The main goals and activities of the ERA-CLIM2 project are discussed along with some key results, including the first ensemble of coupled reanalysis of the twentieth century.
The European Reanalysis of Global Climate Observations 2 (ERA-CLIM2) is a 4-yr research project funded by the European Union Seventh Framework Program (EU FP7; Grant Agreement 607029; see appendix for a list of the Consortium’s institutions) that started on 1 January 2014 following the successful completion of its predecessor project ERA-CLIM. The research initiated in these two projects underpins a concerted effort in Europe to build the information infrastructure needed to support climate monitoring, climate research, and climate services, based on the best available science and observations.
ERA-CLIM2 is one of several collaborative research projects designated by the European Commission as precursors to the EU’s Copernicus Climate Change Service (C3S). Indeed, ERA-CLIM2 activities on data rescue, satellite data rescue and reprocessing, coupled data assimilation, and reanalysis production have had a strong impact on the design and implementation of the C3S. Furthermore, activities performed in ERA-CLIM2 aimed at improving the assimilation of observations at intermedium boundaries (i.e., sea surface temperature and sea ice concentration observations) will benefit ocean and sea ice reanalyses and operational marine forecast activities performed within the EU’s Copernicus Marine Environment Monitoring Service (CMEMS).
The main aim of the ERA-CLIM2 project was to enable production of state-of-the-art reanalyses of the coupled climate system for the twentieth century. This follows earlier work on coupled, long-term climate reanalysis conducted at the National Center for Atmospheric Research (NCAR) and the National Centers for Environmental Prediction (NCEP) (Saha et al. 2010; Compo et al. 2011) and at the European Centre for Medium-Range Weather Forecasts (ECMWF) (Dee et al. 2014).
Climate reanalyses are physically consistent datasets derived from observations that document the recent evolution of the global atmosphere, ocean, land surface, cryosphere, and the carbon cycle. Reanalysis data are generated by a sequential process called data assimilation, which combines first-guess estimates defined by short-range numerical model forecasts with vast amounts of data from a range of observing platforms (surface, upper air, and satellites). Climate reanalyses provide long (multidecadal) time series of gridded estimates for many different climate variables, which are used to study past weather events, estimate climatologies, monitor climate change, and supply crucial data on climate needed for science and applications.
The production of a reanalysis is a complex activity that requires large computational resources, access to observations from many providers, and expertise in multiple disciplines. During the past three decades, successive reanalyses of the global atmosphere have been produced by a few, key institutions: NCAR and NCEP (Kalnay et al. 1996; Saha et al. 2010), the Japan Meteorological Agency (JMA; Onogi et al. 2007; Kobayashi et al. 2015), and the National Aeronautics and Space Administration (NASA; Schubert et al. 1993; Rienecker et al. 2011; Gelaro et al. 2017), in addition to ECMWF. A global reanalysis extending back to the late nineteenth century was first produced by NOAA in collaboration with the Cooperative Institute for Research in Environmental Sciences (CIRES), using only surface pressure observations and prior estimates of sea surface temperature (SST) and sea ice distributions to avoid the effects that large changes in the observing system could have had on the reanalysis (Compo et al. 2011).
ERA-CLIM2 has contributed to advancing reanalysis science and the production of climate reanalyses in four main areas:
Observation data rescue and postprocessing: Activities under this theme included a large effort on data rescue for historic in situ weather observations around the world and substantial work on the reprocessing of satellite climate data records and enabling the use of historical satellite data for reanalysis;
Data assimilation methods: Activities under this theme aimed to progress the development and testing of “coupled assimilation methods,” capable of including observations from different Earth system components (land surface, ocean, sea ice, atmosphere, chemical components, etc.) to produce a more consistent estimate of the Earth system evolution, especially at the intermedium boundaries;
Reanalysis production: Activities under this theme aimed to generate the innovative reanalysis datasets, such as the first European coupled ocean–land–atmosphere reanalysis of the twentieth century, and to provide access to the reanalysis data;
Evaluation and uncertainty estimation: Activities under this theme focused on the assessment of reanalyses’ quality, along with how products differ from previous uncoupled products, and on the development of methods for estimating uncertainty in reanalyses.
REANALYSIS AS A TOOL TO MONITOR THE CLIMATE.
