the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Dec 2024
Brief communication: Improving lake ice modeling in ORCHIDEE-FLake model using MODIS albedo data
Abstract. The Flake lake model embedded in the ORCHIDEE land surface model was recently updated to better represent winter ice cover. MODIS albedo data and the Great Lakes ice cover fraction dataset over the Laurentian Great Lakes were used to calibrate and validate a new parameterization of the lake albedo accounting for a partial ice cover fraction. The results show large improvements in the simulation of the ice phenology of 200 lakes of various sizes reported in the Global Lake and River Phenology database. The agreement with the observations is improved for all lake size categories, with the largest and deepest lakes showing larger error reductions on the duration of the ice cover period. This study highlights the importance of considering partial ice cover to correctly model lake albedo in cold regions and thus to simulate realistic mass and energy exchanges at the land-atmosphere interface.
- Preprint
(859 KB) - Metadata XML
-
Supplement
(831 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-2907', Anonymous Referee #1, 08 Jan 2025
General Comments:
The authors of the paper are utilizing a very commonly used MODIS albedo retrieval and parameterizing this variable into the FLAKE model that is then incorporated into the ORCHIDEE land surface model. The aim of this manuscript is to highlight that by parameterizing the albedo retrievals from MODIS that it better constrains the albedo parameter of the lake ice thickness model, thereby improving the ice-on and ice-off dates, providing a better assessment of ice phenology for lakes of various sizes, from small up to the Laurentian Great Lakes. In my assessment, the authors have done well to incorporate the MODIS data into the FLAKE model used in the ORCHIDEE land surface model, and I do believe that the improvements they are reporting are conceivable.
The presentation and assessment of the outputs of the model, in the Observed/Prior/Post experiment is lacking in detail, glossing over much of the potential detail that could have been provided for both the Great Lakes GLIC dataset, and the GLRP database. There needs to be significant work done to a) show the inter-year distribution of albedo parameter improvement in the Great Lakes dataset, and b) show how the improvement of ice phenology parameters varied across distance and time. The authors do well to describe the spatial distribution of lakes within the study among multiple countries, but do not present a study site figure, or segment the results by administrative or geographical boundaries (which have influences on how and the frequency of phenology reports). Section 3.1. presents a significant improvement to the model since the albedo is being parameterized, but it does not read as an important contribution, mainly because it is only shown through an average of 10 years for each lake. More detail in the inter-year distribution of results needs to be shown.
With improvement to the presentation of the study site, methods, and results section this manuscript would be an important contribution to the body of literature.
Specific Comments:
Page 1 Line 19 – 20: “The role of lakes… demonstrated in various works”
Briefly describe what the role of lakes actually are, not that other people say they’re important.
Page 2 Lines 37-38: “In these large-scale models… linked to the varying bathymetry”
Please provide references for this statement.
Page 2 Line 44: “free water and ice show very different characteristics”
In what way? Please be specific
Page 2 Lines 47 – 48: “When the air temperature falls… on its bathymetry and weather conditions”
This is an oversimplification of how freezing occurs – what about general density of water at 4 degrees, and the mixing that happens?
Page 2 Line 60: “daily albedo time series at a few hundred metres”
Be specific- how many metres?
Page 3 Line 89: “Crossing the albedo raster images”
Do you mean “overlaying”?
Page 3 Lines 90 – 91: “To avoid radiance contamination by the lake shore pixels… to mask such mixed pixels”.
This is a surprising way to mask out all non-water pixels. For instance, what about exposed soil/sand? Those pixels would not have vegetation but would have higher albedo than water, so it could still erroneously be included for small and medium lakes.
Page 3 Line 100 – 102: The GLIC provides a slew of ice concentration data – but how was it derived? It’s important to note that it was derived using a combination of paper charts, SAR, visible/infrared imagery and met data, not just interpolated.
Page 4 Line 110: “SYKE”
Define.
Page 4 Line 124: “LMDZ”
Define.
Page 7: There is no workflow diagram shown here – which is necessary to help the reader understand the new implementation and model experiment.
Page 7 Lines 206 – 207: “the only two lakes we could formally identify in the HydroLAKES database”
This needs a better explanation. Are these the only two lakes with names? How does one identify the lakes?
Page 7 Lines 213 – 214: larger surface roughness observed… and reduce the overall surface reflectance”
Would the surface roughness increases not actually increase reflectance as well? Surface roughness on water could cause white caps?
Page 8 Line 238 – 239: “Here, we present daily values averaged over the 20098-2018 period…”
There are a lot of dates averaged together here, and one year may be very different from another, especially considering the size of the Great Lakes. There is no presentation of the standard deviation of the albedo values, the interquartile range of the deviation of the albedo and no presentation of the spread of observations, prior or post. The results that are presented are a significant wrapping up of the data, presenting 10 years in one line.
Page 8 Line 242: “The results clearly show”
Conjecture, please avoid using the word “clearly”.
Page 9 Lines 244 – 245: “The RMSEs are considerably reduced… Supporting Information)/.
Some of this needs to be included in the manuscript – it’s important for the reader to see.
Page 8 Line 247: “then reducing”
Do you mean “thus reducing”?
Page 8 Line 250: “not shown here)”
When highlighting a particular year to discuss deviation in expectations of results, this needs to be shown.
