the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
The Impact of Improved Spatial and Temporal Resolution of Reanalysis Data on Lagrangian Studies of the Tropical Tropopause Layer
- 1Harvard University Department of Earth and Planetary Sciences
- 2Harvard University School of Engineering and Applied Sciences
- 1Harvard University Department of Earth and Planetary Sciences
- 2Harvard University School of Engineering and Applied Sciences
Abstract. Lagrangian trajectories are frequently used to trace air parcels from the troposphere to the stratosphere through the tropical tropopause layer (TTL), and the coldest temperatures of these trajectories have been used to reconstruct water vapor variability in the lower stratosphere, where water vapor’s radiative impact on Earth’s surface is strongest. As such, the ability of these trajectories to accurately capture temperatures encountered by parcels in the TTL is crucial to water vapor reconstructions and calculations of water vapor’s radiative forcing. A potential source of error for trajectory calculations is the resolution of the input data. Here, we explore how improving the temporal and spatial resolution of model input data impacts the temperatures measured by Lagrangian trajectories that cross the TTL during boreal winter using ERA5 reanalysis data. We do so by comparing the temperature distribution of trajectories computed with data downsampled in either space or time to those computed with ERA5's maximum resolution. We find that improvements in temporal resolution from 6 hour to 3 or 1 hour lower the cold point temperature distribution, with the mean cold point temperature decreasing from 185.9 K to 185.0 K or 184.5 K for trajectories run during boreal winters of 2010 to 2019, while improvements to vertical resolution from that of MERRA2 data (the GEOS5 model grid) to full ERA5 resolution also lower the distribution but are of secondary importance, and improvements in horizontal resolution from 1° x 1° to 0.5° x 0.5° or 0.25° x 0.25° have negligible impacts. We suggest that this is caused by excess vertical dispersion near the tropopause when temporal resolution is degraded, which allows trajectories to cross the TTL without passing through the coldest regions, and by undersampling of the four--dimensional temperature field when either temporal or vertical resolution is reduced.
- Preprint
(2178 KB) -
Supplement
(14753 KB) - BibTeX
- EndNote
Stephen Bourguet and Marianna Linz
Status: final response (author comments only)
-
RC1: 'Comment on acp-2022-159', Alison Ming, 27 Apr 2022
Bourguet and Linz analyse the effect of different spatial and temporal resolution of ERA5 data on the cold point temperature distribution obtained using a Lagrangian transport model. They find that using a temporal resolution of 1h versus 6h is the primary contributor to the differences in cold point distribution with vertical resolution being of secondary importance. The paper is a thorough analysis of the effect of resolution on the cold point temperature and offers some explanations as to how to biases in temperature arise by examining how the trajectories sample the cold point.
Overall, the paper is well written and I only have a few minor comments detailed below.
It is unclear when you are using the DJF 2010 to 2019 integrations. Am I correct in thinking that most of the paper uses the 2017 integrations but the 2010 to 2019 integrations are in the supplementary material only? One suggestion would be to label (say A to G) the 7 experiments in Table 1 and use those labels in the figure captions.
Page 2 L24 Worth explaining the water vapor tape recorder in a bit more detail as it will help a reader who is unfamilliar with this region of the atmosphere to understand the subsequent discussion.
Page 4 L107 ERA5 + ERA5.1
Page 6 L147 I am confused. I thought that you were running the trajectory model on hybrid pressure levels but here you are initializing the trajectories on isentropic surfaces. Are you interpolating from isentropic levels to initialise or is the model using isentropic levels?
Page 7 L167 Change 90 days to 3 months or vice versa for consistency
References: Please check your references carefully. There are various issues with the urls. E.g. Line 366 and many others
-
RC2: 'Comment on acp-2022-159', Anonymous Referee #2, 03 May 2022
This paper analyzes how the temporal and spatial resolution of model input data impact the cold point temperatures sampled by Lagrangian back trajectories. The trajectory simulations and comparisons done using ERA5 data are innovative and provide important information for the topics related to upper troposphere lower stratosphere transport. I suggest publication after the authors address my comments and make revisions accordingly.
#1. The authors investigated the impact of the resolution of the input meteorological data on the cold point temperatures (CPT) sampled by trajectories. The conclusions are largely about how the distributions of CPT differ among different trajectory runs. This is useful information, however, it has its limitation since it does not show whether increased resolution results in an improved simulation compared to the real atmosphere. For example, which input resolution results in a more accurate CPT prediction and thus better water vapor prediction compared to the ERA5 water vapor itself or to the observation? The Lagrangian cold point temperatures are important for water vapor simulation, as stated by the authors, but the authors provided limited information (texts or figures) on water vapor predicted from the CPT. These are the major weaknesses of this paper. I highly recommend such analyses be added and discussed.
