COADS Release 2 Data and Metadata Enhancements for Improvements of Marine Surface Flux Fields

S.D. Woodruff¹, H.F. Diaz¹, J.D. Elms², and S.J. Worley³

¹ Climate Diagnostics Center; NOAA/ERL (R/E/CD); 325 Broadway; Boulder, CO 80303 USA

² National Climatic Data Center; NOAA/NESDIS (E/CC22); 151 Patton Avenue; Asheville, NC 28801 USA

³ National Center for Atmospheric Research; Data Support Section; P.O. Box 3000; Boulder, CO 80307 USA
 

Received 25 April 1997; accepted 11 September 1997
 
 
 

Abstract. The Comprehensive Ocean-Atmosphere Data Set (COADS) has been updated through a cooperative U.S. project since 1981, including vital international contributions. Quality controlled marine surface observations from ships have been supplemented in more recent years to include moored environmental buoys, drifting buoys, and near-surface measurements from oceanographic profiles. The data set now covers 142 years, 1854-1995. Monthly statistics of pseudo-fluxes and basic marine variables are calculated for each year using observed data falling within 2° latitude × 2° longitude boxes (1°×1° summaries are also available for 1960-93). Enhancements in data and metadata planned by the year 2000 as part of COADS Release 2 (~1820-1997) will concentrate on the basic observational records. In addition to new data sources, which will augment flux estimates through expanded coverage, planned enhancements include: a) usage of selected metadata from WMO Pub. No. 47 (ship instrumentation history) to improve the observational records back to about 1973; b) improvements in the reliability of the wind speed ("estimated/measured") indicator; and c) bias adjustments of wind speed Beaufort estimates and anemometer measurements.

Correspondence to: S. Woodruff

1 Introduction

Ship observations are our only major source of early historical data to estimate oceanic surface energy fluxes. Satellites now provide data with increased spatial and temporal coverage. However, in situ measurements remain critically important (Rossby et al., 1995)-they are used as a baseline to remove biases in satellite data and provide measurements of variables not presently estimated from satellite instruments, e.g. surface barometric pressure and air temperature.

The Comprehensive Ocean-Atmosphere Data Set (COADS) is the most extensive and widely used set of surface marine data currently available for the world ocean back to 1854, the year marking the beginning of an internationally organized system for recording shipboard meteorological observations. COADS presently extends through 1995, with the historical marine observations taken by such "ships-of-opportunity" (including merchant and naval vessels) supplemented in COADS since around the turn of the century by near-surface sea temperatures derived from the uppermost levels of oceanographic profiles, and since around the mid-1970s by measurements from increasingly extensive arrays of drifting and moored buoys.

The basic observed variables in COADS include sea surface and air temperatures, wind, humidity (wet bulb or dew point temperature), barometric pressure, cloudiness, weather, and wave and swell fields. The observations have been quality controlled, and monthly summary statistics calculated for 2° latitude × 2° longitude boxes (available for the entire 1854-1995 period), and for 1° latitude × 1° longitude boxes (presently available only for 1960-93). The 2° (1°) summaries include eight observed variables, plus 11 (14) derived variables, including heat and momentum pseudo-fluxes (neglecting transfer coefficients).

Section 2 of this paper provides a brief historical overview of marine surface observing and the development of COADS, which represents a continuing cooperative U.S. effort between the National Oceanic and Atmospheric Administration (NOAA) and the National Center for Atmospheric Research (NCAR). Section 3 discusses progress on data digitization and acquisition, and other specific data and metadata enhancements planned for upcoming major COADS updates. These efforts will result in expanding the spatial and temporal coverage (~1820-1997) by adding new data, improving quality controls, and better observational and statistical products including specific changes targeted to improve the availability and usability of oceanic flux quantities.
 