The Earth’s climate has traditionally been studied by statistical analysis of observations of particular weather elements such as pressure, temperature, wind, and rainfall. These meteorological observations are temporally and spatially incomplete and are often presented in terms of long-term averages to identify evidence of climate change. Reanalyses provide a more complete source of data to understand and monitor the climate. In a reanalysis, the weather observations collected in past decades are fed into a modern forecasting system designed to assimilate observations from existing and planned platforms (e.g., forthcoming satellite instruments), which provides a physically consistent description of the Earth system. By constantly correcting a model first guess (defined by short-range forecasts) toward the available observations, reanalyses combine the advantages of having a three-dimensional first guess and available data (further details in “Data assimilation methods for reanalysis” section). Therefore, reanalyses are physically consistent and spatially complete and encompass many variables for which observations are not always available. Irregular and intermittent observation sampling throughout the reanalysis period, especially during early periods, might, however, prevent the reanalysis dataset from showing temporal homogeneity at both global and regional scale (e.g., Ferguson and Villarini 2012; see also “Summary and conclusions” section).
Since its creation in 1975, ECMWF has been a key player in the production of reanalyses. The initial focus was on producing atmospheric reanalyses covering the modern observing period, from 1979. The first of these reanalyses, the First Global Atmospheric Research Program (GARP) Global Experiment (FGGE), was produced in the 1980s at the Geophysical Fluid Dynamics Laboratory (GFDL; Ploshay et al. 1992), followed by the 15-yr ECMWF Re-Analysis (ERA-15), the 40-yr ECMWF Re-Analysis (ERA-40) (Uppala et al. 2005), and the ECMWF interim reanalysis (ERA-Interim) (Dee et al. 2011). The next reanalysis in this series, the fifth major global reanalysis produced by ECMWF (ERA5), is now in production after many years of research and development.
Generating a reanalysis for climate monitoring is very challenging because it needs to be extended further back in time when the observing system was very sparse, especially before the availability of satellite data from the 1970s onward, and even more so before the arrival of radiosonde measurements in the 1930s. To tackle the unavoidable issue of the ever-changing observational network, the European Union started the ERA-CLIM project to investigate data selection for reanalyses covering the whole twentieth century. As part of the ERA-CLIM project, ECMWF produced the uncoupled atmospheric reanalysis ERA-20C, which covers the period January 1900–December 2010 (Poli et al. 2016). ERA-20C assimilated only conventional observations of surface pressure and marine wind, obtained from well-established climate data collections. ERA-20C delivered 3-hourly products describing the spatial and temporal evolution of the atmosphere, land surface, and waves.
As part of the FP7 ERA-CLIM2 project, the reanalysis capabilities developed in the ERA-CLIM project have been extended to the ocean and sea ice components. A new assimilation system, the Coupled ECMWF Reanalysis (CERA) system, has been developed to simultaneously ingest atmospheric and ocean observations in the coupled Earth system model used for ECMWF’s ensemble forecasts (Laloyaux et al. 2016, 2017). This approach accounts for interactions between the atmosphere and the ocean during the assimilation process and has the potential to generate a more balanced and consistent Earth system climate reconstruction (Fig. 1). CERA has also been found to reproduce better the observed negative SST–precipitation relationships on monthly time scales because it can resolve atmospheric feedbacks on SST, although in some areas the response in CERA is overdone compared with observations (Fig. 2). Efforts are being made to investigate whether this improvement will improve the prediction of precipitation (Feng and Haines 2017). One of the key deliverables of the ERA-CLIM2 project is CERA-20C, the first 10-member ensemble of coupled climate reanalyses of the twentieth century. It is based on the CERA system, which assimilates only surface pressure and marine wind observations as well as ocean temperature and salinity profiles. The data are openly available from the ECMWF platform (http://apps.ecmwf.int/datasets/).
Longitudinal Hovmöller diagrams at 1°N in the equatorial Pacific (90°W–180°) of high-pass-filtered wind stress (colors; N m–2) and SST (contours; interval: 0.25 K) for (a) ERA-20C and (b) CERA-20C over the period 1 Apr 1973–27 Mar 1974. CERA-20C is able to represent tropical instability waves (TIWs) because the ocean and the atmosphere are both responding in a consistent way. In ERA-20C, there are no TIWs and wind stress signals.
Citation: Bulletin of the American Meteorological Society 99, 5; 10.1175/BAMS-D-17-0199.1
SST–precipitation correlations for their monthly fluctuations in (a) observational data (HadISST2 and GPCP), (b) ERA-20C, and (c) CERA-20C (control run), over 1979–2010. Unshaded areas are where the correlations do not pass the significance test at the 90% confidence level. Note the agreement between observations and CERA-20C in the heavily precipitating regions.
Citation: Bulletin of the American Meteorological Society 99, 5; 10.1175/BAMS-D-17-0199.1
There is now a strong need for detailed information of CO2 fluxes and carbon pools from the climate modeling community who want to understand and quantify the carbon cycle at global and regional scales and from policy-makers and citizens who want to make well-informed decisions on CO2 emissions at regional and local scales. For this reason, ERA-CLIM2 is also producing associated global reanalyses of carbon fluxes and stocks using terrestrial biosphere and ocean biogeochemistry models, which are forced by CERA-20C.
