the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 15 Sep 2025
SPREADS: From Research to Operational Open-Source Data Assimilation System
Abstract. The Scalable PaRallelised EArth Data Assimilation System (SPREADS) is a next-generation data assimilation system developed at CMCC-Foundation (Euro-Mediterranean Center on Climate Change) to support operational global forecasts. Built upon the Data Assimilation Research Testbed (DART), SPREADS incorporates key advancements such as First Guess at Appropriate Time (FGAT), enhanced observation handling via D4O (Database for Observations), and high-performance parallelisation to significantly improve computational efficiency and scalability. A major focus of SPREADS is the assimilation of a vastly increased number of asynchronous satellite-based radiances, which have been shown to substantially enhance analysis quality. Designed for coupled atmosphere-land-ocean-ice assimilation, SPREADS forms the core of the CMCC Earth SYstem Modelling and Data Assimilation (ESYDA division) operational forecast system. This paper presents the modifications made to DART, evaluates preliminary results, and outlines future developments toward fully coupled data assimilation.
- Preprint
(2286 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2025-4294', Anonymous Referee #1, 17 Nov 2025
-
RC2: 'Comment on egusphere-2025-4294', Anonymous Referee #2, 25 Nov 2025
The manuscript describes the development of SPREADS which is a new open-source, ensemble-based data assimilation system derived from DART. Its main goal is to extend DART toward operational global NWP use by introducing FGAT capability, modularized assimilation steps, the d4o SQL-based observation database, and several code-level parallelization improvements. The authors present diagnostics for multiple observation types, discuss the impact of bias-correction schemes, and compare SPREADS to ERA5. The manuscript positions SPREADS as a scalable and transparent platform for coupled DA.
My main concerns relate to the writing quality and interpretation of results. The manuscript often reads as a sequence of statistics rather than a cohesive scientific narrative, and several sections would benefit from substantial reorganization to improve flow (particularly Section 3). Many figures are too low-resolution to interpret, which limits readability. More importantly, several evaluation comparisons are difficult to interpret or potentially misleading. The manuscript frequently compares diagnostic changes across dates within the same experiment without controlling for changes in the observing system or other evolving conditions (e.g., different dynamic setups, seasons, etc.). Likewise, comparisons with ERA5 would be more meaningful if a control or reference experiment were included to quantify SPREADS’s incremental value. Overall, the work is promising, but the scientific communication and evaluation methodology need stronger justification and more careful framing.
Please see the attached document for my specific comments.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,006 | 73 | 16 | 1,095 | 18 | 16 |
- HTML: 1,006
- PDF: 73
- XML: 16
- Total: 1,095
- BibTeX: 18
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Carla Cardinali
Giovanni Conti
Marcelo Guatura
Sami Saarinen
Luis Gustavo Gonçalves De Gonçalves
Jeffrey Anderson
Kevin Raeder
Scientists have developed research systems to test new ideas in data assimilation, but these often lack the efficiency and robustness needed for operational use. We addressed this gap with key innovations: a flexible observation database, first guess at the appropriate time, and modular, parallelised software enabling the assimilation of millions of observations.
Scientists have developed research systems to test new ideas in data assimilation, but these...
The authors introduce their global data assimilation system and present the results of a preliminary experiment. They assimilated satellite radiances from AMSU-A using a two-step bias correction (scan-angle and air-mass bias corrections). However, a considerable bias in brightness temperature (approximately −0.2 to −0.3 K) remains even after applying the air-mass bias correction (top-left panel of Figure 4). The prior bias should ideally be close to zero; otherwise, the analyses may be degraded. I believe this issue should be resolved before publication.
Major comments:
Minor comments:
Line 197-199: A diagram might help readers easily understand the time slots.
Line 504 “At the initial time”: The experimental period starts from July 2017 (line 337). Why is December 1st 2017 the initial time?
Figure 10: To whiten the absolute values less than 1.56 seems optimistic.
It would be great if the authors could increase the resolution of some figures (e.g., Figure 4).