Overview of current atmospheric reanalyses

Submitted by Cathy.Smith@noaa.gov on Thu, 10/07/2010 - 12:47


Ask a Question

Current / State-of-the-art: 

ASR | COSMO-REA | CERA-20C | ERA-20C | ERA-20CM | ERA-Interim | JRA-55, JRA-55C, JRA-55AMIP | MERRA-2 |  NCEP CFSR | NOAA-CIRES 20CRv2c

Possible issues: Consider other datasets for use in new research projects: NCEP/DOE II  | NCEP/NCAR NCEP NARR 

Superseded / Caution use for new research projects: ERA-40 | ERA-15 JRA-25

Model Change: NCEP CFSR (2011 and after)

Updated in real-time for public use (days behind): NCEP/DOE II  | NCEP/NCAR NCEP NARR  | NCEP CFSR? |  JRA-55 (2 days behind from JMA suite) |
Updated in near real-time for public use (months behind)ERA-Interim | JRA-55 | MERRA-2
Updated irregularly for public use (years behind): CERA-20C |ERA-20C | ERA-20CM | NOAA-CIRES 20CR NOAA-CIRES 20CRv2cNASA MERRA


Overview Comparison Table (Reanalyses.org)

Overview Comparison Table (ClimateDataGuide)

Overview Comparison Table (as of 2016, S-RIP)

Notes,questions, and discussion by dataset


Arctic System Reanalysis (ASR): 2000-2012

The Arctic System Reanalysis (ASR), a high-resolution regional assimilation of model output, observations, and satellite data across the mid- and high latitudes of the Northern Hemisphere for the period 2000 – 2012 has been performed at 30 km (ASRv1) and 15 km (ASRv2) horizontal resolution using the polar version of the Weather Research and Forecasting (WRF) model and the WRF Data Assimilation (WRFDA) System.

Source: Polar Meteorology Group, Byrd Polar & Climate Research Center, The Ohio State University

Time Range: 2000-2012

Assimilation: WRFDA-3DVAR

Dataset Output Times and Time Averaging: 3-hourly for surface and upper air fields, Monthly means of selected variables

ASRv1 – 30 km
Model Resolution: 30 km, 71 sigma levels
Dataset location: http://rda.ucar.edu/datasets/ds631.0/
ASRv2 30 km is expected early 2017

ASRv2 – 15 km (Currently updating through 2016 - available end of 2017)
Model Resolution: 15 km, 71 sigma levels
Dataset location: https://rda.ucar.edu/datasets/ds631.1/

Data Access: Polar Meteorology Group

References | ClimateDataGuide


COSMO Regional Reanalyses: 1995-present

Several reanalysis systems are developed to generate high-resolution regional reanalysis data sets for Europe based on the NWP model COSMO. The first available data set at a horizontal resolution of 6km covers the entire European continent for the years 1995-2015 with output of model variables being available for every hour. A subsequent second data set covers large parts of Central Europe in a convection-permitting setup with 2km horizontal resolution and is available for the period 2007-2013.  

Reanalysis consortium: Hans-Ertel-Centre - Climate Monitoring and Diagnostics, German Meteorological Service (DWD) - Climate and Environment, Meteorological Institute of the University of Bonn, Institute for Geophysics and Meteorology of the University of Cologne

Time Range: 1995-2015

Assimilation: Continuous nudging, separate DA modules for soil moisture, snow, SST and sea ice, latent heat nudging scheme

Dataset Output Times: 15 minutes for surface, hourly for upper air fields

COSMO-REA6 – 6.2 km
Model Resolution: 6.2 km, 40 eta levels
Period available: 1995-2015, production for year 2016 is currently under way, will be updated regularly
Domain: Europe (CORDEX-11 domain)
Production of a second (updated) version of COSMO-REA6 is expected to start late 2017 / early 2018

COSMO-REA2 – 2 km
Model Resolution: 2 km, 50 eta levels
Period available: 2007-2013, production for 2014-2015 is currently under way, will be updated regularly
Domain: Central Europe

Data Access: COMSO reanalysis website



ECMWF CERA-20C: 1901-2010

CERA-20C is a global 20th-century reanalysis which aims to reconstruct the past weather and climate of the Earth system including the atmosphere, ocean, land, waves and sea ice. CERA-20C is part of the EU-funded ERA-CLIM2 project and extends the reanalysis capability developed in ERA-20C to the ocean and sea-ice components.

A new assimilation system (CERA) has been developed to simultaneously ingest atmospheric and ocean observations in the coupled Earth system model used for ECMWF’s ensemble forecasts. It is based on a variational method with a common 24-hour assimilation window. Air-sea interactions are taken into account when observation misfits are computed and when the increments are applied to the initial condition. In this context, ocean observations can have a direct impact on the atmospheric analysis and, conversely, atmospheric observations can have an immediate impact on the analysed state of the ocean.

CERA-20C assimilates only surface pressure and marine wind observations (ISPDv3.2.6 and ICOADSv2.5.1) as well as ocean temperature and salinity profiles (EN4). The air-sea interface is relaxed towards the sea-surface temperature from the HadISST2 monthly product to avoid model drift while enabling the simulation of coupled processes. No data assimilation is performed in the land, wave and sea-ice components, but the use of the coupled model ensures some dynamical consistency.

The evolution of the global weather for the period 1901–2010 is represented by a ten-member ensemble of 3-hourly estimates for ocean, surface and upper-air parameters. The resolution of the atmospheric model is set to TL159L91 (IFS version 41r2), which corresponds to a 1.125° horizontal grid (125 km) with 91 vertical levels going up to 0.1hPa. The ocean model (NEMO version 3.4) uses the ORCA1 grid, which has approximately 1° horizontal resolution with meridional refinement at the equator. There are 42 vertical ocean levels with a first-layer thickness of 10m.

Data Access: ECMWF

References | Data documentation | ClimateDataGuide


ECMWF ERA-20C: 1900 - 2010

ERA-20C is ECMWF's first atmospheric reanalysis of the 20th century, from 1900-2010. It is an outcome of the ERA-CLIM project.

ERA-20C was produced with the same surface and atmospheric forcings as the final version of the atmospheric model integration ERA-20CM. A coupled Atmosphere/Land-surface/Ocean-waves model is used to reanalyse the weather, by assimilating surface observations. The ERA-20C products describe the spatio-temporal evolution of the atmosphere (on 91 vertical levels, between the surface and 0.01 hPa), the land-surface (on 4 soil layers), and the ocean waves (on 25 frequencies and 12 directions). The horizontal resolution is approximately 125 km (spectral truncation T159). Note, atmospheric data are not only available on the native 91 model levels, but also on 37 pressure levels (as in ERA-Interim), 16 potential temperature levels, and the 2 PVU potential vorticity level. Monthly means, daily, and invariant data are available. The temporal resolution of the daily products is usually 3-hourly.

The assimilation methodology is 24-hour 4D-Var analysis, with variational bias correction of surface pressure observations. Analysis increments are at T95 horizontal resolution (aprox. 210 km). The analyses provide the initial conditions for subsequent forecasts that serve as backgrounds to the next analyses. A 10-member ensemble was produced initially, to estimate the spatio-temporal evolution of the background errors.

The observations assimilated in ERA-20C include surface pressures and mean sea level pressures from ISPDv3.2.6 and ICOADSv2.5.1, and surface marine winds from ICOADSv2.5.1. The observation feedback from ERA-20C is available. It includes the observations but also departures before and after assimilation and usage flags.

