the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Submitted as: model experiment description paper 07 Jan 2021
Submitted as: model experiment description paper | 07 Jan 2021
Impact of Initialized Land Surface Temperature and Snowpack on Subseasonal to Seasonal Prediction Project, Phase I (LS4P-I): Organization and Experimental design
- 1University of California – Los Angeles, CA 90095, USA
- 2Institute of Tibetan Plateau Research, Chinese Academy of Sciences, China
- 3CNRM, University of Toulouse, Météo-France, CNRS, Toulouse, France
- 4University of Arizona, Tucson, USA
- 5Earth System Science Interdisciplinary Center (ESSIC), University of Maryland, College Park, USA
- 6Japan Agency for Marine Earth Science and Technology (JAMSTEC), Japan
- 7Lawrence Livermore National Laboratory, Livermore, CA 94550, USA
- 8International GEWEX Project Office, George Mason University, USA
- 9Hans Ertel Centre for Weather Research, Germany
- 10Korea Institute of Atmospheric Prediction Systems, South Korea
- 11Institute of Atmospheric Physics, Chinese Academy of Sciences, China
- 12Meteorological Research Institute, Japan Meteorological Agency, Japan
- 13Hokkaido University, Japan
- 14Indian Institute of Tropical Meteorology, Ministry of Earth Sciences, India
- 15Bureau of Meteorology, Australia
- 16European Centre for Medium-range Weather Forecasts (ECMWF), UK
- 17Chinese Academy of Meteorological Sciences, China Meteorological Administration, China
- 18School of Atmospheric Sciences, Nanjing University, China
- 19Nanjing University of Information Science Technology, Nanjing 210044, China
- 20Pacific Northwest National Laboratory, Richland, WA 99352, USA
- 21San Diego State University, USA
- 22Tsinghua University, China
- 23Beijing Normal University, China
- 24National Institute for Space Research (INPE), Brazil
- 25National Center for Atmospheric Research (NCAR), USA
- 26National Center for Environmental Prediction (NCEP)/National Weather Service/National Oceanic and Atmospheric Administration (NOAA), USA
- 27Environment and Climate Change Canada, Canada
- 28University of Maryland, College Park, USA
- 29Euro-Mediterranean Centre on Climate Change Foundation (CMCC), Italy
- 30National Meteorological Information Center, China Meteorological Administration, China
- 31National Meteorology Center, China Meteorological Administration, China
- 32National Climate Center, China Meteorological Administration, China
- 33University of California, Santa Barbara, USA
- 34Atmospheric Sciences Research Center, University at Albany, State University of New York, Albany, NY, 12203
- 35State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics (LASG), Institute of Atmospheric Physics, Chinese Academy of Sciences, China
- 36Yonsei University, South Korea
- 37Sun Yat-Sen University, China
- 38Northwest Institute of Eco-environment and Resources, Chinese Academy of Sciences, China
- 39University of Connecticut, USA
- 40Tokyo Metropolitan University, Japan
- 1University of California – Los Angeles, CA 90095, USA
- 2Institute of Tibetan Plateau Research, Chinese Academy of Sciences, China
- 3CNRM, University of Toulouse, Météo-France, CNRS, Toulouse, France
- 4University of Arizona, Tucson, USA
- 5Earth System Science Interdisciplinary Center (ESSIC), University of Maryland, College Park, USA
- 6Japan Agency for Marine Earth Science and Technology (JAMSTEC), Japan
- 7Lawrence Livermore National Laboratory, Livermore, CA 94550, USA
- 8International GEWEX Project Office, George Mason University, USA
- 9Hans Ertel Centre for Weather Research, Germany
- 10Korea Institute of Atmospheric Prediction Systems, South Korea
- 11Institute of Atmospheric Physics, Chinese Academy of Sciences, China
- 12Meteorological Research Institute, Japan Meteorological Agency, Japan
- 13Hokkaido University, Japan
- 14Indian Institute of Tropical Meteorology, Ministry of Earth Sciences, India
- 15Bureau of Meteorology, Australia
- 16European Centre for Medium-range Weather Forecasts (ECMWF), UK
- 17Chinese Academy of Meteorological Sciences, China Meteorological Administration, China
- 18School of Atmospheric Sciences, Nanjing University, China
- 19Nanjing University of Information Science Technology, Nanjing 210044, China
- 20Pacific Northwest National Laboratory, Richland, WA 99352, USA
- 21San Diego State University, USA
- 22Tsinghua University, China
- 23Beijing Normal University, China
- 24National Institute for Space Research (INPE), Brazil