2 Historical overview and recent COADS updates

International agreement to systematically record weather observations in ships' logbooks, including measurements of temperatures and barometric pressure, and estimates of wind direction and wind speed based on the Beaufort force, was only obtained starting in 1854 (Maury, 1854). Eventually, the U.S. and other maritime nations developed programs to archive and digitize ship logbook data, whose international exchange was not formalized until WMO (1963) Resolution 35 (Cg-IV). Similarly, about 1921 an international code for radio weather messages from merchant ships was introduced (UK Meteorological Office, 1977), but digital data from the Global Telecommunication System (GTS) or its precursors were digitally archived only much later, with data from U.S. sources available only starting in 1966.

Prior to WMO Resolution 35, copies of many of the available digital collections of foreign keyed logbook data were obtained by NOAA's National Climatic Data Center (NCDC), e.g., through bilateral agreements. To facilitate data processing on punched card "decks" including automated quality control (QC), NCDC consolidated the foreign data, plus available U.S. data, into a uniformly formatted database referred to as Tape Data Family-11 (TDF-11) (NCDC, 1968). The TDF-11 data were used to produce the revised U.S. Navy Marine Climatic Atlases of the World (e.g., U.S. Navy, 1981).

With increasing recognition of the potential value of long-term marine records for climate research (Fletcher et al., 1981), inter-agency cooperation began in January 1981, when NOAA's Environmental Research Laboratories (ERL), and the Cooperative Institute for Research in Environmental Sciences (CIRES), planned with NCDC to combine additional marine data sources that had become available since TDF-11 into an easily used data set. NCAR soon joined in detailed planning and execution of this continuing project, which has produced COADS.
 

2.1 COADS Release 1 (1854-1979)

The consolidated TDF-11 data became the main core of data for 1854-1979 used to complete COADS Release 1 (Slutz et al., 1985; Woodruff et al., 1987). The TDF-11 data were augmented by ship data from a variety of other sources, including GTS data since 1966, and additional in situ data from buoys and sea temperatures taken from oceanographic profiles. In Release 1 processing the individual observations were quality controlled ("trimmed"), and 14 monthly statistics were calculated for 8 observed and 11 derived variables including pseudo-fluxes, using 2° latitude × 2° longitude boxes. For many users, "group" files of 4 variables × 8 selected statistics (Table 1) provide efficient access to the most commonly used statistics at manageable data volume.

"Trimming" in COADS specifically refers to the process of screening individual observations against upper and lower quality control (QC) limits defined for each 2° box and month. Flags were set in the Compressed Marine Report (CMR) format used to store Release 1 individual observations, to indicate whether observations passed or failed the screening. Observations exceeding the limits were then omitted from the trimmed 2° monthly summaries (but such observations were retained with flags in the CMR format, i.e., data were never irretrievably discarded by QC). For Release 1, the trimming limits were set at the 3.5 standard-deviation (sigma) level using three climatological periods (1854-1909; 1910-49; 1950-79).
 

2.2 Phased updates to COADS: interim updates and Release 1a (1980-95)

Following COADS Release 1, "interim" updates were completed on an approximately annual basis to extend the period of record through 1991, until a full update for 1980-92, Release 1a, was completed in 1993 (Woodruff et al., 1993). Two updates to Release 1a have since been completed. In 1995, data were added to extend Release 1a through 1993, and data for 1988 and 1992 were reprocessed at that time to correct problems in dew point and wet bulb temperatures impacting U.S.-keyed data. In early 1997, temporal coverage was extended through 1995, and the existing data for 1990-95 were updated with delayed receipts and minor processing improvements.

A variety of data additions was made to Release 1a, including replacement of many GTS ship reports by matching International Maritime Meteorological (IMM) keyed logbook reports exchanged under WMO Resolution 35, because of typically higher quality and observational completeness. GTS measurements from drifting or moored buoys were also replaced by quality controlled data from Canada's Marine Environmental Data Service (MEDS), and from NOAA's Pacific Marine Environmental Laboratory (PMEL) and the National Data Buoy Center (NDBC). In addition, special fishing fleet data from the Inter-American Tropical Tuna Commission (IATTC) helped substantially improve coverage in data-sparse regions of the equatorial Pacific Ocean.

