=============================================================================== International Comprehensive Ocean-Atmosphere Data Set (ICOADS): Release 2.4 General Information about Statistics 23 September 2007 ===================================================================== Document Revision Information (previous version: 27 February 2004): Updates for Release 2.4. ------------------------------------------------------------------------------- This document gives a general description of the sets of monthly summary statistics available from ICOADS, presently covering 1800-May 2007. The summaries were calculated for each year and month over the entire period-of-record using 2-degree latitude x 2-degree longitude boxes. Starting in 1960, when spatial coverage improves, they were also calculated using 1x1-degree boxes. The summaries were based on Release 2.4 Long Marine Reports Fixed-length (LMRF) for that period. Coverage was extremely sparse, and thus statistics were not calculated, prior to 1800. Twenty-two observed and derived variables (Table 1) were summarized for each year-month. For each variable, 10 statistics (Table 2) were calculated. The statistics were broken up into "groups" of four variables per group, in Monthly Summary Group (MSG) format (version 1, designated as MSG1). Each group (numbered 3-7, plus 9) contains the 10 statistics for each of the four variables (Table 3). For technical details for the MSG format, see , and for quality control ("trimming") and data selection rules used to construct the products, please refer to . In the following, each 2-degree and 1-degree product acronym is defined, accompanied by a brief product synopsis. Also, the temporal coverage currently available for each product is listed. Finally, we answer a list of frequently asked questions. Product, domain, and group abbreviations ---------------------------------------- std: Standard statistics The standard statistics were derived using 3.5 standard deviation (sigma) trimming limits and were limited as nearly as possible to observations taken from ships (where identifiable). enh: Enhanced statistics The enhanced statistics were derived using 4.5 sigma trimming limits, so as to accommodate more extreme climate event, and using a broad collection of marine observations from ships, buoys, fishing fleet vessels, and near-surface oceanographic profile temperatures. glo: Global domain Domain limits: global in latitude and longitude. 1-degree boxes whose boundaries fall on units of latitude and longitude (e.g., 0-1E, 89-90N). All the 2-degree products also utilize a global domain, with box boundaries falling on even degrees of latitude and longitude (e.g., 0-2E, 88-90N). equ: Equatorial domain Domain limits: equatorial latitude band 10.5N to 10.5S; global with respect to longitude. 1-degree boxes shifted half a degree in latitude (only) in comparison to the global domain, such that the center-latitude of the central row of boxes is the equator (e.g., 0-1E, 0.5S-0.5N). g_: Group numbering: g1, g2, g3, g4, g5, g6, g7, g8, g9. b_: Box size: 1 for 1-degree, or 2 for 2-degree, lat/lon boxes. 2- and 1-degree products in MSG format -------------------------------------- MSG: Monthly Summary Groups Six groups (numbered 3-7, plus 9), each containing four variables x 10 statistics. MSG.b2.std.glo MSG.b1.std.glo 2- or 1-degree standard statistics for the global domain. MSG.b2.enh.glo MSG.b1.enh.glo 2- or 1-degree enhanced statistics for the global domain. MSG.b1.std.equ 1-degree standard statistics for the equatorial domain. MSG.b1.enh.equ 1-degree enhanced statistics for the equatorial domain. Temporal coverage of MSG products --------------------------------- MSG.std.glo.b2 [1800-----------------------------------------------2007] MSG.enh.glo.b2 [1800-----------------------------------------------2007] MSG.std.glo.b1 [1960-----------------------2007] MSG.enh.glo.b1 [1960-----------------------2007] MSG.std.equ.b1 [1960-----------------------2007] MSG.enh.equ.b1 [1960-----------------------2007] Frequently asked questions about the statistics ----------------------------------------------- 1) Are the summaries interpolated or analyzed? No. No interpolation or extrapolation is done to fill in temporal or spatial data gaps. Only boxes that have observations will have data records. No analysis is done to smooth or modify the statistics. 2) Are the summaries adjusted for changes in observing practices? No. The available summaries incorporate no adjustments for changes in observing practices, such as changes from bucket to engine room intake measurements of sea surface temperature, or changes from estimated (Beaufort) to anemometer measured wind speeds. 3) How were the statistics (e.g., mean, median, sextiles) calculated? Refer to Release 1, supp. A, and to . If there is a single observation in a box, that value is used to calculate the statistics (e.g., the mean is that value divided by one). 4) For a given variable, is the number of observations as provided in MSG format used to calculate all the statistics? Yes, for all statistics except the mean day-of-month and the fraction of observations in daylight. For those two statistics the numbers of observations may be fewer (and are not stored in MSG format). This is because day and hour may be missing in LMRF (but latitude, longitude, year, and month may not). However, any differences are generally minor since most data sources contain extant day and hour, with the exception of significant amounts of (Maury Collection) data prior to about 1861. 