A new version of the CERA system has also been developed based on a higher-resolution coupled model with the full observing system. This system has been used to produce the second coupled reanalysis planned by ERA-CLIM2, the Coupled European Reanalysis of the Satellite Era (CERA-SAT), which covers the period 2008–16. Plans are now being drawn to extend backward CERA-SAT to cover also the period 2000–08 and 2016 to the present and to keep it running to provide coupled analyses during the Year of Polar Prediction (2017–19).
The availability of atmospheric, oceanic, and coupled reanalyses allows for new advanced coupled diagnostics of the global energy cycle. For example, Mayer et al. (2017) improved the classic method of evaluating the vertically integrated atmospheric energy budget such that it is independent of reference temperature and consistent with diagnosed ocean heat budgets. It is also worth mentioning Pietschnig et al. (2017, manuscript submitted to J. Geophys. Res. Oceans), who demonstrated the value of comparing the net energy transport through all major Arctic Ocean gateways from ocean reanalyses with independent mass-consistent transport estimates from instrumented mooring observations.
OBSERVATION DATA RESCUE.
Reanalyses efforts strongly depend on observations of the atmosphere, land surface, ocean, and cryosphere. Observations are assimilated into a coupled general circulation model in order to produce the reanalysis in our project, but these are also used in several other steps. They are used to constrain the boundary conditions, to calibrate certain relations and to validate the final product. Particularly when going back in time, not all observations can be readily used in reanalysis production. A large fraction of historical meteorological observations has never been digitized because the data have thus far not been considered valuable. Even in the rather recent past, the availability of satellite products (and the computer code to read and process these data) is an issue that needs to be addressed.
Major efforts were undertaken in ERA-CLIM2 to collect and make available observations for reanalyses (Brönnimann et al. 2018). Such an undertaking requires a much broader vision than the production of a specific reanalysis, since these efforts should be seen as essential to providing future more realistic reanalyses. Availability of historical observations becomes a legacy, and producing reanalyses or other data products must be seen as a continuous effort.
Within ERA-CLIM2, millions of radiosonde, pilot balloon, and other ascending instruments’ profiles were digitized, which provides a three-dimensional view of the atmosphere back to the early twentieth century. Although the radiosonde data were not incorporated into CERA-20C produced in ERA-CLIM2, they were used for its validation.
These data could be used in the future to produce a new version of a reanalysis that includes historical upper-air data, following the preliminary results obtained by a pre-satellite-era reanalysis performed by Hersbach et al. (2017) for the period 1939–67. By the end of the project, all digitized upper-air data will be available in assimilation-ready format with bias adjustments for radiosonde temperatures (Haimberger et al. 2012) extending back to 1939. Any future reanalyses will be able to incorporate this vast amount of upper-air data and will thus build on the ERA-CLIM2 efforts.
Surface pressure and mean sea level pressure data were also digitized for several countries in both the Northern and Southern Hemispheres (NH and SH), some with sparse observation networks, and sent to the International Surface Pressure Databank (ISPD). These have been assimilated in new ISPD versions to be used in future reanalyses. Many other daily and subdaily land surface observations of temperature, relative humidity, surface wind, cloud cover, precipitation, evaporation, and sunshine duration were rescued. These have also been subjected to quality-control procedures and can be used for reanalysis comparison purposes (Brönnimann et al. 2018).
Snow is an important component of the climate system, at the interface of the land surface, vegetation, and atmosphere. It is highly relevant for various fields such as ecology, water resources, transport, and tourism. By digitizing large amounts of historical snow data and combining this with satellite products, ERA-CLIM2 generated snow products for various uses, including (but not limited to) reanalyses. Snow courses are specified paths of a few kilometers in length around a location; along this path, different snow properties are measured regularly. Thanks to ERA-CLIM2, about 1.2 million snow course observations have been collected from close to 400 stations compiled and made available.
Satellite data are the backbone of today’s reanalysis datasets, but they became available only in the mid-1960s and were built for the purpose of weather monitoring. Their use in reanalysis requires a quantification and correction of long-term effects due to systematic differences between satellite instruments of the same kind and changes in the characteristics of satellites and performance of sensors during their operational lifetime in space. A reprocessing of the data that applies the corrections is fundamental to serve the generation of physically consistent data records of geophysical variables by reanalysis.