- 25National Center for Atmospheric Research (NCAR), USA
- 26National Center for Environmental Prediction (NCEP)/National Weather Service/National Oceanic and Atmospheric Administration (NOAA), USA
- 27Environment and Climate Change Canada, Canada
- 28University of Maryland, College Park, USA
- 29Euro-Mediterranean Centre on Climate Change Foundation (CMCC), Italy
- 30National Meteorological Information Center, China Meteorological Administration, China
- 31National Meteorology Center, China Meteorological Administration, China
- 32National Climate Center, China Meteorological Administration, China
- 33University of California, Santa Barbara, USA
- 34Atmospheric Sciences Research Center, University at Albany, State University of New York, Albany, NY, 12203
- 35State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics (LASG), Institute of Atmospheric Physics, Chinese Academy of Sciences, China
- 36Yonsei University, South Korea
- 37Sun Yat-Sen University, China
- 38Northwest Institute of Eco-environment and Resources, Chinese Academy of Sciences, China
- 39University of Connecticut, USA
- 40Tokyo Metropolitan University, Japan
Abstract. Sub-seasonal to seasonal (S2S) prediction, especially the prediction of extreme hydroclimate events such as droughts and floods, is not only scientifically challenging but has substantial societal impacts. Motivated by preliminary studies, the Global Energy and Water Exchanges (GEWEX)/Global Atmospheric System Study (GASS) has launched a new initiative called Impact of initialized Land Surface temperature and Snowpack on Sub-seasonal to Seasonal Prediction
(LS4P), as the first international grass-root effort to introduce spring land surface temperature (LST)/subsurface temperature (SUBT) anomalies over high mountain areas as a crucial factor that can lead to significant improvement in precipitation prediction through the remote effects of land/atmosphere interactions. LS4P focuses on process understanding and predictability, hence it is different from, and complements, other international projects that focus on the operational S2S prediction. More than forty groups worldwide have participated in this effort, including 21 Earth System Models, 9 regional climate models, and 7 data groups.
This paper overviews the history and objectives of LS4P, provides the first phase experimental protocol (LS4P-I) which focuses on the remote effect of the Tibetan Plateau, discusses the LST/SUBT initialization, and presents the preliminary results. Multi-model ensemble experiments and analyses of observational data have revealed that the hydroclimatic effect of the spring LST in the Tibetan Plateau is not limited to the Yangtze River basin but may have a significant large-scale impact on summer precipitation and its S2S prediction. LS4P models are unable to preserve the initialized LST anomalies in producing the observed anomalies largely for two main reasons: i) inadequacies in the land models arising from total soil depths which are too shallow and the use of simplified parameterizations which both tend to limit the soil memory; and ii) reanalysis data, that are used for initial conditions, have large discrepancies from the observed mean state and anomalies of LST over the Tibetan Plateau. Innovative approaches have been developed to largely overcome these problems.
- Preprint
(4170 KB) -
Supplement
(130 KB) - BibTeX
- EndNote
Yongkang Xue et al.
Status: final response (author comments only)
-
RC1: 'Review of Xue et al. "Impact of Initialized Land Surface Temperature and Snowpack on Subseasonal to Seasonal Prediction Project, Phase I (LS4P-I): Organization and Experimental design"', Rene Orth, 18 Feb 2021
Review of gmd-2020-329 by Xue et al.
"Impact of Initialized Land Surface Temperature and Snowpack on Subseasonal to Seasonal Prediction Project, Phase I (LS4P-I): Organization and Experimental design"This paper introduces the LS4P project including its motivation, goals, and instructions on the modelling experiments. The key idea is to study the effect of soil temperatures in mountaineous areas on subsequent precipitation in downstream areas through remote effects of land-atmosphere interactions. Models and observations hint that such effects could be existing, and could consequently be exploited for weather/climate forecasting. An accurate representation of these effects across models is challenging, and the project is aiming to improve this.
