ICOADS Web information page (Monday, 28-Apr-2014 17:59:15 UTC):

Landmask files used for quality control

Two related landmasks, at 1° latitude x 1° longitude and 2°x2° spatial resolution, are currently used to detect marine reports falling over land areas for ICOADS, with usage of these landmasks extending back to COADS Release 1 (Slutz et al. 1985). The 2° landmask was created from the 1° landmask, with a few modifications.

The 1° landmask (LNDLMT) was obtained from NCDC in 1981 (Figure 1), together with climatological QC limits (see this webpage). Together with the limits, LNDLMT is used within the "NCDC-QC." This quality control processing is defined by a complex flowchart in supp. J of Slutz et al. (1985). It marks 1° boxes as land or ocean. There is no intermediate "coastal" value to indicate that a box contains both land and ocean.

Since LNDLMT was developed so many years ago, the methods used probably are no longer documented at NCDC. However, LNDLMT generally followed the approach stated in NCDC (1968) that "ship positions had to be in ocean, sea, or lake areas." There are some exceptions, however. For example, LNDLMT has the Mississippi River marked as water, as opposed to some other large navigable rivers. Also, LNDLMT has the Ross Ice Shelf marked as land.

The primary ICOADS 2° landmask (LLN2F1) was created using the four constituent LNDLMT 1° values for each 2° box as follows:
  1. if all four 1° boxes were ocean: 2° = ocean
  2. if all four 1° boxes were land: 2° = land
  3. otherwise: 2° = coastal
In addition, the Mississippi River was modified in LLN2F1 to classify it as land, and a box over part of Florida to classify it as coast. The resulting LLN2F1 (Figure 2), also shown in supp. G of Slutz et al. (1985), distinguishes 4,375 boxes over land ("landlocked") from the remaining 11,827 boxes over coastal (881) or ocean (10,946) areas.

Applicable to currently available observational (IMMA) and monthly summary (MSG) products, LNDLMT and LLN2F1 are used in ICOADS processing, including to set QC flags ZNC and LZ (stored in the Icoads attm), as follows:
  1. If a report falls over a land box as indicated by LNDLMT, it is flagged ZNC=7 (M; erroneous) by NCDC-QC. Note that ZNC=7 can also be set by NCDC-QC for other reasons, such as missing hour or day (see item 4).
  2. If a report falls over a land box as indicated by LLN2F1, the landlocked flag (LZ) is set LZ=1 (report over land).
  3. During creation of the final IMMA user product, reports falling over land according to the 2° check (LZ=1) are not output. Thus reports falling over ocean or coastal boxes according to LLN2F1 are accepted in that product (however landlocked and other suspect reports are retained in the "intermediate" IMMA product, from which the final product is constructed).
  4. MSG monthly summary statistics are calculated from the final IMMA product, thus the 2° landlocked check is pre-applied. In addition, from 1950 forward 2° and 1° MSG statistics are not calculated for reports falling over land according to the 1° check (ZNC=7). Prior to 1950 the ZNC flag is ignored during calculation of MSG because otherwise excessive amounts of data were rejected due to missing hour (primarily daily pre-1860 observations from the Maury Collection; details here).

The settings for ZNC and LZ may conflict. This can happen because coastal 2° boxes are considered equivalent to ocean boxes in setting LZ. Thus by definition at least one of the constituent 1° boxes within a coastal 2° box is over land according to the 1° check. Or, the Mississippi and Florida modifications made to LLN2F1 introduce some differences (Figure 3). Finally, ZNC=7 may arise from other data problems detected by NCDC-QC (as noted above).

Using North America for example, Figure 3 also illustrates some inaccuracies in both LNDLMT and LLN2F1. Our eventual aim is to replace these outdated landmasks with more modernized versions, e.g. a 1° landmask (lstags) developed several years ago by Dick Reynolds (Figure 4). Figure 5 shows differences between it and LNDLMT.

A consideration for the use of landmasks for earlier historical ship data is that latitudes and longitudes from early shipboard observers and of ports may be greatly reduced in accuracy from values available today. Hence use of a more forgiving (e.g., 2°) landmask may still be desirable for early data.

For easier access, we reformatted the three landmask files LNDLMT, LLN2F1, and lstags (shown in Figures 1, 2, and 4) into the following data (and control; ctl) files in the Grid Analysis and Display System (GrADS) format:

LNDLMT LLN2F1 lstags.onedeg.dat
LNDLMT.ctl LLN2F1.ctl lstags.onedeg.ctl

References

NCDC (National Climatic Data Center), 1968: TDF-11 Reference Manual. NCDC, Asheville, NC, 138 pp. (pdf; 2.5MB)

Slutz, R.J., S.J. Lubker, J.D. Hiscox, S.D. Woodruff, R.L. Jenne, D.H. Joseph, P.M. Steurer, and J.D. Elms, 1985: Comprehensive Ocean-Atmosphere Data Set; Release 1. NOAA Environmental Research Laboratories, Climate Research Program, Boulder, CO, 268 pp. (NTIS PB86-105723).

[Documentation and Software][Links to additional]

U.S. National Oceanic and Atmospheric Administration hosts the icoads website privacy disclaimer
 Document maintained by icoads@noaa.gov
Updated: Apr 28, 2014 17:59:15 UTC
http://icoads.noaa.gov/mask.html