Release 1a processing employed the QC procedures used for Release 1, with some minor improvements, and also implemented numerous data corrections. One significant error arose from an ambiguity in the IMM format, such that longitudes of French logbook data reportedly back to 1954 were converted with a 10° error, applicable to observations from 90°E-90°W across the international date line. Fortunately, diagnostic tests during Release 1a (and later Release 1b) processing confirmed Dutch and British findings that the problem was actually confined to a portion of the Release 1a period (~1981-88). We therefore replaced existing (identifiable) French logbook data in COADS during that period with properly located data provided by France.

An updated Long Marine Report format (LMR) was also developed as part of Release 1a, including a fixed-length version (LMRF) intended to meet the requirements of nearly all users. These new formats incorporate a variety of improvements and extensions in comparison to the abbreviated CMR format, including more information about platform type and any available platform ID information (ship call signs or buoy numbers).

Starting with Release 1a, electronic documentation for COADS became available for internet access using the Web (http://www.cdc.noaa.gov/coads/) or anonymous ftp. The metadata contents are gradually being updated and expanded to cover extensions to Release 1a, and updates to earlier periods as discussed below.
 

2.3 Data for Global Atmospheric Reanalysis: Releases 1a (1980-95) and 1b (1950-79)

Release 1a individual observations represent a crucial input to the National Centers for Environmental Prediction (NCEP)/NCAR Global Atmospheric Reanalysis Project (Kalnay et al., 1996). In addition, we are providing Reanalysis with enhanced and updated data for earlier periods within the planned total scope of Reanalysis (back to approximately the late 1940s). An update for 1950-79, designated as Release 1b, was completed in November 1996 and earlier observations will be provided if required by Reanalysis.

A variety of data additions was also made to Release 1b. For example, Russia has provided its Marine Meteorological Data Set of ship data from the former Soviet Union extending back to 1888 (1980-94 data were used for Release 1a), plus Arctic data back to 1950 from manned "North Pole" (NP) stations on ice floes. Surface-level sea temperatures extracted from oceanographic profiles were added from the 1994 World Ocean Atlas (WOA) (Levitus and Boyer, 1994; Boyer and Levitus, 1994). Another important new input for Release 1b was drifting buoy data gathered by MEDS for the FGGE period (1978-79).

Two errors corrected or mitigated as part of Release 1b processing involved widespread temperature biases in GTS data: The first occurred in data from U.S. Air Force Global Weather Central (GWC; deck 888), which represents the principal GTS source for COADS during 1973-79. GWC temperatures (air, sea surface, and dew point) were 0.2°C lower than they should be for the period 1973 through approximately April 1977, stemming from confusion over a change in the Kelvin base value in the original Air Force format. The second problem involved a truncation correction applied many years ago to Monterey Telecommunications (1966-73) air temperatures, which produced an average Monterey bias of approximately +0.4°C because many early temperatures were reported in whole degrees. Reversal of the correction yielded better agreement with other sources during 1966-69, whereas during 1970-73 retention of the original correction by NCDC (i.e., no adjustment by COADS) was deemed more appropriate.
 

2.4 2° and 1° monthly summaries based on Release 1a and 1b data (1950-95)

As shown by, e.g., Wolter et al. (1989) and Wolter (1997), the 3.5 sigma trimming limits developed for Release 1 have proven overly restrictive for extreme climate anomalies such as the 1877-1878 and 1982-83 El Niño/Southern Oscillation (ENSO) events. As discussed in Sect. 3, inhomogeneities within ship data arise from changes in spatial and temporal sampling, instrumentation and observational practices, and data processing. In addition, however, concerns have arisen about the validity of mixing data (especially winds and temperatures) from ships with that from buoys, sea temperatures from oceanographic profiles, and data from fishing vessels (e.g., Barnett, 1984; Wilkerson and Earle, 1990; Woodruff et al., 1993).