-------------------
Recommendation:
I think the paper requires major revisions.This paper comprehensively presents the LS4P project which brings together the land-climate modelling community. The key idea about exploiting remote effects of land-atmosphere interactions for weather predictability is promising, and fits well with other recent studies illustrating so-far largely overlooked effects of land surface status and fluxes in downstream areas. Also considering soil temperature in this context is innovative as it also reflects to some extent the moisture/energy state of the land surface, and quite some satellite and ground-based data are available which are partly insufficiently exploited, particularly in comparison with the more prominent soil moisture. Nevertheless, I also see some shortcomings in this paper which should to be addressed to make the paper suitable for publication in Geoscientific Model Development:
(1)
I have some doubts about the application of the mask as described in section 3.2.
While equation 1a is clear, I do not understand why equation 1b is needed;
in this context also lines 403-408 and Figure 2 are unclear to me.
More generally, I think that the models' memory is a dynamic feature which should
be addressed through adapting the modelled (soil temperature and moisture) *dynamics*
rather than their *states*.
This way, I feel if the tuning parameter "n" could exaggerate initial corrections
which could degrade the simulations in the early forecasting period.
Why not simply correcting the model/forecast biases through post-processing and without
interferring with the actual model simulations?
Finally, as the mask correction requires observed soil temperature information,
maybe I missed that but I was wondering how this is done in places where this is not available?(2)
The description and motivation for selection of the ground truth data is insufficient
in my opinion. The text in lines 249-273 lists many datasets but does not indicate
how they are applied. A summary table of all employed datasets would be nice, including
their spatial and temporal extent, advantages, and variables provided.
I understand that you are using the CMA dataset as this is based on a relatively large
number of ground measurements. This makes sense, but it would be interesting how (well)
this dataset extrapolates between these measurements and how many (fewer?) measurements
are employed by other datasets such as the state-of-the-art ERA5 reanalysis.
Next to this, I was wondering why you are not employing available satellite-based
land surface temperature products http://data.globtemperature.info/ ?(3)
The manuscript is comprehensive but also quite long. To make it more concise it would be
helpful to shorten where possible I think. Below, I have indicated some examples where content
is repeated and where I would see potential to shorten the text.
Moreover, a summary table listing the main information regarding tasks 1-5 in section 3.2
would improve the readability.(4)
The authors refer a lot to snow effects in sections 1 and 2, and I like these ideas.
However, snow is not mentioned at all in sections 3, 4 and 5, and apparently
only implicitly (through LST) part of the analyses.
This should be clarified, and the role of snow in the project as described in
sections 1 and 2 should be toned down.(5)
As with the snow, also predictability is prominently mentioned in sections 1 and 2
and even in the abstract and title, but the detailed description of the project and the simulations
does not refer at all to this. So also here I would suggest to either include details
on how the predictability could/will be assessed, or tone it down in the beginning of the
manuscript.
I do not wish to remain anonymous - René Orth.
------------------Specific comments:
line 29 & 179: "Data groups" is not clear
line 35: Summer precipitation in which area?
line 49: "stubbornly low", not everywhere, there are quite some regional variations of precipitation forecast skill
lines 60-80: in this context you could cite Orth and Seneviratne (2017) where we compare the impacts of SST vs soil moisture
for land climate globallylines 83-85, and elsewhere: I am missing some justification why you chose to focus on (usually data sparse) mountain areas, and why their downstream impacts are expected to be higher than that of flat regions with possibly larger surface heat fluxes.
line 86: "very close", you mean strongly related here?
line 89 & Figure 1 caption: I do not fully understand how this was computed. Do you select years with warm/cold Mays, respectively, and then you track the temperature difference between these years through all months?
line 100: SSiB, abbreviation not explained
line 122: insert "the" before "snow darkening effect"
lines 127/128: could you please specify/explain "large diversity" a bit more?
line 146: insert "the initiative" after "historical development of"
lines 178/179: there is no need to put both written-out and numeric numbers there
lines 195/196: I think this is an important part of the project, can you give some more details on this data base?
lines 210/211: Sorry for mentioning an own study again, but Orth and Seneviratne (2017) can be instructive here I think
lines 221-225: repetition, could be removed
line 246: insert "the" before "Tibetian Plateau"
line 248: please be more specific which of these datasets are useful for the project, and why
lines 267-269: could be removed
lines 282-283: you mean anomaly precipitation rates of +1.32 mm/day?