Data Access: ECMWF | NCAR

References | ClimateDataGuide


ECMWF ERA-20CM Model integration (no data assimilation): 1900 - 2010

The ERA-20CM atmospheric model integrations were produced in the framework of the ERA-CLIM project.

There are two versions, ERA-20CM and ERA-20CMv0, each comprising of a 10-member ensemble. The first version is 'final', the second is 'experimental'. The 'experimental' version was an early production and should not be used to initiate new research.

The model integration is forced by radiative forcing from CMIP5 and also by sea-surface temperature (SST) and sea ice cover from HadISST2.

Access: ECMWF

References | ClimateDataGuide

ECMWF Interim Reanalysis (ERA-Interim): 1979-present

ERA-Interim was originally planned as an 'interim' reanalysis in preparation for the next-generation extended reanalysis to replace ERA-40. It uses a December 2006 version of the ECMWF Integrated Forecast Model (IFS Cy31r2). It originally covered dates from 1 Jan 1989 but an additional decade, from 1 January 1979, was added later. ERA-Interim is being continued in real time. The spectral resolution is T255 (about 80 km) and there are 60 vertical levels, with the model top at 0.1 hPa (about 64 km). The data assimilation is based on a 12-hourly four-dimensional variational analysis (4D-Var) with adaptive estimation of biases in satellite radiance data (VarBC). With some exceptions, ERA-Interim uses input observations prepared for ERA-40 until 2002, and data from ECMWF's operational archive thereafter. See Dee et al. (2011) in the references below for a full description of the ERA-Interim system.

Data Access: ECMWF | NCAR

References | ClimateDataGuide


ECMWF 40 Year Reanalysis (ERA-40): Sep 1957-Aug 2002

Completed in 2003, ERA-40 is a global atmospheric reanalysis of the 45-year period 1 September 1957 - 31 August 2002. It was produced using a June 2001 version of the ECMWF Integrated Forecast Model (IFS Cy28r3). The spectral resolution is T159 (about 125 km) and there are 60 vertical levels, with the model top at 0.1 hPa (about 64 km). Observations were assimilated using a 6-hourly 3D variational analysis (3D-Var). Satellite data used include Vertical Temperature Profile Radiometer radiances starting in 1972, followed by TOVS, SSM/I, ERS and ATOVS data. Cloud Motion Winds are used from 1979 onwards. Various data from past field experiments were used, such as the 1974 Atlantic Tropical Experiment of the Global Atmospheric Research Program, GATE, 1979 FGGE, 1982 Alpine Experiment, ALPEX and 1992-1993 TOGA-COARE.

Data Access: ECMWF  | NCAR

References | ClimateDataGuide

ECMWF 15 Year Reanalysis (ERA-15): Jan 1979-Dec 1993

Completed in 1996, ERA-15 is a global atmospheric reanalysis of the 15-year period 1 January 1979 - 31 December 1993. It was produced using an April 1995 version of the ECMWF Integrated Forecast Model (IFS Cy13r4). The spectral resolution is T106 (about 190 km) and there are 31 vertical levels, with the model top at 10 hPa (about 31 km altitude). Observations were assimilated using a 6-hourly Optimum Interpolation analysis (OI). Satellite data used were limited to cloud-cleared TOVS radiances and Cloud Motion Winds from GOES, GMS, and METEOSAT. Pseudo-observations of surface pressure (PAOBS) were also used, as well various data from past field experiments: 1979 FGGE, 1982 Alpine Experiment (ALPEX), TOGA, SUBDUCTION, and those found in the COADS dataset.

Data Access: ECMWF  | NCAR

References | ClimateDataGuide

Japanese 25-year Reanalysis (JRA-25): 1979-2004, JCDAS: 2005-Jan.2014

The Japanese 25-year Reanalysis (JRA-25) represents the first long-term global atmospheric reanalysis undertaken in Asia. Covering the period 1979-2004, it was completed using the Japan Meteorological Agency (JMA) numerical assimilation and forecast system and specially collected and prepared observational and satellite data from many sources including the European Center for Medium-Range Weather Forecasts (ECMWF), the National Climatic Data Center (NCDC), and the Meteorological Research Institute (MRI) of JMA. A primary goal of JRA-25 is to provide a consistent and high-quality reanalysis dataset for climate research, monitoring, and operational forecasts, especially by improving the coverage and quality of analysis in the Asian region.  JRA-25 was conducted by JMA and CRIEPI (Central Research Institute of Electric Power Industry).  It had been continued as JCDAS (JMA Climate Data Assimilation System) operated by JMA in near real time basis. The data assimilation systems of JRA-25 and JCDAS are the same.  Users can use JRA-25 and JCDAS as one continuous reanalysis dataset. JCDAS data provision was terminated in early Feburay 2014 because it was replaced with JRA-55 in operational. The available data period of JRA-25/JCDAS is for 35 years and 1 month (January 1979 to January 2014).

Data Access: NCAR

Homepage | Atlas | References | ClimateDataGuide

Japanese 55-year Reanalysis (JRA-55): 1958-2012,  and extended to present
     [JRA-55C(1972-2012) and JRA-55AMIP(1958-2012)] <-- not extended to present

JMA has carried out the second reanalysis project named the Japanese 55-year Reanalysis (JRA-55) (nicknamed JRA Go! Go!) using a more sophisticated NWP system, which is based on the operational system as of December 2009, and newly prepared past observations. The analysis period is extended to 55 years starting from 1958, when the regular radiosonde observations became operational on the global basis. Many of deficiencies in JRA-25 have been diminished or reduced in JRA-55 because many improvements achieved after JRA-25 have been introduced. JRA-55 provides a consistent climate dataset over the last half century. JRA-55 has been continued in near real time basis after 2013. If you need real time basis (2 days behind) data, please access to JMA. The real time basis data are also provided to other cooperating organizations (Note: half year behind). Note that the products for extended period after 2013 are also called JRA-55. As "JRA-55 family", there are two subproducts JRA-55C and JRA-55AMIP produced by MRI/JMA. JRA-55C assimilated conventional observations only. JRA-55C covers from 1972 to 2012. Before 1971, use JRA-55 instead because no satellite data was assimilated in JRA-55 before 1971. JRA-55AMIP (AMIP type run of JRA-55, with no observations) covers 1958 to 2012. JRA-55C and JRA-55AMIP data are available from DIAS and NCAR. Note that JRA-55C and JRA-55AMIP are not extended to present. JRA-55 Atlas (climate charts) is now available.

Data Access: JMA | DIAS (JRA-55) (JRA-55C) (JRA-55AMIP) | NCAR (JRA-55: Daily 3-Hourly and 6-Hourly DataMonthly Means and Variances) (JRA-55C: Daily(3-hourly,6-hourly), Monthly) (JRA-55AMIP: Daily(3-hourly,6-hourly), Monthly) | ESGF/NASA/WCRP | ECMWF |

Homepage | References | JRA-55 Atlas |



NASA Modern Era Reanalysis for Research and Applications (MERRA): 1979-2016(Feb)

MERRA is a NASA reanalysis for the satellite era using a major new version (circa 2008) of the Goddard Earth Observing System Data Assimilation System Version 5 (GEOS-5) produced by the NASA GSFC Global Modeling and Assimilation Office (GMAO). The Project focused on historical analyses of the hydrological cycle on a broad range of weather and climate time scales and placed the NASA EOS suite of observations in a climate context.