To begin addressing some of these complex problems, the Release 1a monthly statistics were produced in two versions. The "standard" version provides closest compatibility with Release 1 data. Here data were restricted as nearly as possible to only ship observations, and trimmed using the Release 1 (1950-79) 3.5 sigma limits. A second "enhanced" version was prepared using automated platform types in addition to ships, and processed using the trimming limits expanded to 4.5 sigma, thereby maximizing coverage and providing a more accurate representation of extreme climate anomalies.

Prior to about 1980 the data mixture becomes largely, and earlier (~1900) exclusively, ship data, but adverse trimming effects under the 3.5 sigma limits are still of concern, especially in view of new data corrections and data additions. Therefore, standard and enhanced sets of 2° monthly statistics, as defined for Release 1a processing, were also completed for 1950-79 as part of Release 1b. In addition, global and equatorial (10.5°N-10.5°S) sets of standard and enhanced monthly summaries were created at higher spatial resolution (1° latitude × 1° longitude) for 1960-93 (e.g., Fig. 1), in support of NOAA's Pan-American Climate Studies (PACS) Program, and including three new variables (Table 1).
 

3 Future update plans

Releases 1a and 1b have provided users with a variety of new and updated products, both monthly summaries for 2° and 1° boxes, and improved observational formats (LMR and LMRF). As an important preparatory step towards Release 2, we intend to re-process 1854-1949 data so as to provide users of pre-1950 data with the same format and product improvements, as well as to simplify the logistics of data distribution. This update, Release 1c, is planned for completion during 1998 (Release 1a will likely also be extended by one or two years). Release 1c will include some data additions, and format improvements such as inclusion of any ship IDs, although in earlier years it appears that original information is frequently unavailable.

Release 2, a complete re-processing of the entire period-of-record, is planned to follow Release 1c as the next major phase in our update plans. We hope to make the basic Release 2 observational data (~1820-1997), plus associated statistics and metadata, available to users by the year 2000. Following is a discussion of plans for Release 2, including QC and processing improvements that impact all the marine data elements, plus specific changes intended to improve the reliability of ocean flux fields.
 

3.1 Data digitization and acquisition

Inventories produced by Release 1 and earlier projects strongly indicated the need for additional data coverage, for example during the World War I and II periods and the 19th Century (Fig. 2). Since 1989, NCDC has made extensive efforts to fill some of these gaps through digitization of 3.5 million 1912-46 U.S. Merchant Marine reports, plus Arctic manned "ice island" data (Elms et al., 1993). International cooperation has also been actively pursued to broaden the scope of digitization (Fig. 2). For example, through a cooperative agreement with NCDC, China's National Oceanographic Data Center (CNODC) digitized 1.4 million observations from the "Maury Collection" (primarily 1820-1860). The UKMO recently digitized half a million UK records during 1935-39 out of an estimated 15 million post-1850 data in existence (Parker et al., 1995). Similarly, the Japan Meteorological Agency plans by the end of 1997 to digitize an additional one million 1892-1932 records out of possibly 5-6 million unkeyed records from its "Kobe Collection." Decks 118 (1933-53) and 119 (1953-61), which were previously keyed from the Collection by Japan, already represent sizable components of the Release 1 data, e.g., for 1933-41 as shown in Fig. 3.

An important aspect of Release 2 will involve a planned blend of COADS with the UK Meteorological Office (UKMO) Main Marine Data Bank (MDB) for the period 1854 to date, to add unique observations and help assure the quality of both data sets by cross-validation between duplicate reports. The blend should also offer the opportunity for correction of significant problems in both data sets. For example, in "Dutch" (deck 193; 1854-1938) data, which represent a major component of COADS prior to World War II (Fig. 3), sea level pressure was not translated from millimeters to millibars (this involves a correction for gravity) and thus was omitted from the 2° monthly summaries (the observed pressure values are available in supplementary fields). Similar problems impacted the MDB where the pressure values were translated but without the correction for gravity. Resolution of these translation problems represents an important priority for Release 2 to significantly augment the atmospheric pressure fields.