line 294: "around late April", why is this not more specific?
line 316: "the models' performances are then checked" can be removed
line 321: what is a "proper" lapse rate?
lines 323-327: repetition, can be removed
lines 334-335: Not sure if the project is obsolete if models would do a good job, as you could still study LST/SUBT downstream effects, as stated in the project goals in e.g. the abstract
line 349: "around 2010", why is this not more specific?
line 426: what are the "sensitivity" and "control" runs?
line 450: here you could point to Table S1
lines 478-479: ERA-Intermin should be updated to ERA5 (https://climate.copernicus.eu/climate-reanalysis)
lines 473-475: I get the point that you would use the dataset that uses most ground station measurements as reference,
but I am wondering how many stations ERA5 is using over this area?lines 590-591: While it becomes more clear later, it would be good to already motivate here why you are comparing biases with anomalies.
line 594 & Figure 6: I do not understand why you do this multiplication with -1.
line 605: How is this correlation computed? Is it a spatial correlation? Over which domain?
line 615: Please specify the area over which the subsequent drought occurred.
line 637: Why these methods, and not a similar approach as for example in Koster et al. (2016)
lines 643-647: Why and how is the fore-restore method causing inaccurate soil memory?
line 666: Correct tense, 2020 is in the past now :)
line 669-670: "A possible ... will also be prepared" can be removed
Figure 5:
- the quality/resolution is very low, please improveReferences:
Orth, R. and S.I. Seneviratne
Variability of Soil Moisture and Sea Surface Temperatures Similarly Important for Warm-Season Land Climate in the Community Earth System Model
Clim. Dyn. 30(6), 2141–2162 (2017).Koster, R., et al.
Impacts of Local Soil Moisture Anomalies on the Atmospheric Circulation and on Remote Surface Meteorological Fields during Boreal Summer: A Comprehensive Analysis over North America
J. Climate 29(20), 7345-7364 (2016). -
RC2: 'Comment on gmd-2020-329', Anonymous Referee #2, 11 Mar 2021
This paper describes the international / multi-organization LS4P project and provides some initial analyses of data submissions. It is essentially an “introduce the community to the project” paper, with most of the scientific findings to be documented later, as the project progresses.
While the paper is well written, in my opinion it has enough issues to rate a review of “major revision”. On the one hand, I do applaud the experiment organizers for bringing together such a wide-ranging group of participants to address such an important problem. With this diverse a group and a corresponding collection of model outputs from such a diverse set of models, I don’t doubt that the project will bear useful scientific fruit, and a paper like this that introduces the project to the broader community is certainly of value. On the other hand, the paper’s write-up glosses over several critical aspects of seasonal and subseasonal prediction that call some of the project’s long-term strategies into question, at least in terms of how they’re currently described. A revised version should address these shortcomings through substantial qualification (not just a sentence or two here and there) or, better yet, through a substantial rethinking of the approaches to be applied.
1. The underlying assumption of the project appears to be that if a model does not produce an accurate temperature over the Tibetan Plateau, the fault lies with the land model (e.g., in how long the land model maintains an initial condition). The paper states this explicitly on line 465. The truth is, all models have biases in both air temperature and precipitation across the globe, and these biases could have any number of sources. A temperature bias over the Tibetan Plateau might have nothing to do with land model processes. It might instead result from deficiencies in the reproduction of the general circulation, for example, or from some problems with the radiation balance. Forcing the model to have a low temperature bias by imposing a stronger initial temperature anomaly (perhaps even an unrealistic anomaly, through eq. 1) may amount to “getting the right answer for the wrong reason”, which is not a good basis for a forecast experiment. It’s quite possible that forcing a correct temperature through such an initialization when the model wants to do something else for reasons unrelated to land processes might have unexpected negative consequences – especially if the model is artificially modified in one region and not in surrounding regions. Substantial discussion regarding this is needed.