#2. It is confusing that the authors claimed their trajectories as domain filling DJF trajectories, while they only released trajectories on five consecutive days in February. Technically, there are no trajectories initialized in January or in December.
Some specific questions related to this:
How long does it take for a parcel to reach the cold point level from 400 K? Do most parcels reach their cold point on nearby dates?
The trajectories are initialized from the end of February. Given a climatological heating rate of 0.2-0.4 K/day in the TTL in boreal winter (Fueglistaler et al., 2009), it may take a parcel 50-100 days to reach 380 K. Therefore, the CPT the parcels sampled may occur in December or January. Since the authors only released parcels from the end of February, the cold point sampling may be limited to a few nearby days in January or December (This is only a rough estimate). If so, the CPTs are not technically DJF CPTs.
#3. Figures 2-4:
Figure 2: There is no map or PDF for the 1-degree run in the left panel, though the texts mentioned it. Panel e shows a warm bias by the 1-deg run. The authors claim that the warm bias is due to horizontal temperature variability. However, panels a-c only explains the warm bias by the 0.5-deg run relative to the 0.25-deg run.
For the right panel in Figures 2-4, I suggest adding PDFs of predicted water vapor and comparison to ERA5 water vapor or observation (as discussed in the first comment).
The ticks in y axis for panel d in Figures 2-4 are not consistent.
# 4. How are the “collocated” CPT differences computed? This question can be applied to Figures 2-4.
1) Line 179: colocated-> collocated.
2) Does collocated mean at the exact same grid point or just nearby grid points? Is there any spatial interpolation for nearby locations when computing the difference?
3) Are the CPT differences computed for points collocated horizontally only or three-dimensional? For example, Lines 229-230. There are clearly differences in vertical distributions judging from Figs. 4f-g.
4) What is the percentage of trajectories that collocate their CPTs for each run? How about the CPT locations from different runs that do not collocate? How much do those non-collocate points contribute to the temperature PDF shift (This seems to be addressed only for the temporal resolution experiment)?
5) For the horizontal resolution experiment, the authors ruled out the under sampling of the wind in causing the warm bias. Such a conclusion can be drawn if the authors confirm that they use CPT locations obtained from the 0.5 deg run and compare the temperatures from the 0.5 deg data and 0.25 deg data interpolated onto these CPT locations. Such a comparison directly tells whether the warm bias is purely caused by temperature variability in different datasets. But it is not clear how the difference in Fig. 2c is computed. Neither is there such calculation for the 1-deg run, which shows the largest warm bias. Is there a line for the 1-deg run in Fig. 2d? The lack of information makes the authors’ explanations less convincing.
6) Lines 186-192: “horizontal temperature variability” here the wording is a bit ambiguous and not consistent with previous texts. Do the authors mean temperature variability between different resolution data?
Summary and final thoughts for comment # 4: The method of the spatial-temperature difference analyses is not consistent or unclear throughout the paper, which makes their explanations of the temperature PDF shifts less convincing.
It seems from the paper, the CPT differences can be attributed to a) horizontal or vertical locational differences, b) given the same 3-D location, the temperature variability among different resolution data, c) differences in wind sampling. It would be helpful for the authors to include quantitative conclusions (for each of their experiment sets) on how much each factor contributes to the CPT bias in percentage and K.
I also suggest consistency in the text when referring to temperature variability between different runs and other terms alike.
#5. Lines 290-291 vs Lines 295-296. The conclusions here stated seem to contradict each other. Lines 290-291: Do the authors mean far too large of a fraction of the lower resolution trajectories are traced to the stratosphere?
It would also be helpful to reference the figure number in these two paragraphs.
Minor comments:
Line 126: The full title for CDO was not given.
Lines 148-149: Is the vertical velocity w? Are the pressures and potential temperatures interpolated to the parcel’s location or Lagrangianly integrated?
Line 152: …relative humidity with respect to ice
Lines 154-156: Are new trajectories released daily for 20 days in January or just released on one day in January? Which 20 days?
Lines 310-316:
1) “warm bias … 0.5 K and 1.4 K”. vs Line 243: “…the mean cold point temperature for 1, 3, and 6 hour resolutions are 184.8, 185.2, and 186.2 K…” The numbers here do not match.