Early GTS data appear to represent an especially error-prone data source, due to the near-experimental nature of the system in the 1960s and early 1970s, plus archival problems stemming from insufficient attention to long-term data continuity owing to operational requirements (similar problems are emerging now that operational centers are transitioning to WMO's BUFR code for data transmission). Section 2 describes widespread errors in available GTS temperature records during the 1970s and 1960s. Another key problem in the GTS data is the present unreliability of the wind speed indicator (measured/estimated). The status of this indicator can hamper projects to create wind field adjustments for anemometer height and Beaufort equivalent-scale biases (e.g., da Silva et al., 1994; Kent and Taylor, 1997). As a major component of Release 2, we hope to resolve or mitigate these and other problems by using GTS data reconverted from original format sources.
 

3.2 Quality control and duplicate elimination improvements

QC improvements will form an important part of the enhancements planned for Release 2, with continuing attention to the need to ensure that original data values, even those identified as suspect during QC, are recoverable from an observational format. As discussed in Sect. 2, the 3.5 sigma trimming limits have proven overly restrictive, thus the 1950-79 trimming limits were expanded to 4.5 sigma for Release 1a and 1b enhanced statistics. Additional QC changes will be required to re-process pre-1950 data and include new data sources. Wolter (1997) prescribes one possible "adaptive" trimming approach that we plan to consider for feasibility, or, if data distributions in time and space are insufficient to support the adaptive approach, we tentatively plan to utilize an update of the original Release 1 trimming procedure.

QC also forms a major criterion for selection of the "best" duplicate report, as part of subsequent duplicate elimination processing. Lander and Morrissey (1987) noted small numbers of residual (unidentified) duplicates widely separated in space or time in Release 1 data (Steurer, 1987), with ~2-3% estimated to exist in Release 1a data (Kent, personal communication, 1996). Resolution of some of these problems will probably be delayed beyond Release 2, because of the difficulty of implementing generalized platform track-checking, particularly for earlier data where platform IDs may be less reliable or unavailable. However, Release 2 planning will consider the feasibility of some augmented duplicate checks, and of performing more extensive substitution of data elements between reports (e.g., between GTS and logbook data).
 

3.3 Creation of value-added observational products

The climatic significance of many trends in COADS data still remains in doubt because of changes in instrumental and observing practices, as covered in general by, e.g., Roll (1965) and Kent et al. (1993a). Problems have been heavily investigated for temperatures (e.g., Wright, 1986; Kent et al. 1993b; Folland and Parker, 1995) and winds (e.g., Ramage, 1987; Wright, 1988; Cardone et al., 1990). Studies (e.g., Ward and Hoskins, 1996) have used the sea level pressure field to evaluate the wind field, showing that careful analysis of the COADS wind data can be very useful for studies of long-term climate change. Others have described problems in estimating oceanic fluxes (e.g., Ramage, 1984; Isemer and Hasse, 1991; da Silva et al., 1994; Kent and Taylor, 1995). Time-varying distributions of the observations in time and space represent another important consideration (Fig. 4; see also Barnett, 1984 and Trenberth et al., 1992).

Increasing computing and storage capabilities coupled with COADS product improvements have increased the ability of researchers to independently prepare "value-added" sets of more homogeneous summary products, including data interpolations to produce uniformly gridded fields (e.g., Smith et al., 1996) plus instrumental corrections (e.g., Parker et al., 1994; da Silva et al., 1994). Rather than attempting similar analyses, we plan to concentrate available resources on further improvements to the individual observations (including QC improvements as discussed above). Nevertheless, basic 2° summary statistics have been a valuable stand-alone data product and remain in the plan for Release 2, with further updates to the PACS 1° products, and possibly products using finer temporal resolution where warranted by observational density, as secondary priorities.