Data Access: GES MDISC | ESGF

Home Page | References | FAQ | Atlas | ClimateDataGuide | AMS Special Collection

NASA Modern Era Reanalysis for Research and Applications Version-2 (MERRA-2): 1980-present

MERRA-2 is a NASA reanalysis for the satellite era using a major new version of the Goddard Earth Observing System Data Assimilation System Version 5 (GEOS-5) produced by the NASA GSFC Global Modeling and Assimilation Office (GMAO). MERRA-2 assimilates observations not available to MERRA during the 2010s, and therefore, will continue processing in real time longer than MERRA.  There are numerous improvements and updates to the data assimilaiton, model and observing system. One notable change is the assimilation of aerosol observations, including black and organic carbon, sulfate and dust. Production began in the spring of 2014 and is presently available for access.

Data Access: GES MDISC | FTP Subsetter

Home Page | File Specification | Documentation | AMS Special Collection



NCEP Climate Forecast System Reanalysis (CFSR): 1979-present

The National Centers for Environmental Prediction (NCEP) Climate Forecast System Reanalysis (CFSR) spans 1979 to present. The CFSR was designed and executed as a global, high resolution, coupled atmosphere-ocean-land surface-sea ice system to provide the best estimate of the state of these coupled domains over this period. The T382 resolution atmospheric data spans 1979 to 2010. The current T574 analysis is an extension of the CFSR as an operational, real time CFSv2 product from 2011 into the future.

Data Access: NCEP | NCDC NOMADS | NCAR (includes real time CFSv2) | ESGF

References | ClimateDataGuide


NCEP/DOE Reanalysis II: 1979-near present

NCEP produced a second version of their first reanalysis starting from the beginning of the major satellite era. More observations were added, assimilation errors were corrected and a better version of the model was used.


References | ClimateDataGuide

NCEP/NCAR Reanalysis I: 1948-present

This reanalysis was the first of its kind for NOAA. NCEP used the same climate model that were initialized with a wide variety of weather observations: ships, planes, RAOBS, station data, satellite observations and many more. By using the same model, scientists can examine climate/weather statistics and dynamic processes without the complication that model changes can cause. The dataset is kept current using near real-time observations.


References | FAQ | FGDC | ClimateDataGuide

NCEP North American Regional Reanalysis (NARR): 1979-near present

The NARR reanalysis was done to produce very high resolution output over the North American region. Observational inputs were similar to NCEP I with the addition of assimilated precipitation. The NARR model region was nested in a global, lower resolution model. Outputs are similar to the NCEP I and II models but with more snow, ice and precipitation related variables.


References | FAQ | ClimateDataGuide

NOAA-CIRES 20th Century Reanalysis version 2 (20CRv2): 1871-2012

The 20th Century Reanalysis version 2 (20CRv2) dataset contains global weather conditions and their uncertainty in six hour intervals from the year 1871 to 2012. Surface and sea level pressure observations are combined with a short-term forecast from an ensemble of 56 integrations of an NCEP numerical weather prediction model using the Ensemble Kalman Filter technique to produce an estimate of the complete state of the atmosphere, and the uncertainty in that estimate. The uncertainty is approximately inversely proportional to the density of observations. Additional observations and a newer version of the NCEP model that includes time-varying CO2 concentrations, solar variability, and volcanic aerosols are used in version 2. The long time range of this dataset allows scientists to examine better long time scale climate processes such as the Pacific Decadal Oscillation and the Atlantic Multidecadal Oscillation as well as looking at the dynamics of historical climate and weather events. Verification tests have shown that using only pressure creates reasonable atmospheric fields up to the tropopause. Additional tests suggest some correspondence with observed variations in the lower stratosphere.


Homepage | References | Related Publications | ClimateDataGuide

NOAA-CIRES 20th Century Reanalysis version 2c (20CRv2c): 1851-2012 [2013-2014]

The 20th Century Reanalysis version 2 (20CRv2c) dataset contains global weather conditions and their uncertainty in six hour intervals from the year 1851 to 2012. Surface and sea level pressure observations are combined with a short-term forecast from an ensemble of 56 integrations of an NCEP numerical weather prediction model using the Ensemble Kalman Filter technique to produce an estimate of the complete state of the atmosphere, and the uncertainty in that estimate. The uncertainty is approximately inversely proportional to the density of observations. Additional observations from ISPDv3.2.9 and new boundary conditions from the Simple Ocean Data Assimilation with sparse observational input (SODAsi.2) pentad sea surface temperature and COBE-SST2 monthly sea ice concentration are used.  For the 2013 and 2014 extension, SSTs are from the NOAA daily 1/4 degree Optimal Interpolation version 2. 2014 does not include tropical cyclone pressure observations after April. 

Data Access: ESRL |  NCAR | ESGF | NERSC (every member)

Homepage | References | Related Publications ClimateDataGuide



Have a question or comment on a specific reanalysis dataset or in general?

Use this link to Post Notes, Questions, and Comments on specific reanalyses



Paul Berrisford (not verified)

Sat, 10/07/2017 - 04:12

Dear Yiyi,

In ERA-20C data the monthly means of daily means are averages over the available hours in a day, then over all days in the month, to represent the whole or complete monthly mean. Synoptic monthly means are averages over days in the month for a particular synoptic time eg 12 UTC, or in the case of accumulations, for only a particular accumulation period within the day eg 06 to 12 UTC.

For most applications you would use the monthly means of daily means. You would only use the synoptic monthly means if you wanted to restrict your attention to a particular point in, or portion of, the diurnal cycle.


Paul Berrisford


Fri, 10/06/2017 - 12:46

To be consistent with comparisons with 20CRv2c which or the ERA20C products should we use in our ESGF archive?  synoptic monthly means: In the case of analyses, these are averages throughout the calendar month for each available synoptic hour, whereas in the case of forecasts (all issued daily from 06 UTC), they are averages throughout the calendar month for each available forecast step up to 24 hours. or monthly means of daily means: In the case of analyses, these are averages throughout the calendar month across all the available synoptic hours,  whereas in the case of forecasts (all issued daily from 06 UTC), they are averages throughout the calendar month across all the available forecast steps up to 24 hours, for instantaneous forecasts, or just for step=24 hours, for accumulated forecasts.

Yiyi (not verified)

Tue, 02/21/2017 - 18:20

Hello everyone,

Currently I am using the cloud fraction and radiation flux monthly data from MERRA-2, 20CRv2c, ERA-Interim, JRA-55 and CFSR. As far as I know, these variables are calculated in the forecast portion of forecast-analysis update cycles. So I was wondering over which period the radiation fluxes and cloud variables are calculated (e.g., the first 12 hours of each forecast) in each reanalysis. I would be really appreciated if you can provide information for any one of them.



Wesley Ebisuzaki (not verified)

Wed, 12/14/2016 - 09:00


For your storm surge model you probably want surface winds (primary), surface pressure (secondary) and geopotential and temperature (tertiary).

The GFS analyses use satellite scatterometers to estimate the surface wind speed over water.  Both NCEP/NCAR and NCEP/DOE do not use these instruments.  However, CFSR, ERA-Interim, MERRA and JRA-55 do use scatterometers.