Page 8 Line 250 – 251: We have seen that for al these lakes, the spatial variability is very large…”
How can this be shown within the manuscript?
Page 9 Line 262: “according to their mean depth”
Reminder the reader what shallow, medium and deep mean here
Page 9 Lines 264 – 269: “Given that deep lakes… fractional ice cover parameterization”
This needs to be included in the methods, not introduced in the results.
Page 9 Line 270: “Figure 2 clearly…”
Conjecture
Page 9 Line 271 – 279: This description of the data would be well served in a table that is included in the manuscript.
Page 9 Line 280 – 283: “Besides, the fact that the ice-off… most of the lakes observed”
Is this not more likely because the albedo changes significantly quickly in the spring compared to the Fall, where ice onset may be congelation ice with albedo increasing less quickly?
Figure 2: Are there values for data representing one year? Multiple years? Is there more or less agreement based on the year observed? What about the RMSE of the sim-observed values? There are a lot of potential tests that can be run to showcase how the correction is behaving across space and time. Also, GLRP is worldwide, you could segment the data across administrative or geographic boundaries.
Figure 2 caption, Line 289: “GLIC database”
Is this not the GLRP database? -
RC2: 'Comment on egusphere-2024-2907', Anonymous Referee #2, 27 Jan 2025
The Flake lake model within the ORCHIDEE land surface model has been updated by the Authors to better simulate winter ice cover. Using MODIS albedo data and the Great Lakes Ice Cover fraction dataset, the authors calibrated and validated a new lake albedo parameterization that accounts for partial ice cover. This update significantly improved the simulation of ice phenology for 200 lakes of various sizes, as reported in the Global Lake and River Phenology database. The improvements were notable across all lake sizes, with the largest and deepest lakes showing the greatest reduction in error for ice cover duration. The study underscores the importance of considering partial ice cover to accurately model lake ice.
Overall, the work presents a useful contribution to large-scale modelling in partially ice-covered regions but does leave some lack of clarity on how this performs in fully ice-covered regions (e.g. by implying Lake Erie did not improve as much as the rest due to the more uniform ice cover). Are the 200 lakes this is tested on in the warmer regions that have partial ice cover? Or the full GLRP, which includes many northern lakes with full ice cover? And are the SYKE lakes in southern Finland? Or in the north with solid ice covers? While acknowledging this is a brief communication, some important details have been omitted, and once addressed, this paper could be considered for publication.
Line 73: Hydrolakes – While agreed that it is the best option available, the limitations in lake depths should be noted here as outlined in Messager et al., 2016 (estimating the depths based on limited information in some locations), and what the implications are for your study.
Line 90: Perhaps comment on why grouping the lake into similar depth tiles rather than just run for each 500m pixel? What if there is fractional ice coverage within each depth grouping? Also is the MODIS water mask not a viable option rather than using a vegetation index?
Line 132: is the medium category is missing here?
Line 180: “In Flake, the minimum and maximum albedos are equal to 0.1 and 0.6 respectively, for both snow and ice. They were revised by Bernus and Ottlé (2022) following Semmler et al. (2012) and Pietikaïnen et al. (2018), and set to 0.3-0.5 and 0.77-0.87”
These ice albedos are a lot lower than what have been published for lakes in Central Ontario. Are those albedo values to represent a broken ice cover only? or only during the melt season?
~ line 200 – I think it would be pertinent to show the results here. Also, add more explanation – if Lake Erie has the largest ice fraction, and the worst RMSE does that not indicate that your new method is not suitable for lakes with a full ice cover and should only be used on a partial ice cover? Or is the albedo selected based on the ice cover fraction? This needs some clarity.
Line 205: “Therefore, we compared these specific dates estimated from the MODIS albedo time series to those observed in the GLIC database and found that over the period 2010 - 2015, our method has an average error of 2 days for Lake Nilakka and 5 days for Lake Nasijarvi, the only two lakes we could formally identify in the HydroLAKES database.”
Unclear – why are there Finnish lakes in the GLIC database? Clarify why only the 5 years? Can you not use maps to identify the other lakes in the database? The explanation here needs more clarity. And again, does this indicate that these lakes had partial ice cover? Or did the results improve for full ice cover here, but not for Lake Erie?
Line 230: which lakes? Where? From what I recall the GLRP is not really global, it’s mostly North American - are your test lakes mainly in Canada then? in regions with full ice covers? The work needs a map to show which lakes you are using.
Line 249: “This is particularly the case for Lake Erie in 2008 and 2015 or for Lake Huron in 2015 (not shown here).”
Consider showing the examples and why the warm/cold year performs well/not well. Perhaps in the supplemental section if length is an issue for the brief communication. Variability is important.
Line 265: “Given that deep lakes present partial coverage most of the time and that the observations may not be representative of the whole lake, we decided, after various tests, to diagnose the ice phenology dates with the simulated surface temperature as was done in Bernus and Ottlé, 2022.”
Which deep lakes have partial coverage? Where? How far north? this is something that needs to be clearly addressed throughout.
Figure 2: Do you mean the global river and lake ice database? Not the great lakes ice cover?
Citation: https://doi.org/10.5194/egusphere-2024-2907-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 126 | 40 | 6 | 172 | 20 | 11 | 8 |
- HTML: 126
- PDF: 40
- XML: 6
- Total: 172
- Supplement: 20
- BibTeX: 11
- EndNote: 8
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1