Therefore, a major additional planned deliverable for Release 2 will involve the creation of value-added individual marine reports containing some adjusted data elements, and other improvements to make the data easier to use. Tentatively, we plan to retain original values in their existing field positions in the observational format, and at the same time establish new value-added fields in the format to contain adjusted data elements and external metadata. Wind represents a defining component of heat and momentum flux estimates, but which appears to be contaminated with major observational biases, and thus will represent the main focus of our improvements.
 

3.3.1 Wind speed improvements

Following up on the recommendation of a recent workshop (Diaz and Isemer, 1995), we intend to focus on adjustment of wind speed estimates, including incorporation of a new Beaufort equivalent scale. WMO continues to work on new recommendations for replacement of the present (Code 1100) scale. However, national and time-varying biases may delay this recommendation and preclude establishment of a single solution for all data, in which case an interim replacement in the Release 2 time-frame might be guided by comparisons of Kent and Taylor (1997).

As an initial step, original Beaufort numbers will be recovered and made readily available in the observational format, either extracted from supplemental fields or through a reverse conversion where it is known the original reported values were in Beaufort numbers. About 1949, observers began a transition to reporting estimated winds in knots (UK Meteorological Office, 1948; Ramage, 1987) or meters per second (but generally tied to the Beaufort scale), thus the value we recover might represent a best estimate of the intended Beaufort number. As a secondary priority, we plan to consider the feasibility of anemometer-height adjustments for measured wind speeds from ships and buoys.

Particularly for more recent data, the wind speed indicator (measured/estimated) represents critical metadata to allow appropriate corrections to wind data. Unfortunately, due to data processing and archival problems (Woodruff, 1995), the wind speed indicator (measured/estimated) is frequently unreliable. Some improvements in the indicator are planned for certain homogeneous decks based on original documentation, or through comparisons of duplicates including preparation of new GTS datastreams as discussed in Sect. 3.1.
 

3.3.2 Inclusion of selected metadata elements from WMO Pub. 47

We plan to include in the revised observational format some selected metadata, specifically including anemometer heights to better enable adjustments for measured winds. A variety of individual ship metadata has been published annually in WMO (1955-) Pub. No. 47, but editions are available in digital form only starting in 1973 and the original format contained some variations that hindered computer access. Fortunately, the UK Southampton Oceanography Centre (SOC) has cleaned up and uniformly reformatted the original Pub. 47 metadata since 1973 into a more usable digital form (see Kent and Taylor, 1997), which we intend to partially integrate into COADS. SOC has blended (via each ship's call sign) the reformatted metadata with COADS Release 1a data, and the combination has been employed as the basis for the production of a new global air-sea flux climatology in which corrections have been made for biases arising from observing procedure at the level of individual ship reports (Josey et al., 1996).
 

3.3.3 Present weather and cloud adjustments

Present weather and cloud observations have been used in flux and precipitation estimates based on COADS (e.g., da Silva et al., 1994; Petty, 1995). Improvements to present weather will focus on problems introduced during WMO's 1982 code change, which allowed present (and past) weather to be omitted when there was no significant weather to report. A new "station/weather" indicator (i_x) also was introduced in 1982, to differentiate non-significant weather from weather not observed. For climate statistics, it is important to use i_x, e.g., to re-establish a non-significant weather code to avoid biasing the statistics. Unfortunately, i_x was not reliably archived in most data sources until 1984 or 1985, and our efforts will include steps such as substitution of i_x values between duplicates, and setting the value-added present weather field to a non-missing code.

As part of Release 2, total cloudiness as included in the COADS 2° and 1° statistics should be reprocessed according to improved techniques (e.g., Wright, 1986; Hahn et al., 1992 and 1995). Associated improvements may be implemented as part of the value-added fields in the observational format.
 

4 Conclusions

It remains vital that long-term historical data sets such as COADS, in conjunction with detailed metadata, be extended as far back and as completely as possible, and made readily available to support the estimation of oceanic energy fluxes as well as other areas of climate and global change research. This importance is highlighted through growing interest in international observing programs, such as the Global Climate Observing System (GCOS) and its ocean component the Global Ocean Observing System (GOOS), by governments and researchers throughout the world.