Following the example of the work done by other groups (see references quoted in the first section), ERA-CLIM2 also devoted resources to reprocess satellite data of infrared and microwave radiances from geostationary imagers and microwave sounders, radio occultation bending angle profiles for several satellites, and atmospheric motion vectors from different instruments in geostationary and polar orbit. In addition, as part of satellite data rescue activities (Poli et al. 2017) radiative transfer calculations for some early satellite instruments were conducted, taking into account the characteristics of the instruments. These reprocessed data will allow a more comprehensive use of early satellite data, which should lead to improved future reanalyses.
DATA ASSIMILATION METHODS FOR REANALYSIS.
ERA-CLIM2 has been produced using a state-of-the-art version of the ECMWF data assimilation system capable of combining observations in the atmosphere and ocean with a coupled ocean–land–atmosphere model. Part of the work within ERA-CLIM2 was devoted to improve such a data-assimilation system, also testing new methods and ideas that could be used in future reanalysis productions.
A crucial part of the coupled climate system is the interface between the ocean/sea ice and atmosphere (Feng and Haines 2017). In the existing CERA system, the sea surface temperature is constrained to follow a global observational analysis, such as the Hadley Centre Sea Ice and Sea Surface Temperature, version 2, dataset (HadISST2) (Titchner and Rayner 2014) or the Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) (Donlon et al. 2012), which are calculated externally. Enabling the next CERA system to assimilate directly the wealth of high-quality satellite SST observations should allow the system to combine them with the temperature profile data and the coupled model background in a more consistent manner, thereby improving the accuracy of the reanalysis. Care has to be taken to deal properly with biases in the satellite data that could otherwise introduce spurious trends. In addition, the in situ ocean data are sparse, particularly in the early part of the twentieth century.
The ocean data assimilation system used in CERA is a state-of-the-art three-dimensional variational system called the Nucleus for European Modelling of the Ocean Variational assimilation system (NEMOVAR). It includes developments made to improve the assimilation of sea ice concentration during the satellite era. Sea ice concentration estimates have error characteristics that make it difficult to assimilate in most data assimilation algorithms so techniques have been tested that transform the sea ice concentration into a form that has Gaussian errors. It has been evolved into a more flexible framework for assimilating data into the NEMO ocean model, which includes changes (to the original NEMOVAR configuration) to allow the ocean data to be assimilated using sophisticated techniques similar to those used in the atmosphere (Weaver et al. 2016).
An example of how the NEMOVAR three-dimensional variational data assimilation (3D-VAR) system can effectively utilize sparse observations is illustrated in Fig. 3. Sparse observations can be assimilated using empirical orthogonal functions (EOFs), computed to model the large-scale error covariances. The EOFs can be used in conjunction with the existing local-scale error covariance model (e.g., Karspeck et al. 2012), thus allowing the spreading of information over large spatial scales when the data are sparse, while making good use of the higher-resolution measurements during the satellite era. In ERA-CLIM2, this has been investigated for the assimilation of SST measurements and the improvement has been plotted in the bottom panel of Fig. 3 versus the classic background statistics that does not contain EOF information.
Examples of (a),(c) sea surface temperature observations (°C) and (b),(d) temperature profile observations (°C) from (top) Jan 1953 and (middle) Jan 2010. (bottom) The percentage change in error in SST from assimilating the Jan 1953 data using an updated version of NEMOVAR.
Citation: Bulletin of the American Meteorological Society 99, 5; 10.1175/BAMS-D-17-0199.1
The new version of NEMOVAR can also be configured to use information from an ensemble of model runs to improve the assimilation, which is a more sophisticated system than the algorithm used in the production of ERA-CLIM2. In the present CERA system, the ocean bias correction scheme is not applied during the ocean data assimilation as it normally needs to be estimated from a priori run. However, now with CERA-20C being finished, it becomes practically possible to implement this scheme throughout the whole twentieth century. Although implementing the bias correction scheme avoids the ocean model producing a spurious reaction to adjust the imbalance between the ocean and atmosphere initial conditions, tests have to be performed to investigate if the implementation will benefit the atmosphere analysis. It is also worth mentioning that, also thanks to ERA-CLIM2 resources, a four-dimensional version of the ocean assimilation system has been developed and is under testing.
Currently, in the CERA system the coupled model is introduced at the outer-loop level [see Fig. 9 of Dee et al. (2014) for a schematic], by coupling the ECMWF’s Integrated Forecasting System (IFS) for the atmosphere, land, and waves, to the NEMO model for the ocean and to the two-level Louvain-la-Neuve Sea Ice Model (LIM2) for sea ice. This means that air–sea interactions are continuously taken into account when observation misfits are computed and when the increments are applied to the initial condition. This allows feedback between the ocean and atmosphere models (Laloyaux et al. 2016, 2017). Investigations into whether the atmosphere observations could be used to directly correct the ocean model, and vice versa, have been carried out in ERA-CLIM2 (Storto et al. 2018). This technique, known as “strongly coupled data assimilation,” could allow even better use of sparse observations. There are many open research questions in the development of strongly coupled data assimilation, so ERA-CLIM2 partners were involved in the organization of a Coupled Data Assimilation workshop in October 2016, sponsored by the World Meteorological Organization (WMO), to discuss these with the wider international research community.