The first generation reanalyses had the resolution to resolve synoptic-scale+ features.  For many phenomenon, such as ENSO, this is adequate.  For storm surges, you want something that that can resolve much finer scale features.  The modern reanalyses (CFSR, ERA-Interim, MERRA, JRA-55) are run at a much higher resolution and are more suited.  However, they are run at a much lower resolution than the current GFS (GFS:T1534 vs CFSR:T384/T574)

The near-surface winds are strongly influenced by the physics in the model. The CFSR will more similar to the GFS, so the CFSR is looking more favorable. The CFSR products have not restriction for commercial use which is another advantage.  One disadvantage of the CFSR is difficulty in obtaining it from the official NOAA archive (NCEI) especially data from 2010 and onwards.  (I don't follow the situation at NCEI and the data may be available.)

My suggestion is to use the CFSR if you can get the data.  If you are affected by the commercial restriction, then the MERRA-2 reanalysis is available without a restriction for commercial use.  If the commercial restriction doesn't apply, I don't know which of the modern reanalyses gives the best surface winds.



Alberto Canestrelli (not verified)

Thu, 11/24/2016 - 08:16

Dear all, we are group working on storm surge forecasting. We are interested in using GFM as input for our statistical and deterministic models for storm surges. We have a couple of questions on this regard:

1) Our statistical models need a large historical dataset for calibration (mainly surface wind speed, surface pressure, geopotential and temperature), therefore we were planning to use reanalysis for calibration. In the website (http://www.esrl.noaa.gov/psd/data/gridded/reanalysis/ ) it looks like the best reanalysis are NCEP/DOE Reanalysis II  (2.5 deg resolution) and 20th Century Reanalysis (V2 and V2c) (2 deg resolution). Which one do you think is the one closer to the GFS operational model, which has a 0.25 degrees resolution, and which we will use operationally?

2) Our Deterministic (shallow water) models need a smaller dataset for calibration (only surface pressure and surface wind ). 5 or 10 years would be enough. Are the past GFS analysis at 0.25 degrees of resolution available for the past 5-10 years? Again we will use GSF operationally, therefore we need to calibrate using the same model. 

Thanks a lot. Regards


Ashwin (not verified)

Wed, 09/21/2016 - 20:27

Is the same sigma coordinate used in the reanalysis data from UCAR FNL ? 0.25 x 0.25 degree resolution ? Are the 28 levels the same ?

Dear Ashwin,

I will need more information to be able to provide any assistance on this. What is the UCAR FNL? Do you have a web address?

What is the reference reanalysis you are asking about? Is it JRA55, ERA-Interim, CFSR?  You mention 28 levels. That could be 20CR or NCEP-NCAR reanalysis, but 0.25 resolution is more like a modern full-input reanalysis.

I will look forward to hearing from you.

Best wishes,


sujan ghimire (not verified)

Fri, 09/09/2016 - 04:57

I am doing solar modelling and have downloaded the data but it is at 0.00 which is UTC and our region is UTC +10. so it means that the data is for 10AM here which is not useful and next hour is 6 which is 16hrs so it means 4 PM over here. So with the 10AM data and 4 PM data how can i do the modelling?

How can i download the daily data on hourly basis? Any idea highly appreciated.

Paul Berrisford (not verified)

Wed, 10/12/2016 - 12:04

In reply to by sujan ghimire (not verified)


There is no hourly data in ERA-Interim. The analyses are 6 hourly and the twice daily forecasts (from 00 and 12 UTC) provide 3 hourly output to T+24 hours on pressure levels and at the surface, with output becoming less frequent to T+240, and to T+12 hours on model levels. Only a sub-set of this data is available on the point and click web interface (http://apps.ecmwf.int/datasets/data/interim-full-daily/levtype=sfc/), but all can be obtained by using script access to the data, see https://software.ecmwf.int/wiki/display/WEBAPI/Access+ECMWF+Public+Datasets.

Regards, Paul Berrisford

Julio Cardona (not verified)

Fri, 06/10/2016 - 10:33

Hello I am new to this reanalysis data, I would like to know where can I download data from precipitation and monthly average temperature from 1950 for the region of north central Mexico with good resolution and what software or application you recommend to view and extract data downloaded from the server because the files have .nc commonly extension. I am an engineering student at the Autonomous University of Zacatecas in Mexico and I'm doing my teisis in the analysis of drought, for which I need the data. Beforehand thank you very much

Many applications can read netCDF files. See unidata http://www.unidata.ucar.edu/software/netcdf/software.html and http://www.esrl.noaa.gov/psd/data/gridded/tools.html As far as which dataset, all reanalyses have precipitation and temperature (though precip is usually model output and not assimilated). See this table for resolution and time coverage. https://reanalyses.org/atmosphere/comparison-table There are also observed temperature and precipitation files. See http://www.esrl.noaa.gov/psd/data/gridded/tables/precipitation.html and http://www.esrl.noaa.gov/psd/data/gridded/tables/temperature.html NCAR archives many datasets, in addition. http://rda.ucar.edu Cathy

Wenjun Cui (not verified)

Fri, 06/03/2016 - 14:10

Hello all, I have a question about CFSR precipitation data. According to Saha et al. (2010), for land-surface analysis, the model-generated precipitation is replaced by the a mix of observation-based (CMAP and CPCU) and model-generated precipitation as forcing data. So I am wondering, are the CFSR precipitation data I downloaded generated by model or generated by merging observations and model out? Thanks, Wenjun

Dear Wenjun, A little more information is needed to make a helpful reply. What "CFSR precipitation data" are you using? What is the source of this data? Where did this person download it and what is the exact name of the file you are looking at and the link to where you obtained it? There are several distributors of CFSR data. Thanks for clarifying,

Hello Dr. Compo, I downloaded CFSR monthly precipitation data from the UCAR/NCAR Research Data Archive, the link is "http://rda.ucar.edu/datasets/ds093.2/#!description". The variable I downloaded is total precipitation, and the file name is "pgbh.gdas.yyyymm.grb2.nc". Thank you for your time, Wenjun

Courtesy of Jesse Meng and Suru Saha of NCEP When you download the CFSR data, you are getting the model generated precipitation. The mixed precipitation approach is used in land analysis only, not in the product. Hope this answers the questions.

Rocio (not verified)

Mon, 05/30/2016 - 10:29

Hi all. I'm pretty new to all this stuff. Can anyone please give me a simplified, step by step explanation of how to get and use data from any of these reanalysis websites? I'm a student at a University in Cuba. I stumbled on this and I'd like to use data from this for my final dissertation.I need temperature, preassure and relative humidity data from at lees 1985 untill nowadays, with the best resolution I can get, I'll be using python to analize and plot the data. I'd really appreciate any help I can get. Thank you.

Dear Rocio, You can click on the Data Access links above, or it may be more convenient to use some plotting web tools first. See the page http://reanalyses.org/atmosphere/how-obtainplotanalyze-data . You didn't mention at what time resolution you need the data: subdaily, daily, or monthly? The table at http://reanalyses.org/atmosphere/comparison-table may help you select some datasets to examine. As to "best", it entirely depends on your application. I am afraid you will need to experiment some and see what works for you. Best wishes,

Anonymous (not verified)

Fri, 05/20/2016 - 05:39

Good day, I want to obtain the moisture bugdet term P-E from the evapotranpiration and Total precipitation data sets, and the evaporation values are negative. any explanation. Also I may want to know the unit of evapotranspiration. thanks

Dear Keerthi, For ERA-Interim "daily" data, the units of evaporation and precipitation are "m", and these quantities are accumulated from the beginning of the forecast for step hours. To convert to mm/day (for the period of step hours) you need to multiply by 1000*24/step. However, if you want a value of evaporation over one day, you could add the values for time=00, step=12 and time=12, step=12 and then multiply by 1000. Note that vertical fluxes are defined to be positive downwards, so precipitation is positive and evaporation is usually negative (though condensation would be positive). Regards, Paul

Hi all,

I've run across what is either a 2-day discrepancy or a misunderstanding on my part of the time coordinate in MERRA2 data accessed through OpenDAP. The description of the 'time' variable is 'days since 1-1-1 00:00:00'. However, the first element of the MERRA2 time vector, corresponding to 1980-1-1 00:00:00, is 722816 (days), whereas when I compute the difference between 1980-1-1 and 1-1-1 using Python's 'datetime', I get 722814 days. I've checked 'datetime' against test data in Appendix C of Dershowitz and Reinhold's book 'Calendrical Calculations', and it appears to be correct. Any ideas? I've appended a short iPython session demonstrating the issue below.