2) “the shifted temperature distribution for the 6 hour data results in a 26% increase in water vapor… from the cold tail).” This estimate is not discussed in the main text before the summary.
Lines 327-329: The fractions (0.58, 0.62, and 0.64) appeared in the summary but not in the main text. In section 2.3 the author stated that the trajectories for the dispersion experiment are released and tracked in January 2017. However, this paragraph is not consistent with those (“DJF 2010 to 2019”). Table 1 shows that the DJF 2010-2019 run is done for 1 hour resolution only.
Lines 337-338: The authors actually can evaluate the water vapor predicted by the CPTs against ERA5 water vapor or observation.
Comments on the summary section in general: The authors’ summary of their results and the significance/impact of their findings is a bit repetitive and dispersive. For example, both the first and last paragraph of the summary section mentions that increasing horizontal resolution beyond 1 deg does not bring significant improvement. Another example is that the impact due to temporal resolution is mentioned sporadically and repetitively throughout the section.
References
Fueglistaler, S., Legras, B., Beljaars, A., Morcrette, J.-J., Simmons, A., Tompkins, A. M., & Uppala, S. (2009). The diabatic heat budget of the upper troposphere and lower/mid stratosphere in ECMWF reanalyses. Quarterly Journal of the Royal Meteorological Society, 135(638), 21–37. https://doi.org/10.1002/qj.361
-
RC3: 'Comment on acp-2022-159', Anonymous Referee #3, 07 May 2022
This manuscript adopted the backward kinematic Lagrangian trajectory model to investigate the impact of spatial and temporal resolution of input meteorological fields on cold point temperature (CPT) simulations , therefore the impact on water vapor mixing ratio at entry into the stratosphere. The quantitative evaluation results provide an important value on the simulations of troposphere to stratosphere transport and moisture distribution. I suggest to consider publication after the authors appropriately address the comments.
General points:
- The manuscript assessed the simulation of CPT with trajectory model and mentioned in the text the corresponding water vapor changes (e.g., 26% in summary part). It is certainly worth showing the distribution of water vapor at cold point level and comparing with observations.
- The setup for the LAGRANTO runs are not clear.
- In sec 2.2, the authors mentioned that the trajectories were calculated backwards for 3 months from the end of February to the beginning of December. Is this time at the release points or when the CPT was found along the trajectory?
- Lines 131-135, the release of parcels are not clear to me. “At the end of each day for 5 consecutive days”. How many days in total per experiment from the end of February to the beginning of December? How to choose the 5 consecutive days? The figure S4 shows the PDF comparisons for two latitude ranges. How about the PDF of the cold point latitude?
- Iines 147-152, the trajectories tracked below 340 K were used for cold point analysis. What’s the percentage of the trajectories used in this analysis compared to total initialized? For “the fraction of trajectories traced below 340 K at each timestep”, is the black contours in figs 2-4 a&b related the fraction here? Or how was the fractions shown as black contours in figs 2-4 calculated?
- Figures 2-4
- In 2d, the PDF for 1.0 is missing; in 4d, the PDF for 3 hour is missing.
- In 3, given the cold point pressure and potential temperature show a bimodal distribution, I suggest to check the same PDFs separately in three main regions: west Pacific Ocean, South America and central Africa to discuss the possible reason for the bimodal structure.
- Lines 216-219, is this related to the limitation of interpolation in lower vertical resolution grid?
Minor comments:
- The saturation mixing ratio should add “relative to ice” .
- Line 221, it is 0.35 K in original reference.
- Consider add the calculated water vapor mixing ratio in text, beside the difference or percentage, such as lines 231 and 240.
- Lines 229-230 mentioned the difference first and later lines 241 gave the values. Consider put together or mention the mean values early.
- Line 248, are these values the calculated water vapor or the differences?
- Lines 249-255, is the horizontal coverage of the cold point sampled every 6 hours different from the 1 hour trajectories? Were the “edge” cold points still there?
- The summary part is not quite corresponding to the order of the results part. Such as summarized the variance in displacement first. Put discuss on temporal resolution ahead of vertical resolution and talk about temporal resolution again from line 327.
Stephen Bourguet and Marianna Linz
Data sets
LAGRANTO resolution project data and code Stephen Bourguet https://doi.org/10.5281/zenodo.6410194
Stephen Bourguet and Marianna Linz
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 281 | 59 | 8 | 348 | 24 | 1 | 1 |
- HTML: 281
- PDF: 59
- XML: 8
- Total: 348
- Supplement: 24
- BibTeX: 1
- EndNote: 1
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1