Land and ocean carbon reanalyses are also being produced as part of ERA-CLIM2, and the techniques used to produce them are being developed for use in future reanalyses. Various new data streams have been tested for improving the accuracy of the land carbon reanalysis both through improved state estimation and through improving knowledge of the model parameters (Peylin et al. 2016). The methods used to couple ocean biogeochemical models to the physical ocean–atmosphere reanalysis system have also been investigated in order to provide improved information about the ocean carbon cycle.
ENSEMBLES OF REANALYSES TO ESTIMATE CONFIDENCE.
The accuracy of any physical measurement is limited. Furthermore, the spatial resolution of measurements and of assimilated gridded data is limited, and thus there may be deviations of the gridpoint values from the true values. Confidence (or uncertainty) is best described by the probability distribution of these deviations. The distribution itself can be characterized by a few parameters such as standard deviation or by a limited-size ensemble of realizations drawn from the distribution, from which the user may derive statistics. The latter approach consumes much more data storage but also leaves more choices for informed users.
With reanalysis products getting more mature, it is now possible to estimate their accuracy and assign to the reanalysis data some level of confidence. In ERA-CLIM2, this has been achieved by applying an ensemble approach, based on several complete realizations of all quantities.
Both the twentieth-century reanalysis CERA-20C and the coupled reanalysis of the satellite era, CERA-SAT, consist of 10 realizations, run in parallel. The 10 members can be used to estimate a range of possible states for all the reanalysis output variables. They have been generated using an ensemble of data assimilation (EDA) in the atmosphere (Isaksen et al. 2010; Bonavita et al. 2017) and with perturbing the positions of in situ ocean observations, the air–sea fluxes, and the SST following ECMWF’s Ocean Reanalysis System 5 (ORAS5) (Zuo et al. 2014). The SST perturbations lead to variations in important output parameters such as surface precipitation. Figure 4 shows how CERA-20C precipitation has improved over the earlier twentieth-century uncoupled ERA-20C (Poli et al. 2016) over less well-observed land areas, where ERA-20C significantly underforecasts high rainfall amounts. Figure 4 also shows how the calculated precipitation varies within the ensemble. For example, for strong precipitation episodes over Africa (Fig. 4a, upper right) the spread has been increased by about 20%, indicating that the ensemble is reliable.
Monthly regional-mean short-term precipitation forecasts from ERA-20C (black) and CERA-20C (gray) against Full Data Reanalysis, version 7 (Schneider et al. 2016), monthly precipitation estimates from rain gauges provided by the GPCC: (a) Africa, (b) India/monsoon region, and (c) South America. Time interval considered is 1901–2000 at 1° spatial resolution and monthly temporal resolution.
Citation: Bulletin of the American Meteorological Society 99, 5; 10.1175/BAMS-D-17-0199.1
CERA-20C: THE FIRST EUROPEAN 110-YR COUPLED OCEAN–LAND–ATMOSPHERE REANALYSIS.
One of the key results of the ERA-CLIM2 project is CERA-20C, the first European 110-yr coupled ocean, land, and atmosphere reanalysis of the twentieth century. Such a dataset will allow the community to investigate climate variability and trends, also taking into account how the three-dimensional ocean evolves. The accurate representation of variability on interannual and decadal time scales is a requirement for climate applications of reanalysis data, such as reconstructing the time evolution of the atmosphere surface temperature and of the ocean heat content (Fig. 5). Following the completion of CERA-20C, as part of the project we have also completed CERA-SAT, which includes 8 years of coupled reanalysis of the satellite era. These two datasets have demonstrated that it is feasible to aim to reconstruct the three-dimensional climate of the Earth system using a coupled approach. They provide state-of-the-art coupled gridded fields of the Earth system, which can be used not only to understand climate variability but also to estimate predictability of existing models (by providing initial conditions spanning 110 years) and to revisit cases of extreme interest for society, as we discussed above.
Maps of the yearly average global-mean anomalies in 2010 relative to the period 1961–90 for (a) the 2-m temperature and (b) the upper-ocean heat content in CERA-20C. The two time series show the evolution of the yearly average global mean.
