the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 13 Feb 2025
GIRAFE v1: a global climate data record for precipitation accompanied by a daily sampling uncertainty
Abstract. Here we introduce the first version of the Global Interpolated RAinfall Estimation (GIRAFE v1), the first dedicated global climate data record for precipitation by the Satellite Application Facility on Climate Monitoring (CM SAF) by the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT). GIRAFE is based on precipitation rate estimates obtained from observations by a variety of passive microwave radiometers onboard low-Earth orbit satellites and related retrieval algorithms and frequent and highly resolved infrared observations by geostationary satellites covering all longitudes (Geo-Ring). GIRAFE v1 is available globally at 1-degree resolution as daily accumulations and monthly means for the years 2002–2022, with an implementation of continuous production planned for 2025 onwards. The daily product is accompanied by a dedicated sampling uncertainty estimate based on decorrelation scales in space and time in infrared-based instantaneous precipitation fields. The methods for the generation of GIRAFE v1 are described in detail, followed by results of dedicated quality assessment and intercomparison activities. GIRAFE v1 reproduces reference datasets with a performance similar to established precipitation products, especially as those that are – like GIRAFE v1 – not adjusted to ground-based observations. Likewise, GIRAFE v1 proves to be suitable for the analysis of regional precipitation extremes, e.g. in their relation to sea surface temperatures. The main objective in the production of GIRAFE v1 is climate applications, for which we find the dataset highly suitable according to the dedicated stability and homogeneity analysis. The GIRAFE v1 data record is hosted by CM SAF and is freely available at https://doi.org/10.5676/EUM_SAF_CM/GIRAFE/V001 (Niedorf et al., 2024).
- Preprint
(2679 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 22 Apr 2025)
-
RC1: 'Comment on essd-2024-568', Anonymous Referee #1, 06 Mar 2025
reply
GENERAL COMMENTS
This manuscript reports the creation and analysis of a global rainfall climate data record. By combining observations from LEO PMW and GEO IR, GIRAFE provides the precipitation rate at 1° daily and monthly globally from 2002 to 2022, leveraging three different algorithms (HOAPS, PNPR-CLIM, and PRPS). Uniquely, it includes sampling uncertainties based on decorrelation scales in the IR observations.This is a generally well-written manuscript introducing a potentially useful dataset for the scientific community. The emphasis on climate-scale consistency is a key advantage of the product and the availability of uncertainty estimates is a unique feature that addresses a longstanding challenge. That being said, I find that the lack of analysis on the uncertainties is an aspect of the manuscript that could be improved. As well, the choice to use PMW observations to determine the precipitation rate and IR observations to calculate the precipitation fraction is somewhat unusual and can benefit from some justification. As such, I recommend Major Revision.
SPECIFIC COMMENTS
L85-86: Is there a reason for setting the cutoff at 55°N/S? Geostationary coverage does go further poleward; e.g., the NOAA CPC merged 4-km IR has a 60°N/S coverage while GridSat B1 reaches up to 70°N/S. Also, is the "Geo-Ring" mentioned herein related to the GEO-Ring effort led by EUMETSAT (https://www.eumetsat.int/geo-ring-and-isccp-ng-workshop)? If so, please mention; if not, I suggest renaming the term here to avoid confusion.L196: What is the MEI threshold for strong events?
L219-224: This framework of using PMW observations for R and IR observations for F is somewhat counterintuitive compared to the usual method of averaging the precipitation observations from each source. Can the authors discuss their rationale for this framework? This also eases the reader into the following paragraphs, which took me a while to work through because of its unusual approach.
L313: Under what circumstance would the exponential fit fail?
L321-322: Using an absolute number of days as an indicator can be problematic since the number of days in each month is not constant; e.g., 10 days means 35.7% of the time in a typical February but 32.2% of the time in January. Why not just provide the number of missing days itself?
L477-479: But if the issue is with GPCC, would this not also affect CMORPH and ERA5 (which are independent of GPCC) at the same point as well? And GPCP, which currently has a breakpoint there too, should not have it (since it uses GPCC as an input).
Figure 10: I am somewhat surprised that the 99.9th percentile is so low for the Maritime Continent, which appears to be inconsistent with, e.g., the Rx1day figure in Alexander et al. (2020). (But I must admit that it is hard to tell because of the minuscule figure sizes in that paper.) The authors should explain if this is expected behavior.
Figure 11: While I understand that the main focus is in the middle part of the plot, the left part of the plot (below 300.25 K) caught my eye, since the precipitation-SST relationship for GIRAFE is appreciably different from the other products there. Can the authors explain what may be going on here?
Section 5.6: Given the prominence of uncertainties in this product, I must admit some disappointment in the lack of results on the uncertainties. I was hoping to see a demonstration that the reported uncertainties are indeed representative, for example by performing some regional evaluation and investigating whether the estimates with higher uncertainties have larger errors compared to a reference product. Lacking that, I would like to see some analysis and example use of the uncertainties. If manuscript length is a concern, Sec. 5.4 and 5.5 could be shortened or even removed.
TECHNICAL CORRECTIONS
L74-75: I do not see the dashed line in the figure. Also, there is a typo in the figure in the description of "Daily Products".L88: There is a stray sentence on the loose here.
L91-92: What do the numbers on the left indicate? The range of longitude of the satellite footprint?
L440-441: The gray lines are practically useless at this point; I can barely see its variation. I suggest removing them, which would allow a reduced y-axis range.
Figure 9: In Figure 9c, there is a grayish dotted line in March 2018, which does not appear to correspond to any product in the legend. Please clarify what that is. In general, this diagram is hard to read because of the amount of information and the size of the lines. However, I do not have any ideas what could be done, so I ask the authors to consider if they can improve its legibility.
L486: The authors should define stability, or at least explain what it means in qualitatively. Otherwise, I do not know how to interpret the numbers in Table 6.
Citation: https://doi.org/10.5194/essd-2024-568-RC1
Data sets
GIRAFE v1: CM SAF Global Interpolated RAinFall Estimation version 1 A. Niedorf, S. Finkensieper, H. Konrad, R. Roca, M. Schröder, S. Cloché, G. Panegrossi, P. Sanò, C. Kidd, R. A. Jucá Oliveira, K. Fennig, T. Sikorski, M. Penning de Vries, A. Radovan, F. Dietzsch, M. Pondrom, N. Selbach, and R. Hollmann https://doi.org/10.5676/EUM_SAF_CM/GIRAFE/V001
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 206 | 76 | 6 | 288 | 6 | 7 |
- HTML: 206
- PDF: 76
- XML: 6
- Total: 288
- BibTeX: 6
- EndNote: 7
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1