Scott Paine
Smithsonian Astrophysical Observatory

In [1]: import pydap.client

In [2]: import datetime

In [3]: dataset = pydap.client.open_url('http://goldsmr4.sci.gsfc.nasa.gov:80/dods/M2I1NXASM')

In [4]: var = dataset['time']

In [5]: var.attributes
{'grads_dim': 't',
'grads_mapping': 'linear',
'grads_min': '00z01jan1980',
'grads_size': '324362',
'grads_step': '60mn',
'long_name': 'time',
'maximum': '01z01jan2017',
'minimum': '00z01jan1980',
'resolution': 0.0416666679084301,
'units': 'days since 1-1-1 00:00:0.0'}

In [6]: var[0]
Out[6]: array([ 722816.])

In [7]: d1 = datetime.date(1980, 1, 1)

In [8]: d0 = datetime.date(1, 1, 1)

In [9]: delta = d1 - d0

In [10]: delta.days
Out[10]: 722814

We've seen this 2 day error that some calendar calculations make and which I think the python and the calendar script are making.I think it is counting 2 years as leap years that are not. udunits (which we use) has the correct calculation and I believe MERRA2 does as well. Because the base date (1-1-1) is so far back, many calendar programs don't handle it correctly and we have shifted in our group to try to use a base date after the Gregorian calendar started so all calendar programs handle it correctly. I think this rule is the sticking point "The Revised Julian calendar adds an extra day to February in years that are multiples of four, except for years that are multiples of 100 that do not leave a remainder of 200 or 600 when divided by 900." .

Yiyi (not verified)

Wed, 04/13/2016 - 15:39

Hello there, Since I need to convert sea ice concentration to sea ice extent in MERRA-2 (0.625 x 0.5) and JRA-55 (1.25 x 1.25). I was wondering if these two products can provide gridded surface area information. Any help would be appreciated. Thanks, Yiyi

Daniel (not verified)

Tue, 03/15/2016 - 18:19

Hello, I have been working with the ensemble mean of the 20CR. However, I know that the 20CR ensemble mean is based on 56 members. I would like to work with some of this realizations. First of all, where do I more information about them? It seems that most information deals with the ensemble mean. How do I download them? Is is possible to know before downloading which members represent more extreme climates? Many thanks for you help!

Daniel, See the link on this page https://reanalyses.org/atmosphere/overview-current-reanalyses#TWENTv2c with "NERSC (every member)". This will take you to portal.nersc.gov, which has a "Browse" button. Use 20CR version 2c. Please feel free to respond to this thread with questions.

Anonymous (not verified)

Tue, 03/15/2016 - 10:10

Hi everyone, I am trying to use ERA Interim and CFSR for hydrological modelling in a mesoscale catchment in Africa. I have obtained CFSR data for SWAT modelling from globalweather.tamu.edu and ERA Interim data from http://apps.ecmwf.int/datasets/. However i have a problem using ERA Interim data as I can locate elevation data which is needed by SWAT model. Can some one direct me to where i can download elevation data? Thanks in advance for you help. Nkiaka

Dear Nkiaka, I presume you require the elevation of the surface of the Earth. The invariant (ie unchanging) fields are available from http://apps.ecmwf.int/datasets/data/interim-full-invariant/ and includes the (surface) geopotential. Divide by g (9.80665) to obtain surface geopotential height. Regards, Paul Berrisford

Senya Grodsky (not verified)

Mon, 03/14/2016 - 14:49

Is snowfall included in ERA-Interim total precipitation? Yes. precipitation is the liquid equivalent of all precipitation including rain/snow etc.

Yiyi (not verified)

Mon, 03/07/2016 - 13:42

Hello, Currently I am processing the cloud water content (for entire atmosphere) data in 20CRv2c. And I am wondering if this variable include liquid water, ice water and water vapor? Thanks. Yiyi

Dear Yiyi, Happy to help. A couple of questions. From where did you obtain the data? What format is it in? If it is in netcdf, please send the output of "ncdump -h" on the file. If it is in GRIB, please send the output of "wgrib -V" and only the first message is needed. I am fairly certain that the variable is "cloud liquid water" not "cloud water content", but I would need to see your output to be sure of what you are accessing. Best wishes,

Thanks for your help! I downloaded data from ESRL-PSD 20CRv2c webpage. The data link is ftp://ftp.cdc.noaa.gov/Datasets/20thC_ReanV2c/Monthlies/monolevel/cldwtr.eatm.mon.mean.nc. The data is in netcdf format, and the output of "ncdump -h" is following: float cldwtr(time, lat, lon) ; cldwtr:cell_methods = "time: mean (monthly from 6-hourly values)" ; cldwtr:long_name = "Monthly Cloud Water for entire atmosphere" ; cldwtr:units = "kg/m^2" ; cldwtr:precision = 4s ; cldwtr:GRIB_id = 76s ; cldwtr:GRIB_name = "C WAT" ; cldwtr:var_desc = "Cloud Water Content" ; cldwtr:dataset = "NOAA-CIRES 20th Century Reanalysis version 2c Monthly Averages" ; cldwtr:level_desc = "Entire Atmosphere Considered As a Single Layer" ; cldwtr:statistic = "Ensemble Mean" ; cldwtr:parent_stat = "Individual Obs" ; cldwtr:standard_name = "atmosphere_cloud_condensed_water_content" ; cldwtr:missing_value = -9.96921e+36f ; cldwtr:valid_range = 0.f, 6.5532f ; cldwtr:statistic_method = "Ensemble mean is calculated by averaging over all 56 ensemble members at each time step and then averaging mean over all time steps in a month" ; cldwtr:GridType = "Cylindrical Equidistant Projection Grid" ; cldwtr:datum = "wgs84" ; cldwtr:actual_range = 0.f, 0.3662097f ; Thanks, Yiyi

For land surface modeling, it is useful to know temporal frequency of reanalysis output (hourly, 3-hourly, etc...). Some have this mentioned, others make it hard to figure out. Is there a simple list somewhere? In particular, for some of my applications, temporal frequency of surface meteorology trumps spatial resolution. Not sure if any hourly output exists. Thanks. -ankur, UW-Madison, desai@aos.wisc.edu

Ankur, For offline land modeling, the MERRA-2 data includes our lowest model level data. This is on the terrain following coordinate, so the height AGL is variable, roughly 60m (but also provided). The advantages of lowest model level to land modeling are 1) this level includes a portion of the analysis increment, so the observations that help constrain the prognostic state fields and 2) it does not include the surface layer similarity parameterization interpolation that we use to determine fluxes for our land model. This should allow your land model and surface layer to develop independently of what the reanalyses systems do. In this way you can also compare back to the 2m meteorology and even diffusion coefficients that MERRA-2 provides.