Citation: Bulletin of the American Meteorological Society 99, 5; 10.1175/BAMS-D-17-0199.1
Climate signals in reanalyses are inevitably affected by changes in the global observing system and by the presence of time-varying biases in models and observations. To build confidence in climate change information derived from reanalyses, it is important that information about the data assimilation methodology, the forecast model, and the input observations are made available. It is also necessary to compare results based on reanalyses (ECMWF CERA-20C, NOAA 20CR, ECMWF ORA-20C) with results obtained using more traditional, observation-only climate datasets [the Climatic Research Unit Temperature, version 4 (CRUTEM4) 2-m temperature, Global Precipitation Climatology Centre (GPCC) precipitation data, EN4 ocean temperature (Good et al. 2013)], and to test whether the climate signals in CERA-20C are robust to different analysis methodologies. CERA-20C data have been made freely available from the ECMWF website (www.ecmwf.int/en/research/climate-reanalysis/cera-20c) precisely to favor these comparisons.
Reanalyses are valuable datasets that allow us to revisit past events and to provide more evidence on the weather conditions that could have led to extreme conditions. A demonstration case was discussed by Brugnara et al. (2017), who examined the weather conditions in December 1916, in the middle of the First World War, when a massive snowfall event in the Southern Alps triggered countless avalanches, which killed thousands of soldiers and civilians. This event was studied thanks to ERA-CLIM and ERA-CLIM2 funding in the first part of the ERA-CLIM2 project, using dynamical downscaling of the earlier ERA-20C reanalysis (the uncoupled reanalysis produced by the FP7 project ERA-CLIM, the precursor of ERA-CLIM2) in combination with historical observations. By looking at reanalysis data, the atmospheric conditions that led to such catastrophic events could be understood: a blocking-flow situation, moisture transport from the warm Mediterranean Sea toward the Alps, and a rapidly rising snow line, leading to a dangerous “rain on snow” situation.
SUMMARY AND CONCLUSIONS.
ERA-CLIM2 is a European Union Seventh Framework project started in January 2014, which involves 17 organizations (see appendix). It aims to produce coupled reanalyses, that is, physically consistent datasets describing the evolution of the global atmosphere, ocean, land surface, cryosphere, and the carbon cycle. The main contributions of the ERA-CLIM2 project to climate science have been to rescue and reprocess past conventional and satellite data, improve the capacity for producing state-of-the-art climate reanalyses that extend back to the early twentieth century, along with uncertainties, and generate unique and extremely valuable datasets. One of the main deliverables of ERA-CLIM2 has been CERA-20C, the first European coupled reanalysis of the twentieth century. CERA-20C is now being used to generate a land (water and energy) and a carbon (land and ocean) component (CERA-20C/Land and CERA-20C/Carbon). At the time of writing, the production of the CERA-SAT reanalysis has started; the aim is to complete the period from 2008 to to the present by the end of the project (December 2017). Thanks to ERA-CLIM2, many older data have been rescued and postprocessed and are delivered to relevant database providers so that they can be used in future reanalysis. Furthermore, new assimilation methods (e.g., use of a stronger coupling method between the ocean and the atmosphere and the direct assimilation of SST data) developed and tested within the project are planned to be integrated and used in future reanalysis productions.
Understanding climate change is highly dependent on the availability of global satellite and conventional observational data in the atmosphere, the land, and the ocean and sea ice, as well as the development of coupled ocean–land–atmospheric models and assimilation systems that can ingest these data. A continuous cycle of research and development in all key reanalysis activities (from data rescue and observation reprocessing, to assimilation methods for reanalysis, production and evaluation) is required to provide a continuously improving depiction of the time evolution of the Earth system.
Here are two examples of why we need a continuous stream of investments in the two areas mentioned above:
Within ERA-CLIM2 millions of new observation records, mostly made before the International Geophysical Year 1958, have been discovered, rescued, digitized, and prepared to be inserted in appropriate datasets so that they can be used in future reanalysis production. As this work has been progressing, new data are discovered, but unfortunately, owing to a lack of resources and time (the project finished in December 2017), they will not be rescued, digitized, quality controlled, and prepared to be inserted in the datasets.
As part of the ERA-CLIM2 work, the possibility of assimilating directly sea surface temperature observations has been explored and tested in prototype systems. Furthermore, the possibility of using ensemble methods to estimate flow-dependent background error statistics within the ocean has been developed and tested. Neither of these advances could be included in the current ERA-CLIM2, since tested software was not ready in time for production.
