From ao@GFDL.GOV Tue Dec 20 12:07:57 1994 Received: from gfdl.GOV by niwot.scd.ucar.EDU (NCAR-local/ NCAR Mail Server 04/10/90) id MAA17004; Tue, 20 Dec 1994 12:07:55 -0700 Received: from ao (ao.GFDL.GOV) by gfdl.GOV (4.1/SMI-4.0) id AA13312; Tue, 20 Dec 94 14:08:04 EST From: ao@GFDL.GOV (A. Oort) Received: by ao (4.0/GFDL-Client) id AA02674; Tue, 20 Dec 94 14:08:28 EST Date: Tue, 20 Dec 94 14:08:28 EST Message-Id: <9412201908.AA02674@ao> To: joseph@niwot.scd.ucar.EDU Subject: write-up GFDL Tape Library Cc: ao@GFDL.GOV Status: RO Dear Dennis: As promised, I am sending you here a copy of the write-up on the "GFDL Atmospheric Circulation Tape Library, 1958-1989" (dated June 21, 1994). I adjusted some of the expressions in the legend of Figure 1, since "greater than", etc. did not show up. However, some things like overbars (indicating time averages) do not appear. Could you add them yourself? I am sending out to you some copies of the original write-up together with the statistics. Best wishes, Bram "GFDL Atmospheric Circulation Tape Library, 1958-1989" (June 21, 1994;updated December 20, 1994) This tape library contains monthly-mean gridded analyses on a 73x73 latitude-longitude grid spanning the period May 1958 through December 1989. The files are written in ieee f77 format, i.e., 4 byte real values with a 4 byte control word at the beginning and end of each record. There are up to 12 months contained in each file, and each month is stored consecutively within a file. See the tape cartridge for a listing of which months are in a particular file. Each month's worth of data may be read with the following FORTRAN code: real a(73,73) integer iyear, imonth character ahour*12, alvl*6, name*4 do 10 ilevel=1, 11 do 10 iquan=1, 22 read (8) iyear, imonth, ahour, alvl, name, a 10 continue where: iyear is the integer year (1958-1989) imonth is the integer month (1-12) ahour is the character hour (see below) alvl is the character pressure level (1000 mb - 50 mb) name is the character quantity (u, v, etc.) a is the real data array The above do loop variable `ilevel' indexes the 11 pressure levels, which are, in order: 1000, 950, 900, 850, 700, 500, 400, 300, 200, 100, and 50 millibars. The do loop variable `iquan' indexes the 22 analyzed quantities, which are, in order: u, v, T, z, q, rh, , , , , , , u'v', u'T', u'z', u'q', v'T', v'z', v'q', T'q', div v, and omega. The units of the quantities are as follows: the zonal (eastward) wind component u is in m s-1, meridional (northward) wind component v in m s-1, temperature in degrees C, geopotential height z in m, specific humidity q in g kg-1, relative humidity rh in %, horizontal divergence of the wind div v in 10-6 s-1, and the vertical (pressure) velocity omega in 10-4 mb s-1. Variance and covariance terms use the same units as implied by the basic quantities from which they are derived. The first array index of the data array `a' indexes latitude, beginning at 90S and progressing northward in 2.5 degree increments to 90N. The second array index of the data array `a' indexes longitude, beginning at 180E and progressing WESTWARD at 5 degree increments to 180W. (The date line is repeated, but the value at 180W is not exactly equal to that given at 180E.) The time indicator is a character variable indicating the times (GMT) of observation used in the basic rawinsonde records. It is 00z from May 1958 through April 1968, 00z and 12z from May 1968 through January 1975, and 00z, 06z, 12z, and 18z thereafter. NOTES: 1) The missing value indicator is 1010. 2) All moisture related quantities (q, rh, , , u'q', v'q', T'q') are given the missing value above 300 mb. 3) Five quantities are not available at the beginning of the record, and are given the missing indicator for those months. Those quantities, and the first month for which they have actual values, are: rh (May 1973) rh'2 (May 1973) u'z' (May 1963) v'z' (May 1963) T'q' (May 1973) 4) For the first five years of data (May 1958-April 1963) all data at all levels are missing south of 15S. This is due to the lack of rawinsonde data in the southern hemisphere. 5) In general, the analyses near the earth's surface (at 1000, 950 and 900 mb) tend to be somewhat less reliable than those at the standard pressure levels of 850 mb and above because the basic rawinsonde time series are often incomplete due to missing 1000 mb reports. Above 100 mb the analyses are again less reliable because of fewer reports. An idea of the station distribution during the May 1973-December 1989 period can be obtained from studying Fig. 1 and Table 1. The station distributions for the earlier period May 1958-April 1973 are discussed extensively in Oort (1983); they are comparable with those shown here, except in the Southern Hemisphere where the station distribution has improved markedly since 1973. 6) In the data sparse regions especially over the southern oceans, the analysis program has filled in interpolated values as described in Oort (1983). These values are, of course, not very reliable so that it is good to also consider the station distributions when using the analyses. In fact, in a few cases small negative values are extrapolated into no-data regions for positive definite quantities like the variances. 7) After copying the data, an error was discovered by our colleagues at Atmospheric Environmental Research, Inc., and MIT in one of the fields. The error is in v at 1000 mb for May 1978. At 75W most of the longitude values (51 points) are bad. The best remedy seems to interpolate between 70W and 80W for v1000 during May 1978. 8) For further REFERENCE on the basic input data and the analysis schemes used, see: Oort, A.H., 1983: Global Atmospheric Circulation Statistics, 1958-1973, NOAA Prof. Paper No. 14, 180 pp. (with 47 microfiches) (updated). (Available on request from author, Geophysical Fluid Dynamics Laboratory/NOAA, P.O. Box 308, Princeton, NJ 08542.) Fig. 1. Map of the station distribution at 500 mb used in the analysis of the wind components during the May 1973-December 1989 period (200 months). There are 760 stations marked by an asterisk(*) with (nmo.le.200 and nmo.ge.150), where nmo = number of "good" months (ndays.ge.10 in a month) during the period; 327 stations marked by a zero ("0") with nmo.lt.150 and nmo.ge.50); 245 stations marked by the letter "A" with (nmo.lt.50 and nmo.ge.10).