MERRA-2 and CFSR provide 1 hourly frequency data, and CFSR does have higher spatial resolutions.

Ankur, Courtesy of Suru Saha of NCEP: "CFSR has hourly output for all land surface variables at full model resolution." I see from the links http://reanalyses.org/atmosphere/overview-current-reanalyses#CFSR one to NOMADS NCDC http://nomads.ncdc.noaa.gov/data.php?name=access#cfsr . Also, the link to NCAR should be useful http://rda.ucar.edu/#!lfd?nb=y&b=proj&v=NCEP%20Climate%20Forecast%20System%20Reanalysis . best wishes, gil

New student (not verified)

Sun, 11/08/2015 - 12:30

Hello,I am a student currently working on MERRA data, I need some links or detailed information for the parameters of Radiation ; SWGDN(Surface incident shortwave flux) & SWGNT(Surface net downward shortwave flux). What are the differences in both? How both can have effect on a body lying on Earth surface(probably near equator)? Your help will mean alot to me. Thanks in advance

In MERRA, there are two fundamental components of the radiation in shortwave. Downward, the solar radiation that reaches the surface through clouds and other atmospheric attenuation. Upward, the reflected radiation which depends on the downward radiation, and the albedo. The net shortwave is the downward minus the upward. It is described as net downward to provide the direction (or sign) of the data. For more information on MERRA data, you can refer to the File Specification document, and the budget appendix.


Wed, 10/28/2015 - 15:39

Hi all,

I am currently looking for gridded datasets of evapotranspiration (daily, ideally, at national/global scale). I have derived daily potential evapotranspiration estimates for the U.S. using the Hargreaves-Samani equation and the NCEP/NCAR Reanalysis dataset. However, I would like to compare these estimates to other datasets/derivations if possible. I know that Penman-Monteith is another standard method for calculating, but I have not attempted to calculate using P-M yet because of the greater number of inputs needed.

I am interested in learning more about the potential evaporation rate (W/m2) in the NCEP/NCAR Reanalysis dataset (Kalnay et al., 1996) - Does anyone have familiarity with this particular variable and/or know how it was derived? I know that actual evapotranspiration, potential evaporation, and potential evapotranspiration are often distinguised in the literature, so I also wanted to confirm what the potential evaporation rate in this dataset represents.

I also recently asked ESRL PSD about the potential evaporation and transpiration variables in the NOAA-CIRES 20CR dataset, and Gil Compo was very helpful in providing those references (Mahrt and Ek, 1984 - eq 9; Chen et al., 1996 - eqn 11).

If there are other national (or global) gridded datasets of daily evapotranspiration that you know of (eventually I will need to have PET in mm/day), please let me know.

Thank you in advance!
Best regards,
Meridith Fry (EPA/OPP)

There is a description of how CAPE is calculated (in the post processing) in ERA-Interim, in Part IV Physical processes, Chapter 5 Convection, Section 5.11, of the ERA-Interim model (IFS) documentation: https://software.ecmwf.int/wiki/display/IFS/CY31R1+Official+IFS+Documentation Also note that in ERA-Interim the CAPE at forecast step=0 is erroneously set to zero everywhere.

Fabio (not verified)

Fri, 10/09/2015 - 08:33

Hello, Which dataset do you think has the most accurate data for tropical central africa and has a time-step of at least 6-hours? Thanks

Anonymous (not verified)

Sat, 08/29/2015 - 09:34

I am looking at long term drought in the Southern Africa region and I would like to utilise ERA-INTERIM Reanalysis data. I am just not sure of the difference between Large-scale and total precipitation and which one is better to use for such analysis. I would also like to find out if there are other rainfall parameters that are available that I could use. Miya

In ERA-Interim, the total precipitation is the sum of the large-scale precipitation and convective precipitation. There is also snowfall, large-scale snowfall and convective snowfall available. All these fields are accumulated from the beginning of the forecast and the units are "m of water" or "m of water equivalent".

Yiyi (not verified)

Tue, 08/11/2015 - 16:29

Hello there, I found that in some reanalyses products, they would provide FORECAST and ANALYSIS products. Thus I am wondering that if forecast data is only model output and analysis data is produced by model and assimilation method. Can anyone help to explain the differences between them? Also, why some variables are only available in FORECAST but not in ANALYSIS? Thanks. Yiyi

Mr Al (not verified)

Wed, 07/15/2015 - 10:58

Hi, I too have just started to review some reanalysis surface pressure data and was hoping if someone could clarify a particular point. The data I'm looking at is from: http://iridl.ldeo.columbia.edu/SOURCES/.NOAA/.ESRL/.PSD/.rean20thcent/.V2/.six-hourly/.monolevel/.sfc/.pres/ and is 6hrly surface pressures at 2deg grid increments. My question is whether this is likely to be the spatial mean surface pressure across a grid square (and if so, would the coordinates represent the grid square centre) or whether the dataset should be considered as point data applicable to precise coordinates? thanks, Alan

Nigerian Student (not verified)

Mon, 07/13/2015 - 12:45

Hi guys. I'm pretty new to all this stuff. Can anyone please give me a simplified, step by step explanation of how to get and use data from any of these reanalysis websites? I'm a student at a University in Nigeria. I stumbled on this and I'd like to use data from this for my final dissertation. I'd really appreciate any help I can get. Thank you.

Dear Student, Please provide more information about what you are trying to accomplish. What variables? What time period? What time resolution? There is no single way to get data, and there are many possibilities. What software will be using to analyze and plot the data? Will any of the tools for subdaily, daily-average, or monthly average data available off of go.usa.gov/XTd work for you? If monthly mean data are what you are interested in, you may want to consider using the web-based reanalysis intercomparison tool or one of the other tools listed at http://reanalyses.org/atmosphere/tools. best wishes,

SSA (not verified)

Thu, 05/07/2015 - 21:29

Hello all, can anyone tell me the difference between the sea ice cover and sea ice extent. can i using the sea ice cover data in era iterim to explain how much the sea ice extent from one month to other month???

Alba Cid (not verified)

Tue, 04/28/2015 - 02:21

Hello! I'm using sea level pressure fields from ERA-interim reanalysis and from the 20th century reanalysis. There are some regions where SLP from 20CR is much lower than SLP from ERA-interim. For instance, around lon = 134ºW, lat = 14ºN, there are several drops in the SLP from 20CR not present in ERA-interim, with differences of about 20 mb. Does anybody know if there is a physical explanation for that or if they are outliers? Thank you very much in advance Alba

Dear Gilbert, thank you very much for your answer. Following your suggestion, I made a page at reanalyses.org where a time series comparison can be seen and where I specified the details about the data. The page is https://reanalyses.org/atmosphere/slp-drops-20cr-not-present-era-interim-reanalysis Thanks again! Alba

Alexander Cher… (not verified)

Tue, 02/24/2015 - 07:01

Does anyone know something about precipitation data in the ERA-20C reanalyses. Here is no such variable: http://apps.ecmwf.int/datasets/data/era20c_daily/ but there are new variables like: Total column rain water or Total column snow water (additionally to the standard Total column liquid/ice water). Is the following assumption correct: precipitation rate = Total column rain water + Total column snow water ? Is there any literature on this issue?