Preliminary assimilation experiments have shown that the amount and quality of those data justify a full reanalysis, using earlier satellite observations and all conventional (surface and upper air) data, back to the early twentieth century (Hersbach et al. 2017). Such a reanalysis would realize the potential of the data collected and would lead to a much better description of the climate evolution over the last century. Considering the European activities in reanalysis and climate monitoring, a sensible way to be able to continue to generate increasingly accurate reanalyses would be to fund follow-on activities either through a new stream of European projects or directly as part of the European Union C3S activities. In this way, Europe would continue to maintain a leading position in this field, building on the investments already made and the successes of projects such as ERA-CLIM and ERA-CLIM2.
APPENDIX: THE ERA-CLIM2 CONSORTIUM.
The ERA-CLIM2 Consortium included 17 organizations:
European Centre for Medium-Range Weather Forecasts (Europe)
Met Office (United Kingdom)
European Organization for the Exploitation of Meteorological Satellites (Europe)
University of Bern (Switzerland)
University of Vienna (Austria)
Instituto Dom Luiz Faculdade de Ciências da Universidade de Lisboa (Portugal)
Russian Research Institute of Hydrometeorological Information (Russia)
Mercator Ocean Societé Civile (France)
Météo-France (France)
Deutscher Wetterdienst (Germany)
Centre European de Recherche et de Formation Avancée en Calcul Scientifique (France)
Centro Euro-Mediterraneo sui Cambiamenti Climatici (Italy)
Ilmatieteen Laitos (Finland)
Universidad del Pacifico (Chile)
University of Reading (United Kingdom)
Institut National de Recherche en Informatique et en Automatique (France)
Université de Versailles Saint-Quentin-en-Yvelines (France)
REFERENCES
Bonavita, M., and Coauthors, 2017: A strategy for data assimilation. ECMWF Research Dept. Tech. Memo. 800, 42 pp.
Brönnimann, S., and Coauthors, 2018: Observations for reanalysis. Bull. Amer. Meteor. Soc., https://doi.org/10.1175/BAMS-D-17-0229.1, in press.
Brugnara, Y., S. Brönnimann, M. Zamuriano, J. Schild, C. Roher, and D. M. Segesser, 2017: Reanalysis sheds light on 1916 avalanche disaster. ECMWF Newsletter, No. 151, ECMWF, Reading, United Kingdom, 28–34.
Compo, G. P., and Coauthors, 2011: The Twentieth Century Reanalysis Project. Quart. J. Roy. Meteor. Soc., 137, 1–28, https://doi.org/10.1002/qj.776.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, https://doi.org/10.1002/qj.828.
Dee, D. P., M. A. Balmaseda, G. Balsamo, R. Engelen, A. J. Simmons, and J.-N. Thépaut, 2014: Toward a consistent reanalysis of the climate system. Bull. Amer. Meteor. Soc., 95, 1235–1248, https://doi.org/10.1175/BAMS-D-13-00043.1.
Donlon, C. J., M. Martin, J. Stark, J. Roberts-Jones, E. Fiedler, and W. Wimmer, 2012: The operational sea surface temperature and sea ice analysis (OSTIA) system. Remote Sens. Environ., 116, 140–158, https://doi.org/10.1016/j.rse.2010.10.017.
Feng, X., and K. Haines, 2018: Uncertainties of sea surface and air temperature in the CERA-20C coupled reanalysis ensemble. Quart. J. Roy. Meteor. Soc., 144, 195–207, https://doi.org/10.1002/qj.3194.
Ferguson, C. R., and G. Villarini, 2012: Detecting inhomogeneities in the Twentieth Century Reanalysis over the central United States. J. Geophys. Res., 117, D05123, https://doi.org/10.1029/2011JD016988.
Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). J. Climate, 30, 5419–5454, https://doi.org/10.1175/JCLI-D-16-0758.1.
Good, S. A., M. J. Martin and N. A. Rayner, 2013: EN4: Quality controlled ocean temperature and salinity profiles and monthly objective analyses with uncertainty estimates, J. Geophys. Res. Oceans, 118, 6704–6716, https://doi.org/10.1002/2013JC009067.
Haimberger, L., C. Tavolato, and S. Sperka, 2012: Homogenization of the global radiosonde temperature dataset through combined comparison with reanalysis background series and neighboring stations. J. Climate, 25, 8108–8131, https://doi.org/10.1175/JCLI-D-11-00668.1.
Hersbach, H., and Coauthors, 2017: The potential value of early (1939–1967) upper-air data in atmospheric climate reanalysis. Quart. J. Roy. Meteor. Soc., 143, 1197–1210, https://doi.org/10.1002/qj.3040.
Isaksen, L., M. Bonavita, R. Buizza, M. Fisher, J. Haseler, J. Leutbecher, and L. Raynaud, 2010: Ensemble data assimilations at ECMWF. ECMWF Research Dept. Tech. Memo. 636, 45 pp.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471, https://doi.org/10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.
