Perhaps remove ‘during a melting period’ as the improvements are higher in the accumulation period (unless the intent is to highlight that any improvements go against expectations, in which case make more of it!)
- Figure 5. OL melting period spot needs to be black rather than grey.
- Lines 316-317. Use boxes to draw attention to the regions in Figure 6.
- Fig 6. These appear have a different ratio to Fig 1. Please make sure there are the same number of x,y pixels in both. There also appears to be some striping artefacts in Fig 6f (thin white vertical strips of 1-2 pixels) – what is causing this?
- Fig 7b – does white colour mean no change? Would be good to include this in the colourbar.
- Fig 8 – would be useful to extend the OL median up through (behind or dashed in front of) the other bars for easier comparison
- Line 397 – ‘realistic orbital configurations for a volume scattering SAR mission developed using TAT-C’: What is the (approximate?) repeat time in this study?
- Line 422 – Please rephrase ‘and there are values in the montane forest (17%) due to deep snow capability’
Finally, I’d like to commend the authors for Figs 2, 5 and 9 – these are informative / novel ways of representing the concepts and results (at least to me).
Citation: https://doi.org/10.5194/tc-2022-220-RC1 -
RC2: 'Comment on tc-2022-220', Anonymous Referee #2, 12 Jan 2023
This study uses an OSSE to estimate improvement of SWE retrievals if satellite-based SAR data from an X- and/or Ku-band sensor was assimilated in an LSM. The authors use a calibrated version of the LSM (nature run) and compare it to an uncalibrated run (open loop) and a Data Assimilation run using synthetic observations with simulated extents using the TAT-C software.
This paper is an excellent contribution to the overall field of SAR missions to retrieve SWE. It is very well structured and easy to read with excellent supporting figures. The fact that the analysis was done in a very different study area (Western Colorado), this study complements other similar studies (Garnaud et al., 2019). It provides information on what SWE limit and TCF the SAR mission should be able to detect for this specific domain in Colorado.General comments:
One important limitation to this study that is not really discussed and I feel should be feasible with the current OSSE is the inherent geometrical limitations of SAR sensors (i.e. shadow/overlay) which complicates the retrieval of surface properties in mountain regions such as the region of interest in this study. The TAT-C software should allow to estimate the incidence angle and with the SRTM data, it should be feasible to mask out these blind spots. This evaluation might be outside the scope of this study but I feel it should be discussed a bit further as a limitation of this study and be a future consideration. This would increase the number of masked grid cells that would not have SWE retrievals from satellite observations.
To add to the other reviewer's comment: it would be useful to set the priorities of this study and give examples of what kind of mission would be relevant for these priorities since there is no "one-size fits all" mission. As mentionned the range of SWE values given for the Tundra class is not what you will find in other Tundra environments. Would a mission that would provide such improvement for the Tundra high SWE values work for other Tundra environments knowing the SWE values, snow stratigraphy (grain type/microstructure) and landscape conditions are very different? Adding some discussion on the specific snow conditions of the AOI would be relevant to this study.
To add, this study only focuses on SAR retrieval from backscatter values. But what if the sensor has single-pass altimetry/interferometry capabilities? This would help to retrieve snow depths at least, especially during melt season from differential DEMs. Wouldn't that improve the estimation of SWE from the LSM? This might again be outside the scope of this study based on the priorities but I feel this should be discussed as SAR missions are very rich data sources.
Specific Comments:
L.39-44: No need to list the different PMW sensors here, I would keep "Historically, a series of satellite-based passive microwave radiometers have been used to develop spatially distributed snow depth and SWE information (Cho et al., 2017; Derksen et al., 2005; Foster et al., 2005; Vuyovich et al., 2014).
l.220: to make this OSSE more realistic, what would be the incidence angle range of such a SAR mission configuration?
Fig 4.: provide the TCF ranges for the different elevations. I suspect there is not much TCF over low and mid elevations where there is not much improvements in runs with more TCF capability.
Fig 7.: Change RMSD to RMSECitation: https://doi.org/10.5194/tc-2022-220-RC2
Eunsang Cho et al.
Eunsang Cho et al.
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
311 | 125 | 11 | 447 | 41 | 4 | 3 |
- HTML: 311
- PDF: 125
- XML: 11
- Total: 447
- Supplement: 41
- BibTeX: 4
- EndNote: 3
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1