In ERA-20C there are various precipitation fields at the surface. There are no mean rate precipitation fields - the precipitation fields are accumulated from the beginning of the forecast for +step hours. To obtain the mean rate quantity you need need to divide by the number of seconds in step. The parameter names are: convective precipitation, convective snowfall, large-scale precipitation, large-scale snowfall, snowfall and total precipitation. Total column fields are vertically integrated quantities, not fluxes.

Patrick Boylan (not verified)

Wed, 02/11/2015 - 16:02

Does anyone know why the surface pressures from the surface level files (Sp) do not equal the surface pressure from the model level files (ln(Sp)). I have looked at several months in 2010 at all 4 times and step equal to 0.

The surface pressure (sp) archived at the surface is calculated from the model level data. However, the latter is archived by storing the spectral coefficients of ln(sp) (natural log of sp) whereas at the surface sp is archived as sp in grid point space. Differences between these two fields will arise because of the different packing errors associated with their GRIB packing and due to any differences in the methodologies used for transforming from the spectral representation to the grid point representation.


Thu, 02/05/2015 - 14:46

Does anyone know why the only surface humidity field from the ERA-Interim dataset is 2m dew point and not the more commonly used fields of specific humidity and/or relative humidity (which are also available as pressure level fields)?

The reason ERA-Interim only has 2m dew point temperature and not specific or relative humidity is historical. There is a FAQ about this, which describes how to calculate the 2m specific and relative humidity: http://www.ecmwf.int/en/does-era-40-dataset-contain-near-surface-humidity-data

What do you mean by "modern" humidity, Cathy? The synoptic surface atmospheric observations assimilated by ERA-Interm are reported as dewpoints and the synoptic message that contains the dew points also reports temperature and surface pressure, whose values are needed to compute other humidity variables. This was past practice, and is current practice. What is actually measured today depends on the instrument that is deployed, but the conversion to dewpoints for uniform reporting purposes should follow procedures laid down in WMO regulations. Dew point is no less modern than any other humidity variable. If reanalysis output is to be made available for only one humidity variable, it seems reasonable to me that one should use the same variable that is used in observational reports. Providing output for alternative humidity variables, using documented conversion formulae is certainly an option, though small numerical inconsistencies between the variables would arise from data encoding and interpolation.

Having specific humidity available in addition to Td (which is also important, as you say) is a valuable thing and makes certain calculations much easier as you are not combining T and q in one variable. To get the actual moisture in the air with Td, you have to also calculate the saturation q for the measured temperature so you have to read both Tsfc and Td and do the additional calculation. And, as many of the other reanalyses do not output dewpoint, it makes comparisons with them much easier if you have the same variable. In addition, the calculation of saturation q is a nonlinear function of T, so if you want to compare monthly means of other reanalyses, you have to calculate q (sfc) for each time step in a monthly average and then average (or compute Td from each time step of q and T in the other reanalyses). The non-linearity is also an issue for reanalyses which output ensemble means. Theoretically, Td ought to be calculated separately for each member and then averaged.

Agreed it is a bit of a nuisance to have to go through these steps. We went through them to compute monthly-mean specific and relative humidities in our 2010 ERA-Interim paper (doi: 10.1029/2009JD012442) and go through them regularly in preparing our annual contributions to the BAMS state-of-the climate article. But there are several possible variables - the HadISDH datset based on direct analysis of the synoptic data caters for seven different ones.

Sneha (not verified)

Tue, 12/30/2014 - 04:06

Hello, As per the documentation on MERRA, vertical resolution is 72 levels. However the v-wind given by the dataset has only 42 available levels eventhough the latitude and longitude resolutions remain the same. May I know what is the reason behind this?

The modeling and assimilation in the GEOS5 system that performs the MERRA reanalysis runs at 72 terrain following (eta) coordinates. These are not typically what most meteorological studies would favor, so the models eta coordinate is interpolated to 42 pressure levels. For certain variables it is beneficial to provide data in both coordinates. Please see the MERRA File Specification Document for details: http://gmao.gsfc.nasa.gov/research/merra/file_specifications.php

None of the "operational" analyses (JMA, ECMWF, NMC, NCEP, etc) are considered to be reanalyses, simply because in each of these the record contains discontinuities caused by periodic upgrades and changes in model, analysis methods, resolution, observation streams, and whatever else constituted the improvements from which all the operational systems have benefitted over time. In fact, a major rationale for doing reanalysis at all was the possibility to create long records of weather, using modern analysis/forecast systems, and without the operational discontinuities. Comparisons can certainly be made however, between operational systems and reanalyses, for what that might be worth.

Paul Poli (not verified)

Tue, 11/04/2014 - 01:53

Hello Mike, Glad to exchange with you via this forum. Generally reanalysis producers prefer to see only their latest and shinest product used, not just because it tends to give them 'better press', but rather for the immediate feedback they get from users regarding the new products, and so as to improve them next time around. And thus we collectively tend to instruct users 'please do not use earlier products for new research' (i.e. feel free to continue using it in ongoing projects, but do not start *new* work with these data-- please use instead the latest product). If we went along with this we would actually shorten the list above! That said, in the user survey conducted last year (responded by 2500+, the survey mostly reached ECMWF users though), there were 170 respondents who said they were using ERA-15. I doubt these have all been using ERA-15 for the same work over and over in the past nearly 20 years so this suggests there are probably new users of ERA-15, as there are new users of ERA-40. I'll add ERA-15 there, but we should also add FGGE if everybody found its most ancient data holdings. The whole series of reanalyses makes for a nice historical summary of atmospheric research improvements in the last 3 decades! But this shouldn't confuse users, so we should also probably mark with a special flag here those products that we think are, how to say it better, 'outdated'? Paul Poli


Mon, 11/03/2014 - 17:24

whither ERA-15 in this section?

I know ERA-15 is kind of 'OBE' (overtaken by events) relative to ERA-40 (I worked on both reanalyses)...However...

ERA-15 would be useful in assessing observing system variability and the benefits of observation data set 'cleansing'/improvement that were brought about from the initial comparisons of older obs to modern models in the pioneering reanalyses (NCEP/NCAR I NCEP/DOE II and ERA-15).

from an old reanalysis guy...

best /R Mike Fiorino

And the DAO reanalysis too! (Sorry, I think that was lost some time ago) ERA-15 is still available at the ECMWF data server (http://apps.ecmwf.int/datasets/data/era15/) and is also described at the Climate Data Guide (https://climatedataguide.ucar.edu/climate-data/era-15). As you say, it is probably best for marking the progression of reanalyses, but there are known issues that require new investigations to update to more recent data.

Alina (not verified)

Fri, 10/24/2014 - 11:36

Hello Everyone, I would need to know whether either or all of the following 3 reanalyses: - NCEP/NCAR - ERA-Interim - JRA25 assimilates data from the Tropical Moored Buoy System. I am particularly interested in data between 1979 and the present. Thank you in advance for any information on this subject!