Karspeck, A., A. Kaplan, and R. R. Sain, 2012: Bayesian modelling and ensemble reconstruction of mid-scale spatial variability in North Atlantic sea-surface temperatures for 1850–2008. Quart. J. Roy. Meteor. Soc., 138, 234–248, https://doi.org/10.1002/qj.900.
Kobayashi, S., and Coauthors, 2015: The JRA-55 reanalysis: General specifications and basic characteristics. J. Meteor. Soc. Japan, 93, 5–48, https://doi.org/10.2151/jmsj.2015-001.
Laloyaux, P., M. Balmaseda, D. Dee, K. Mogensen, and P. Janssen, 2016: A coupled data assimilation system for climate reanalysis. Quart. J. Roy. Meteor. Soc., 142, 65–78, https://doi.org/10.1002/qj.2629.
Laloyaux, P., E. de Boisseson, and P. Dahlgren, 2017: CERA-20C: An Earth system approach to climate reanalysis. ECMWF Newsletter, No. 150, ECMWF, Reading, United Kingdom, 25–30.
Mayer, M., L. Haimberger, J. M. Edwards, and P. Hyder, 2017: Toward consistent diagnostics of the coupled atmosphere and ocean energy budgets. J. Climate, 30, 9225–9246, https://doi.org/10.1175/JCLI-D-17-0137.1.
Onogi, K., and Coauthors, 2007: The JRA-25 Reanalysis. J. Meteor. Soc. Japan, 85, 369–432, https://doi.org/10.2151/jmsj.85.369.
Peylin, P., and Coauthors, 2016: A new stepwise carbon cycle data assimilation system using multiple data streams to constrain the simulated land surface carbon cycle. Geosci. Model Dev., 9, 3321–3346, https://doi.org/10.5194/gmd-9-3321-2016.
Ploshay, J. J., W. F. Stern, and K. Miyakoda, 1992: FGGE reanalysis at GFDL. Mon. Wea. Rev., 120, 2083–2108, https://doi.org/10.1175/1520-0493(1992)120<2083:FRAG>2.0.CO;2.
Poli, P., and Coauthors, 2016: ERA-20C: An atmospheric reanalysis of the twentieth century. J. Climate, 29, 4083–4097, https://doi.org/10.1175/JCLI-D-15-0556.1.
Poli, P., and Coauthors, 2017: Recent advances in satellite data rescue. Bull. Amer. Meteor. Soc., 98, 1471–1484, https://doi.org/10.1175/BAMS-D-15-00194.1.
Rienecker, M. M., and Coauthors, 2011: MERRA: NASA’s Modern-Era Retrospective Analysis for Research and Applications. J. Climate, 24, 3624–3648, https://doi.org/10.1175/JCLI-D-11-00015.1.
Saha, S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 1015–1057, https://doi.org/10.1175/2010BAMS3001.1.
Schneider, U., A. Becker, P. Finger, A. Meyer-Christoffer, B. Rudolf, and M. Ziese, 2016: GPCC full data reanalysis, version 7.0: Monthly land-surface precipitation from rain gauges built on GTS based and historic data. Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory. https://doi.org/10.5065/D6000072.
Schubert, S. D., R. Rood, and J. Pfaendtner, 1993: An assimilated dataset for Earth science applications. Bull. Amer. Meteor. Soc., 74, 2331–2342, https://doi.org/10.1175/1520-0477(1993)074<2331:AADFES>2.0.CO;2.
Storto, A., M. J. Martin, B. Deremble, and S. Masina, 2018: Strongly coupled data assimilation experiments with linearized ocean–atmosphere balance relationships. Mon. Wea. Rev., https://doi.org/10.1175/MWR-D-17-0222.1, in press.
Titchner, H. A., and N. A. Rayner, 2014: The Met Office Hadley Centre sea ice and sea surface temperature data set, version 2: 1. Sea ice concentrations. J. Geophys. Res. Atmos., 119, 2864–2889, https://doi.org/10.1002/2013JD020316.
Uppala, S. M., and Coauthors, 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131, 2961–3012, https://doi.org/10.1256/qj.04.176.
Weaver, A. T., J. Tshimanga, and A. Piacentini, 2016: Correlation operators based on an implicitly formulated diffusion equation solved with the Chebyshev iteration. Quart. J. Roy. Meteor. Soc., 142, 455–471, https://doi.org/10.1002/qj.2664.
Zuo, H. M., A. Balmaseda, and K. Mogensen, 2014: The ECMWF-MyOcean2 eddy-permitting ocean and sea-ice reanalysis ORAP5. Part 1: Implementation. ECMWF Research Dept. Tech. Memo. 736, 44 pp.