In this paper we have provided some background on historical marine data and the COADS project, including recent updates and ongoing efforts to update the data set. Related activities include establishing bilateral and national agreements for acquiring additional digitized data, detecting and documenting past data processing errors, developing new observational and summary products to better serve the research community, and holding two international COADS workshops (Diaz et al., 1992; Diaz and Isemer, 1995).

Many of the lessons learned in developing COADS have led to recommendations, both within the U.S. and internationally through WMO, for future improvements in the quality, collection, and archival of marine data and metadata. Transmission via GTS has significantly increased the amount of contemporary marine data available to researchers in near real-time. However, a significant fraction of ship data is still received only through exchange of keyed logbook data, often after a delay of 2-5 years operationally, and decades retrospectively. Therefore, combining of the digitized data with GTS data remains a crucial part of the COADS work in order to maximize temporal and spatial coverage.

Acknowledgements. COADS is the result of a continuing cooperative project between the National Oceanic and Atmospheric Administration (NOAA)-specifically its Environmental Research Laboratories (ERL), National Climatic Data Center (NCDC), and Cooperative Institute for Research in Environmental Sciences (CIRES, conducted jointly with the University of Colorado)-and the National Science Foundation's National Center for Atmospheric Research (NCAR). The NOAA portion of COADS is currently supported by the NOAA Climate and Global Change Program and the NOAA Environmental Services Data and Information Management (ESDIM) Program. We are indebted to S. Lubker and P. Franks for assistance in preparing the figures; to R. Jenne, U. Radok, and K. Wolter for helpful discussions; and to an anonymous reviewer for useful comments.
 

References

Barnett, T.P., Long-term trends in surface temperature over the oceans, Mon. Wea. Rev., 112, 303-312, 1984.

Boyer, T.P. and Levitus, S., Quality Control and Processing of Historical Temperature, Salinity and Oxygen Data, NOAA Technical Report NESDIS 81, 65 pp., 1994.

Cardone, V.J., Greenwood, J.G., and Cane, M.A., On trends in historical marine wind data, J. Climate, 3, 113-127, 1990.

Diaz, H.F. and Isemer, H.-J. (Eds.), Proceedings of the International COADS Winds Workshop, Kiel, Germany, 31 May-2 June 1994, NOAA Environmental Research Laboratories, Boulder, Colo., 301 pp., 1995.

Diaz, H.F., Wolter, K., and Woodruff, S.D. (Eds.), Proceedings of the International COADS Workshop, Boulder, Colorado, 13-15 January 1992, NOAA Environmental Research Laboratories, Climate Research Division, Boulder, Colo., 390 pp., 1992.

Elms, J.D., Woodruff, S.D., Worley, S.J., and Hanson, C., Digitizing historical records for the Comprehensive Ocean-Atmosphere Data Set (COADS), Earth System Monitor, 4, No. 2, 4-10, 1993.

Fletcher, J., Radok, U., and Slutz, R., Historical ocean climate project, Trop. Ocean-Atmos. Newslett., No. 5, 1-11, 1981.

Folland, C.K. and Parker, D.E., Correction of instrumental biases in historical sea surface temperature data, Q. J. R. Meteorol. Soc., 121, 319-367, 1995.

Hahn, C.J., Warren, S.G., and London, J., The use of COADS ship observations in cloud climatologies, Proceedings of the International COADS Workshop, Boulder, Colorado, 13-15 January 1992, Diaz, H.F., Wolter, K., and Woodruff, S.D. (Eds.), NOAA Environmental Research Laboratories, Boulder, Colo., 271-280, 1992.

Hahn, C.J., Warren, S.G., and London, J., The effect of moonlight on observation of cloud cover at night, and application to cloud climatology, J. Climate, 8, 1,429-1,445, 1995.