Gary Barnes (not verified)

Tue, 09/16/2014 - 19:13

We are looking at Hurricane Felicia (2009) and have used the NCEP-NCAR reanalysis wind fields to look at the large scale environment. A colleague suggested that we look at CFSR. Does the CFSR reanalysis access different finer scale data? I understand it offers finer resolution but without finer resolution inputs we would necessarily see much difference? thanks, gary

Yes, CFSR uses finer scale observations than NCEP/NCAR Reanalysis. Automated aircraft reports: at cruising altitude 5-6 minutes between reports 737 cruises at 780 km/hr -> reports every 65-78 km see: http://amdar.noaa.gov/FAQ.html#bytes More reports during ascents/decents. So CFSR can get finer details from the aircraft data. Of course, commercial aircraft won't be flying near Hurricane Felicia. Satellite data: NCEP/NCAR Reanalysis uses temperature retrievals and cloud track winds. The temperature retrievals are reduced to a fixed horizontal and vertical resolution This kept the amount of satellite data more or less constant and prevented the satellite data from overwhelming the conventional data as the spatial and frequency resolution of the satellite data improved. The CFSR uses a direct assimilation of the satellite radiances as well at cloud track winds. Much satellite radiance data is thinned or "averaged" because it is of a higher resolution than the assimilation system. So CFSR can access the finer details from the satellite data. (BTW direct assimilation of the radiances is better than temperature retrievals.) However, the CFSR doesn't assimilate radiance data in cloudy "pixels" and hurricanes tend to be cloudy. So the CFSR has observations to support the finer resolution. However, too near the hurricane, the observation systems don't do a good job of sampling.

Thank you very much for your input. We pulled up the NCEP-NCAR reanalysis and compared it to CFSR. Looking at the 250 u and v components one can see that virtually all the larger scale features (> 5 deg latitude) are similar. The CFSR does generate many more fine scale features that I would be a little worried about, lots of closed contours covering length scales of 1-2 degrees. These tend to appear more frequently as one moves toward the equator. Of course subtle changes in the estimate will trigger a contour to appear or disappear. We see some nuances in the sub-tropical jet in CFSR but I think our interpretation of one vs . the other product would be close. Didn't see anything odd around the hurricane, at least at 250 hPa. Given the increased use of the aircraft data I imagine that the CFSR fields are better from Honolulu toward the west coast. regards, gary

Keith (not verified)

Wed, 09/10/2014 - 14:29

I'm curious whether sounding data from field campaigns (specifically the TEPPS cruise in 1997) would have been incorporated into the reanalyses. According to Serra and Houze (2000) pt I, the data was "transmitted via the Shipboard Envi- ronmental (data) Acquisition System (SEAS) to NOAA and were incorporated into the National En- vironmental Satellite, Data and Information Service model initialization." However, I have been unable to figure out if any (or all) of the reanalyses would have included this data.

Anonymous (not verified)

Mon, 08/25/2014 - 07:51

I've looked at the surface latent and sensible heat flux averaged for a polar cap (north of 70N). Both show odd seasonal cycles, with maximum values during winter, and minimum during summer. Is this mainly due to the sea ice misspecification in the 20CR dataset? The other fluxes (shortwave and longwave radiation at both surface and top of atmosphere) show more "normal" cycles (though somewhat higher values).

Dear Anonymous, As noted at https://reanalyses.org/atmosphere/inter-reanalysis-studies-0 perhaps what you are looking for is in the paper Lindsay, R., M. Wensnahan, A. Schweiger, J. Zhang, 2014: Evaluation of Seven Different Atmospheric Reanalysis Products in the Arctic*. J. Climate, 27, 2588–2606. doi: http://dx.doi.org/10.1175/JCLI-D-13-00014.1 Many additional seasonal intercomparison maps (many more than could be included in the publication) showing the median from the seven reanalyses and the deviations of each from the median are found at http://psc.apl.uw.edu/arctic_reanalyses. Both seasonal means and seasonal trends for the period 1981-2010 are shown. 34 different variables are included: surface fluxes, temperature, humidity, wind, precipitation, and pressure; layer heights and temperatures; and TOA radiative fluxes. best wishes, gil compo for Reanalyses.org

Anonymous (not verified)

Fri, 06/20/2014 - 06:56

I would like to know what is the vertical height levels for the reanalysis data like CFSR,MERRA, & ERA-I? Would there be different height levels possible? Also I want to know about Sigma levels ?

Dear User, The vertical levels for the post-processed data from the sources listed on http://reanalyses.org/atmosphere/overview-current-reanalyses under Data Access are described on each page. All of the datasets you list have multiple levels, some with more than 60 in the vertical. All of the datasets are available in self-describing formats that contain the level information. Sigma levels describe a vertical coordinate that is the level where the pressure is a specified fraction of the surface pressure. For example, sigma 0.995 is the level where the pressure is 0.995 times the surface pressure. If the surface pressure is 1000 hPa, then the pressure at the sigma level 0.995 is 995 hPa. Please reply back if you have additional questions.

andre (not verified)

Wed, 04/23/2014 - 11:16

Hello, What are the NCEP1 products that I should use to calculate Net Heat Flux at the air-sea interface? Qnet = SW- LW - LH - SH. where SW denotes net downward shortwave radiation, LW net upward longwave radiation, LH latent heat flux, and SH sensible heat flux I can find these products at http://www.esrl.noaa.gov/psd/data/gridded/data.ncep.reanalysis.surfaceflux.html Regarding the latent and sensible fluxes I don't have a problem (since there are only two in the NCEP list), but regarding the others I have several. For example: "Net longwave radiation", "Net shortwave radiation", "Upward longwave radiation flux" , "Downward solar radiation flux",... Hope you can help on this. Thanks!

Yes, that equation is correct and your abbreviations are correct. According to someone in our group who looked at your question "I think they should use the net shortwave and net longwave (if there are also upward and downward fluxes). Note that the shortwave from reanalysis has some real problems (it depends a lot on the cloud cover, which is not simulated well in the AGCM used in the reanalsyis). We did a paper on this a while back: Scott, J. D. and M. A. Alexander, 1999: Net shortwave fluxes over the ocean. J. Phys. Oceanogr., 29, 3167-3174. "

SA (not verified)

Tue, 04/01/2014 - 03:07

Hi everyone, I am analyzing hourly rainfall records in many regions of the world. However, I would like to know which reanalysis data is suitable to make some comparison. Thanks in advance for any suggestions :D

Dear SA, What you are asking turns out to be a research question. Several papers have looked into this. See a list at http://reanalyses.org/atmosphere/observational-studies and http://reanalyses.org/atmosphere/inter-reanalysis-studies-0#Precipitation Please report back with what you find. best wishes, gil compo

I can't say anything specific, since your comment has very little detail about where and when you are working. One thing that will be universal when considering reanalysis precipitation is that the quantity is derived from the background model forecast of the assimilation process. So, while the environment that produces precipitation is affected by the observational analysis, precipitation relies on the model physics, and has significant uncertainty. If observations of precipitation are available, you'll want to use those, at the least, in a comparison with the reanalysis data. Some reanalyses produce 1 hourly data, and some assimilate rainfall (see the North American Regional Reanalysis). Biases and uncertainty in reanalysis precipitation depends greatly on where and when you are looking. The papers listed in the pages Gil provided are a start, but be sure to follow those that they cite as well.

Mehdi (not verified)

Mon, 03/03/2014 - 06:09

Hi everyone, I'm trying to understand the vertical wind velocities in the ERA Interim data set in model levels ("eta" hybrid coordinates). According to the netCDF metadata, the provided vertical velocity is in Pa.s-1 which corresponds to a vertical velocity in pressure coordinates. Hence the following question : do we have d(p)/dt instead of d(eta)/dt in the data in model levels ? Thanks for any piece of advice.

Alima (not verified)

Thu, 02/27/2014 - 22:25

Hi everyone, I am looking for the land sea mask for 288*145, please let me know where I can have access it. Thanks in advance

Add new comment

This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.