Nudging allows direct evaluation of coupled climate models with in-situ observations: A case study from the MOSAiC expedition
- 1Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research (AWI), Bremerhaven/Potsdam, Germany
- 2Cooperative Institute for Research in Environmental Sciences, University of Colorado Boulder, Boulder, Colorado
- 3National Oceanic and Atmospheric Administration Physical Science Laboratoriy, Boulder, Colorado
- 4Jacobs University Bremen, Bremen, Germany
- 5Department of Meteorology and Bolin Centre for Climate Research, Stockholm University, Stockholm, Sweden
- 6Institute of Environmental Physics, University of Bremen, Bremen, Germany
- 1Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research (AWI), Bremerhaven/Potsdam, Germany
- 2Cooperative Institute for Research in Environmental Sciences, University of Colorado Boulder, Boulder, Colorado
- 3National Oceanic and Atmospheric Administration Physical Science Laboratoriy, Boulder, Colorado
- 4Jacobs University Bremen, Bremen, Germany
- 5Department of Meteorology and Bolin Centre for Climate Research, Stockholm University, Stockholm, Sweden
- 6Institute of Environmental Physics, University of Bremen, Bremen, Germany
Abstract. Comparing the output of general circulation models to observations is essential for assessing and improving the quality of models. While numerical weather prediction models are routinely assessed against a large array of observations, comparing climate models and observations usually requires long time series to build robust statistics. Here, we show that by nudging the large-scale atmospheric circulation in coupled climate models, model output can be compared to local observations for individual days. We illustrate this for three climate models during a period in April 2020 when a warm air intrusion reached the MOSAiC expedition in the central Arctic. Radiosondes, cloud remote sensing and surface flux observations from the MOSAiC expedition serve as reference observations. The climate models AWI-CM1/ECHAM and AWI-CM3/IFS miss the diurnal cycle of surface temperature in spring, likely because both models assume the snow pack on ice to have a uniform temperature. CAM6, a model that uses three layers to represent snow temperature, represents the diurnal cycle more realistically. During a cold and dry period with pervasive thin mixed-phase clouds, AWI-CM1/ECHAM only produces partial cloud cover and overestimates downwelling shortwave radiation at the surface. AWI-CM3/IFS produces a closed cloud cover but misses cloud liquid water. Our results show that nudging the large-scale circulation to the observed state allows a meaningful comparison of climate model output even to short-term observational campaigns. We suggest that nudging can simplify and accelerate the pathway from observations to climate model improvements and substantially extends the range of observations suitable for model evaluation.
Felix Pithan et al.
Status: open (until 09 Feb 2023)
-
RC1: 'Review of egusphere-2022-706', Anonymous Referee #1, 03 Oct 2022
reply
Review of "Nudging allows direct evaluation of coupled climate models with in-situ observations: A case study from the MOSAiC expedition" by Pithan et al
This manuscript describes evaluation of two coupled climate models and one general circulation model against detailed in-situ and remote sensing observations in the Arctic. The manuscript is generally well written, and the topic is appropriate and interesting for EGUsphere. The methods for nudging the coupled system are interesting and will be useful for others. I think the manuscript should be publishable subject to minor revisions, detailed in specific comments below.
Generally, I think some parts of the text could be more specific in the treatment of the different models. Also, some of the figure presentations could be improved or clarified.
Page 1, L12: what does CAM6 show?
Page 2, L42: Add Gettelman et al 2020, since they also used CAM6 in the S. Ocean.
Page 3, L88: If CAM6 is nudged, you also need to cite Gettelman et al 2020 who nudged CAM6 and looked at the S. Ocean.
Gettelman, A., C. G. Bardeen, C. S. McCluskey, E. Järvinen, J. Stith, C. Bretherton, G. McFarquhar, C. Twohy, J. D’Alessandro, and W. Wu. “Simulating Observations of Southern Ocean Clouds and Implications for Climate.” Journal of Geophysical Research: Atmospheres 125, no. 21 (2020): e2020JD032619. https://doi.org/10.1029/2020JD032619.
Page 5, L107: CAM6 cloud fraction?Page 5, L126: This is confusing. Did you use 1 hour or 24 hour for CM3? Also, the all wavenumber case for CM3: was that 1 hour or 24hour?
Page 6, L137: what is the nudging timescale in CAM6?
Page 9, Figure 3: Describe model lines and gray line in 3d in caption.
Page 10, L216: CAM6
Page 10, L218: is the CAM6 albedo shown in Fig 3e? I cannot see it.
Page 10, L230: Can you put then an estimate of observed cloud cover on Figure 3f? Or is it 100% since the radar has cloud condensate all the time?
Page 10, L231: would the models detect such a cloud?
Page 11, L233: is there condensate in CM1?
Page 11, Fig 6: why no CM1 and CAM6?
Page 12, L249: What about CAM6? If it gets the radiation right, does it get the condensate right too? Please be consistent in the treatments here (e.g. CAM6 is on Figure 7 but not Figures 5 & 6, why?)
Page 12, Figure 7: its hard to distinguish the gray dotted lines from the model dotted lines. Perhaps use a shaded region for the standard deviation.
Page 12, L255: why are they not equal? Change of temperature of the ice itself?
Page 13, L275: shouldn’t the relationship be tighter then in figure 8 if it represents a coefficient in the models?
Page 14, L286: Dark green triangles.
Page 14, L307: Why not include CAM6?
Page 15, L311: These models are nudged all the way to the surface. How much does that matter? What does CAM6 (not nudged below 690hPa) do? Does the spectral nudging matter?
Page 15, L318: can you refer to Fig 6 and put a line for the time of the sounding to guide the reader here?
Page 16, L358: what does CAM6 do for liquid water? Gettelman et al 2020 showed overestimations of supercooled liquid in the S. Ocean.