5) Why is the lower-left (SW) corner of the 2- or 1-degree box included in the MSG format (BLO and BLA), rather than the box center? The box corner should NOT be interpreted as a grid-point center from an analyzed dataset (ref. question 1). The weighted mean position of all observations within a box can be obtained from the corner coordinates plus the longitude and latitude offset statistics (x and y) for each variable. In a well and uniformly sampled box the weighted mean position of all observations will be the box center. However, in a sparsely sampled box or box that overlaps land masses the mean sample position may not be the box center. 6) How is the mean sample position within a box obtained? By: mean longitude = BLO + x, and mean latitude = BLA + y. Note that longitude is always measured in east coordinates. Because the x values can range up to two degrees (one degree) for 2-degree (1-degree) boxes, the resultant range of mean longitude is 0-360E. 7) How were the derived variables computed? The derived variables were computed using individual observations of the constituent variables exactly as shown in Table 1. For example, the mean of D results from calculating (S - A) for each individual (e.g., ship or buoy) report, and then averaging those values. Refer to Release 1, supp. A for computational details, including references to the formulae (available in {profs_short} software) used to calculate specific and relative humidities. 8) How were the pseudo-fluxes calculated? We have chosen not to apply any constants or adjustments to variables when calculating heat and momentum flux parameters. This is because the constants for the bulk formulae and adjustments for atmospheric stability of the boundary layer are not universally agreed upon. Therefore, the heat and momentum flux parameters within ICOADS are actually psuedo-fluxes. Thus for example we do not define or use factors such as density, specific heat of air, etc. 9) If there are 22 variables, why does Table 1 list 23? A large numerical range is possible for wind speed cubed (B = W**3). Owing to MSG format limitations, it was necessary to separate this variable into high- and low-resolution representations, i.e., data increments of 0.5 or 5 m**3/s**3 for B1 and B2. B2 can store the full data range, but B1 offers better resolution. Statistics (s1, s3, s5, m, and s) for the low-resolution field B2 are always available, if there were data. But the B1 statistics provide better data resolution unless they are missing because a given statistic can only be represented in B2. Table 1. Variables in MSG format. Each variable is abbreviated with an upper-case letter (followed "1" or "2" for variable W**3, which is available at two resolutions). Additional notation is as follows: @ is used as a plain text abbreviation for the degree symbol, and m/s indicates meters per second. "Units" gives the smallest increment of the data that has been encoded. ------------------------------------------------------------------------------- No. Abbrev. Variable Units =============================================================================== "Observed" ----------------------------------- 1 S sea surface temperature 0.01 @C 2 A air temperature 0.01 @C 3 W scalar wind 0.01 m/s 4 U vector wind eastward component 0.01 m/s 5 V vector wind northward component 0.01 m/s 6 P sea level pressure 0.01 hPa 7 C total cloudiness 0.1 okta 8 Q specific humidity 0.01 g/kg Derived ----------------------------------- 9 R relative humidity 0.1 % 10 D S - A = sea-air temp. diff. 0.01 @C 11 E (S - A)W 0.1 @C m/s 12 F QS - Q = (sat. Q at S) - Q 0.01 g/kg 13 G FW = (QS - Q)W = (evap. param.) 0.1 g/kg m/s 14 X WU (wind stress 0.1 m**2/s**2 15 Y WV parameters) 0.1 m**2/s**2 16 I UA (sensible-heat--transport 0.1 @C m/s 17 J VA parameters) 0.1 @C m/s 18 K UQ (latent-heat--transport 0.1 g/kg m/s 19 L VQ parameters) 0.1 g/kg m/s 20 M FU 0.1 g/kg m/s 21 N FV 0.1 g/kg m/s 22 B1 B = W**3 (high-resolution) 0.5 m**3/s**3 23 B2 B = W**3 (low-resolution) 5 m**3/s**3 ------------------------------------------------------------------------------- Table 2. Statistics available in MSG format. The statistics are abbreviated by lower-case letters, or a lower-case letter followed by a digit. Additional notation is as follows: @ is used as a plain text abbreviation for the degree symbol. "Units" (with reference to Table 1, if the units depend on the variable) gives the smallest increment of the data that has been encoded. ------------------------------------------------------------------------------- Abbr. Statistic Units =============================================================================== s1 1/6 sextile (est. of m - 1s) Table 1 s3 3/6 sextile (the median) Table 1 s5 5/6 sextile (est. of m + 1s) Table 1 m mean Table 1 n number of observations 1 s standard deviation** Table 1 d mean day-of-month of obs. 2 dy ht fraction of obs. in daylight 0.1 x mean longitude of obs.: 2-degree box size 0.2@ 1-degree box size 0.1@ y mean latitude of obs.: 2-degree box size 0.2@ 1-degree box size 0.1@ ------------------------------------------------------------------------------- * Note: a standard deviation estimate is computable from the sextiles, defined as e = (s5 - s1)/2. ----------- Table 3. Variables included in each trimmed (3-7, 9) group in the MSG format (group 8 is undefined for MSG and refers to the previously available MSTG format). Each group contains four variables (from Table 1) and ten statistics (Table 2) for each variable. ------------------------------------------------------------------------------- Monthly Summary Groups (MSG) 3 4 5 6 7 9 =============================================================================== Sea sfc. temp. (S) Scalar wind (W) Total cloud. (C) D=S - A I=UA M=FU Air temp. (A) Wind U-comp.(U) R E=(S - A)W J=VA N=FV Specific hum. (Q) Wind V-comp.(V) X=WU F=QS - Q K=UQ B1=(W**3)* Relative hum. (R) SLP (P) Y=WV G=FW L=VQ B2=(W**3)* ------------------------------------------------------------------------------- * B1 and B2 are high- and low-resolution representations of B = W**3 (i.e., data increments of 0.5 or 5 m**3/s**3). ----------