the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 12 Feb 2026
Arctic Weather Satellite Assessment and Assimilation at ECMWF
Abstract. The Arctic Weather Satellite (AWS) is a ground-breaking small satellite from ESA. Its goal is to measure microwave sounding radiances of sufficient quality for improving weather forecasts from a rapidly developed, low-cost mission. AWS is a pathfinder for the proposed EUMETSAT Polar System (EPS) Sterna constellation, which would represent a paradigm shift for operational satellite meteorology. The payload of AWS is a newly developed passive microwave (MW) sounder, with traditional temperature and humidity sounding channels near 54 and 183 GHz, plus novel humidity-sounding channels near 325 GHz. In this paper, first the radiometric performance of AWS is evaluated in reference to the ECMWF data assimilation system and heritage sounders, and then assimilation trials are presented to gauge the impact of AWS on forecast performance. The assimilation of AWS follows the all-sky method as applied to other MW radiometers in the ECMWF system, with the notable addition of the first-ever sub-millimetre wavelengths from the 325 GHz channel suite. Channel biases and noise estimates are generally in line with those of heritage instruments; AWS performance is similar to that of equivalent channels of AMSU-A and MWHS-2 in the 50 and 183 GHz bands, respectively, but effective noise for temperature sounding is higher than that of ATMS after spatial averaging. Nine months of experimentation show that adding AWS to the assimilation improves short-range forecasts of humidity, winds, and temperature. Geopotential height and winds are improved in the Southern Hemisphere through day 4. Despite its small size, AWS is a high-performing radiometer with data quality sufficient for operational assimilation in NWP. It has been assimilated operationally at ECMWF since July 2025.
- Preprint
(5398 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-712', Anonymous Referee #1, 06 Mar 2026
- RC2: 'Comment on egusphere-2026-712', Hu Yang, 08 Mar 2026
-
RC3: 'Comment on egusphere-2026-712', Anonymous Referee #3, 16 Mar 2026
General comments:
This paper is a very enjoyable and informative read. The structure is clear and logical and the scientific approach is rigorous. The authors provide a well-documented analysis of the assessment of data quality, preprocessing methods and the assimilation of AWS brigthness temperature data within the ECMWF NWP model. The figures are clear, relevant and well-designed. They contribute to the understanding of the methodologies and results. This study will be a good resource for researchers on the assimilation of AWS data. I have a few minor suggestions which could improve precision in certain sections.
Specific comments:
Table 1:
For humidity-channels, AWS channels 10-15 are presented to be equivalent to 17-22 ATMS channels and 10-15 MWHS-2 channels but their definitions are a little different. For example, the channel 11 of AWS has a frequency of 176.311 GHz whereas the “equivalent” channels of ATMS and MWHS-2 frequency is defined as 183.31 ± 7 GHz. Can considering only one side of the absorption band influence the comparisons of the data of these instruments?
Figure 3:
For feedhorns 3 and 4, there is a peak in the number of observations in the zero-bin of SI or CI (particularly on land). It could be interesting to have some explanations about what these observations represent and why they do not appear in the feedhorn 1.
Figure 4:
On cal/val sample of the channel 4, most observations are screened out over land but some remain, particularly at the edges of the swath. It could be interesting to investigate or explain this behaviour.
Figure 6:
To better assess the Gaussianity of the cal/val samples, overlaying a Gaussian curve (with the mean and standard deviation of the samples) on the histogram would provide a visual confirmation of the distribution’s fit.
l.250:
When discussing the across-scan biases for the feedhorn 1, a reference to Figure 4 would help to illustrate the negative bias observed in channel 4 at the first scan positions.
Technical corrections:
l. 138: The abbreviation “TB” is not defined yet, it could be clearer to write “brightness temperature”.
Citation: https://doi.org/10.5194/egusphere-2026-712-RC3
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 297 | 129 | 20 | 446 | 36 | 41 |
- HTML: 297
- PDF: 129
- XML: 20
- Total: 446
- BibTeX: 36
- EndNote: 41
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments:
It is a pleasure to read this article. The presentation is logically structured and the text reads easily. It is useful to developers at other NWP centres to learn how ECMWF has adapted their all-sky radiance assimilation framework to deal with the new technology of the AWS satellite and its microwave radiometer. I appreciate the quick timeline from the satellite launch to operational assimilation and further to comprehensive documentation of the setup and impact in scientific literature. The authors provide sufficient, but not excessive, detail on almost all aspects. I have only a couple of minor suggestions how the article could potentially be improved.
Specific comments:
I wish the authors could unify the way they refer to the primary temperature-sounding spectral region in the microwave. Most often in the text, they say either “50 GHz band” or “50 GHz channels”, so I would suggest a slight rephrasing on lines 5 and 47 to be consistent throughout the text.
Lines 32-33: I believe some more accuracy is needed to define the CGMS baseline orbits. I will leave it up to the authors to decide, but the text should made it explicitly clear that we are talking about sun-synchronous orbits and that 1:30, 5:30, and 9:30 refer to local times of overpasses at the equator.
Given that the 183 GHz channel definitions in the AWS spectrum are not exactly identical to those in ATMS and MWHS-2 spectra, the concept of “equivalent channels” sounds a bit vague. It would be useful to discuss in more detail how the authors have determined which ATMS / MWHS-2 channel is equivalent to which AWS channel.
Line 138: I suppose “TB space” here refers to brightness temperature observations? More clarity is needed here.
Technical corrections:
On lines 145-146, it would be useful to include a reference to middle panels of Fig. 3 .
There is a reference to Table 3 on line 187, which comes before the first reference to Table 2 (line 197). I would suggest removing the reference on line 187.
There is unnecessary repetition on lines 356-357. I suggest removing “spanning the first three months of AWS active use in the operational system” (essentially the same is told on lines 352-353).