Page 16, L361: Can you say anything about the type of nudging? E.g. spectral v. Full nudging, and vorticity/divergence v. Winds & Temps. Does it matter? What is better? How can you tell?
-
RC2: 'Comment on egusphere-2022-706', Anonymous Referee #2, 20 Jan 2023
reply
Peer review for manuscript: Nudging allows direct evaluation of coupled climate models with in-situ observations: A case study from the MOSAiC expedition
Climate models are important tools for understanding complex interactions of physics and dynamics in the atmosphere. Descriptions of dynamical and especially physical processes are not complete in the models and simplifications made in parametrizations of physical processes causes uncertainty in model results. Therefore, it is very important to evaluate accuracy of models.
Typically accuracy of climate models is estimated by comparing model results against long time series of observations (or data sets which are strongly constrained by observations as reanalyses). As the state of large circulation, which causes a large part of variability of atmospheric conditions in middle and high latitudes, in models are not connected to state of large circulation in the real atmosphere, long time series is needed for comparisons, so that they are able to present a sufficient part of climatological variability which is caused by variations in large scale circulation. The requirement of long time series rules out a lot of observational data which otherwise could be used for evaluation of climate models. Especially in high latitudes, where the permanent observational network is sparse and a lot of observations have been collected from relatively short measurement campaigns. This strongly limits the evaluation of accuracy on climate models in the high latitudes where the presentation of atmospheric conditions in climate models is often worse than in lower latitudes. Therefore, it is highly important to evaluate the capability of climate models to simulate polar climate, which further allows development of models, so that they can better simulate atmospheric conditions also in the high latitudes. The manuscript presents a method how to overcome the requirement of long time series using nudged model simulations. When nudged simulations are utilised, large scale circulation in the model is strongly constrained by the observed large scale circulation. This kind of methodology allows direct comparison of model results and observations. As the large circulation in nudged simulation is constrained by the observed large scale circulation, model physics are responsible for uncertainty of models. This method does not allow estimate biases which are associated with presentation of large scale circulation in the models, but often a large part of the uncertainty is associated with parameterized physical processes. However, the possibility to use short time series of observation provides large advantages for model evaluations and further model development.
The novelty of the manuscript is in its methodology. The set of models that is evaluated in the manuscript is not comprehensive by any means, but I think that is not the scope of the manuscript. However, even this set of models shows interesting differences in their capability to simulate atmospheric conditions and clearly demonstrate the usefulness of nudged simulation for model evaluation and also shows some deficiencies in models especially associated with treatment of snow surface. Therefore, the scientific value of this manuscript supports the publishing of the manuscript in Geoscientific Model Development.
General comments:
1) Nudging is probably familiar for many readers, but I still suggest adding a short general description of nudging in the beginning of the nudging paragraph in the method section. Does nudging cause artificial effects on time series when model state is nudged towards real atmospheric state?
2) As model biases are related to weather conditions, I would start the result section with short description of weather conditions where, in addition to temperature changes, you could shortly describe e.g. cloud conditions, cloud cover and cloud liquid and ice water content, stratification, longwave and shortwave radiative fluxes, turbulent heat fluxes and how they are related.
3) The differences in surface temperature and skin temperature between the models are well explained in the method section, but it is still sometimes tricky to follow which is meant by surface/skin temperature in some parts in the results section. Therefore, I would suggest paying attention to clarity when surface/skin temperature is discussed in the result section and add remainder about their meanings when it is not very clear which temperature is in question.
4) Longwave radiation often has a remarkable effect on surface and near surface temperatures. However it has not received a lot of attention in the manuscript. If you have observational data of longwave radiative fluxes, I would suggest adding comparison of longwave radiative surface fluxes between models and observations, and how differences in cloud cover and cloud water (liquid and ice) content affect longwave radiative fluxes as well as how differnces in longwave radiative fluxes affect surface temperatures.
Specific comments:
Lines 100 – 101, Are the different values for rhcrit and rhsat used for the whole column below 700hPa or only in the inversion layer if a temperature inversion exists below 700 hPa?
Lines 141 – 144, How representative observations are for the whole grid cells used in comparisons? Were the conditions in the area of grid cell homogeneous enough (e.g. occurrence open water causes inhomogeneity) that point measurement could represent average condition in the grid cell?
Lines 188 – 189, How coarser resolution affects the delay?
Line 195 Do you compare surface temperatures from models against observed skin temperature? The next paragraph maybe gives an answer, but maybe it is good to clarify also here.
Line 270, Which kind of weather conditions are associated with unstable stratification in AWI-CM models?
Lines 286 – 297, Have you calculated the relationship between sensible heat flux and difference between skin temperature and 2m temperature in AWI-CM3/IFS. How would it look? Maybe the relationship between sensible heat flux and difference between skin temperature and 2m temperature looks better than relationship between sensible heat flux and difference between surface temperature and 2m temperature because in the method section it has been mentioned that AWI-CM3/IFS uses skin temperature for turbulent heat fluxes.
Lines 330 – 331, Could stronger cooling in AWI-CM1/ECHAM be associated with clouds?
Lines 344 – 346, Has the overestimation of cloud ice content so large an impact on radiation that it can compensate the underestimation of cloud liquid water content?
Overall, the manuscript is well written conclusion based on evidence of results. Methods are appropriately described allowing readers to understand how study is done. Therefore, I recommend publishing the manuscript in Geoscientific Model Development after minor revision.
Felix Pithan et al.
Felix Pithan et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
469 | 133 | 9 | 611 | 5 | 2 |
- HTML: 469
- PDF: 133
- XML: 9
- Total: 611
- BibTeX: 5
- EndNote: 2
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1