This is ECMWF-users' newsletter #29. To get copies of previous newsletters, use anonymous FTP to ncardata.ucar.edu [128.117.8.111] and change directory to datasets/ds111.2/news. Or, use a WWW browser and open http://www.ucar.edu/dss/datasets/ds111.2.html and select "news". 4 June 1997 CONTENTS: o Basic Consolidated Dataset (ds111.2) updated through March 1997 o Reanalysis problems o Reanalysis documents o Other reanalysis products o Software changes for Fortran 90 ============================================================================ Basic Consolidated Dataset (ds111.2) update The Basic Consolidated Dataset (ds111.2) has been updated with the first quarter of 1997 data. You can obtain the listing of the volume serial numbers containing the data from the file datasets/ds111.2/MASTER via anonymous FTP to ncardata.ucar.edu. Or, use a WWW browser and open the URL http://www.ucar.edu/dss/datasets/ds111.2.html. This file is also on the Mass Store under the name /DSS/D/DS1112. The data files (all in the /DSS/ directory on the MSS) are: Y38316 1997JAN Y38317 1997FEB Y38318 1997MAR ============================================================================ Reanalysis problems September 1990 surface data appears to be included twice; I will fix this eventually. Rex Gibson of ECMWF recently visited NCAR and gave a seminar on the ERA (ECMWF Re-Analysis) project. Here are some notes based on his seminar, and on his answers to my later questions. The major problem discovered with the reanalyses is that the analysis produces systematic pressure increments in one direction, leading to cooling at the lowest levels. Since relative humidity is analyzed, this leads to constant drying near the ground. This has caused a desertification effect in the analyses particularly in the western Amazon. Evaporation and precipitation are affected. This problem was corrected late in the reanalysis project; data from September 1980 through December 1986 are affected. Fluxes are also affected. The information in the research manuals distributed for the WCRP/TOGA dataset (ds111.x) is largely applicable to the reanalyses, except for the physical parameterization. A new set of manuals is "in the pipeline" and as soon as it is available we (NCAR) will get copies. A comparison between the reanalyses and operational analyses (WCRP/TOGA) shows that reanalyses are better up until the operational analyses went to higher (T213) resolution (17 September 1991). Future plans are for "ERA-40" -- 40 years of reanalyses. However they are very early in the planning stages and don't expect to even begin production (if it even comes to pass) until around 1999. Rex Gibson can be reached at err@ecmwf.int, or you can email the entire Re-Analysis group at era@ecmwf.int, if you have further questions on the model physics or details. ============================================================================ Reanalysis documentation We have a draft copy of the "ERA Description" document, which is similar to the "Technical Attachment" document that is available for the operational analyses. Unfortunately we don't have permission to distribute it at this time, but I can use it to answer questions until we get the final copy (later this month, I hope). The document lists changes to the model, problem dates, and some background information on input data, preprocessing, data assimilation, and production. When the final documents are prepared, Rex Gibson hopes to have them online at either the ECMWF web site, or at the NOAA NIC site. I will distribute the address when they are available. ============================================================================ Other reanalysis products We are still awaiting the higher resolution grids, and the monthly means, from ECMWF. There is a temporary hold while ECMWF and NCAR re-evaluate our agreement on data exchange. Todd Ringler of CSU has computed monthly means and has offered to make them available. I will add them to our reanalysis datasets as "working" (unofficial) VSNs until we receive the official means from ECMWF. Anyone who wants to use these means should contact me for the MSS VSN names. I hope to make them available in a few weeks. ============================================================================ Software changes for Fortran 90 With the decommissioning of shavano, many researchers are switching to Fortran 90 as Fortran 77 is not available on all machines. The ECMWF-supplied software for Cray usage uses conventions which are not legal in Fortran 90 in the vectorization routine. For the present, I have created a version of the Cray-specific code which does not vectorize, which compiles under Fortran 90. It might run a tiny bit slower, but not enough to make any difference in the tests I made. If you use the version of subs_cray.f which is on our ftp machine, you will need to retrieve a new copy. If you compile your code with the MSS version, it has been updated with the non-vectorizing version. Other changes you will need to make to your software under F90 are to add these lines to the beginning of your script: source /opt/modules/modules/init/csh module load modules PrgEnv and to compile under f90: f90 tsrc.f -L/usr/local/lib -L/usr/lib -lmss -lncarg \ -lncarg_gks -lncarg_c -lncaro -lncarm -lX11 instead of the equivalent cft77 and segldr calls. For general Fortran 90 migration questions on NCAR computers, please contact the SCD consultants. ============================================================================ If you would like to be added to or taken off this mailing list, please contact me. If you are interested in ordering any ECMWF/TOGA data, or any other data from NCAR's archives, please contact me. I can be reached at ilana@ncar.ucar.edu or by phone at (303)497-1214. Ilana Stern Data Support Section National Center for Atmospheric Research