2. The overall strategy seems to ignore the fact that forecast models generate their forecasts relative to their own climatologies. A model that is known to be biased warm may produce an anomalously cold 2003 over the Tibetan Plateau (compared to what it usually produces there), and that would be useful information even if the forecasted temperatures are still warmer than the average observed TP temperature. The point is that people know how to account for long-term model biases. They would properly consider a forecast model’s result to be “2003 will be colder than usual by 5 degrees” rather than “the temperature will be 20C”. The emphasis here on matching the observed temperature in absolute magnitude seems inappropriate to a discussion of forecast systems. See, e.g., the NMME forecast anomaly pages [https://www.cpc.ncep.noaa.gov/products/NMME/seasanom.shtml], which show forecasted anomalies relative to each model’s climatology.
3. Forecast systems also produce a range of forecasted values through the running of ensembles, and any one ensemble member could represent what happens in nature. The experimental analysis protocol, however, emphasizes the importance of having the *ensemble mean* match the observed anomaly. This is inappropriate. The key question is, do any of the ensemble members look like the observations? (And, in conjunction with point #2 above, the truly key question is, do any of the ensemble member anomalies *relative to the forecast model’s climatology* look like the observed anomaly?) A model cannot be considered wrong if one of its ensemble members looks like the observations. Insisting that an ensemble mean match a specific year’s temperature seems wrong.
4. The model results concerning May Tibetan Plateau temperature anomalies and June precipitation anomalies in east Asia is perhaps suggestive but far from indicative of a causal relationship. Even if the agreements in 6a/6c and 6b/6d do suggest that one pattern led to the other (it could very well be coincidental), I don’t see how the Tibetan Plateau in 6a/6c can be isolated as the source of the agreement in 6b/6d. Significant qualification of this figure’s implications is needed.Minor comments:
-- Just to clarify: Are the warm and cold years the same for each month shown? If not, it’s not clear what this figure says about the persistence of warm and cold anomalies (line 92).
-- lines 106-108. This sounds strange. Can the authors clarify the link between temperature and water amount? While I see that more water leads to a slower seasonal transition, it’s not obvious that at a single point in time, temperature tells you something about water amount.
-- I studied Figure 2 for a long time and still can’t make sense of it. Why, for a cold year initialization, does the initial condition for a model with a cold bias get set to climatology whereas that with a warm bias does not? Also, please clarify in the caption: are the biases discussed here errors for the particular year of simulation, or are they long term climatological biases? I’m guessing the former, since Task 2 would need to be done for the latter; in that case, though, the use of the term “bias” is confusing here. Bias should refer to a long-term climatological error (reflecting a model deficiency), not to the error at a specific time (which should reflect both bias and random error). Overall, Figure 2 is not helpful for explaining the approach. And again, based on my earlier comments, I’m not convinced the Tmask strategy is appropriate anyway.
-- Equation 1 appears to be a means to impose an artificially large temperature anomaly at the start of a simulation so that the anomaly is maintained realistically during the forecast. As far as I can see, there’s no physical basis for the equation; it’s fully empirical and could lead to initial temperatures that make little physical sense (e.g., colder than the model ever gets). More qualification is needed regarding how artificial this construct is. (And again, based on my earlier comments, it may not be appropriate to fix the temperature error in this way, since it may have a source other than the land model.)
-- line 211 (and elsewhere): Replace “SST” with “ocean state”, since SST is only a small part of what seasonal forecast systems rely on from the ocean. Arguably, subsurface ocean temperature distributions are more relevant.
-- Clarification regarding figure 7: is this the average of the 2003 anomalies relative to the different models’ climatologies, or is it the average (over all years) model T-2m and precipitation minus the average (over all years) observations? I’m guessing the latter, given the remarkable agreement with figure 6cd. The latter can truly be considered a bias, but the term bias was used differently elsewhere in the paper.
Yongkang Xue et al.
Data sets
LS4P-I evaluation datasets for the paper Organization and Experimental design (Version v1) Xue, Y. and Diallo, I. https://doi.org/10.5281/zenodo.4383284
Yongkang Xue et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 510 | 152 | 8 | 670 | 22 | 5 | 8 |
- HTML: 510
- PDF: 152
- XML: 8
- Total: 670
- Supplement: 22
- BibTeX: 5
- EndNote: 8
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Impact of initialized Land Surface temperature and Snowpack on Sub-seasonal to Seasonal Prediction(LS4P) and provides the first phase experimental protocol (LS4P-I). The LS4P introduces spring land surface temperature/subsurface temperature anomalies over high mountain areas as a crucial factor that can lead to significant improvement in summer precipitation prediction.