Isemer, H.-J. and Hasse, L., The scientific Beaufort equivalent scale: effects on wind statistics and climatological air-sea flux estimates in the North Atlantic Ocean, J. Climate, 4, 819-836, 1991.

Josey, S. A., Kent, E. C., Oakley, D., and Taylor, P. K., A new global air-sea heat and momentum flux climatology, International WOCE Newsletter, 24, 3-5, 1996.

Kalnay, E. et al., The NCEP/NCAR 40-year reanalysis project, Bull. Amer. Meteor. Soc., 77, 437-471, 1996.

Kent, E.C. and Taylor, P.K., A comparison of sensible and latent heat flux estimates in the North Atlantic Ocean, J. Phys. Oceanogr., 25, 1,530-1,549, 1995.

Kent, E.C. and Taylor, P.K., Choice of a Beaufort equivalent scale, J. Atmos. Oceanic Technol., 14, 228-242, 1997.

Kent, E.C., Taylor, P.K., Truscott, B.S., and Hopkins, J.S., The accuracy of voluntary observing ships' meteorological observations-results of the VSOP-NA, J. Atmos. Oceanic Technol., 10, 591-608, 1993a.

Kent, E.C., Tiddy, R.J., and Taylor, P.K., Correction of marine air temperature observations for solar radiation effects, J. Atmos. Oceanic Technol., 10, 900-906, 1993b.

Lander, M.A. and Morrissey, M.L., Unexpected duplicate ship reports in the Comprehensive Ocean-Atmosphere Data Set (COADS), Trop. Ocean-Atmos. Newslett., No. 38, 13-14, 1987.

Levitus, S. and T.P. Boyer, World Ocean Atlas 1994, Volume 4, Temperature, NOAA Atlas NESDIS 4, U.S. Government Printing Office, Washington, D.C., 117 pp., 1994.

Maury, M.F., Maritime Conference held at Brussels for devising a uniform system of meteorological observations at sea, August and September, 1853, Explanations and Sailing Directions to Accompany the Wind and Current Charts, 6th Ed., Washington, D.C., pp. 54-88, 1854.

NCDC (National Climatic Data Center), TDF-11 Reference Manual, NCDC, Asheville, N.C., 1968.

Parker, D.E., Folland, C.K., Bevan, A., Ward, M.N., Jackson, M., and Maskell, K., Marine surface data for analysis of climatic fluctuations on interannual to century timescales, Natural Climate Variability on Decade to Century Timescales, Martinson, D.G., Bryan, K., Ghil, M., Hall, M.M., Karl, T.R., Sarachik, E.S., Sorooshian, S., and Talley, L.D. (Eds.), National Academy Press, Washington, D.C., 1994.

Parker, D.E., Folland, C.K., and Jackson, M., Marine surface temperature: observed variations and data requirements, Climatic Change, 31, 559-600, 1995.

Petty, G.W., Frequencies and characteristics of global oceanic precipitation from ship-board present weather reports, Bull. Amer. Meteor. Soc., 76, 1,593-1,616, 1995.

Ramage, C.S., Can shipboard measurements reveal secular changes in air-sea heat flux?, J. Climate Appl. Meteor., 23, 187-193, 1984.

Ramage, C.S., Secular changes in reported surface wind speeds over the ocean, J. Climate Appl. Meteor., 26, 525-528, 1987.

Reynolds, R.W. and Smith, T.M., Improved global sea surface temperature analyses using optimum interpolation, J. Climate, 7, 929-948, 1994.

Roll, H.U., Physics of the Marine Atmosphere, Academic Press, New York, N.Y., 426 pp., 1965.

Rossby, T., Siedler, G., and Zenk, W., The voluntary observing ship and future ocean monitoring, Bull. Amer. Meteor. Soc., 76, 5-11, 1995.

da Silva, A.M., Young, C.C., and Levitus, S., Atlas of Surface Marine Data 1994, Volumes 1-5, NOAA/NESDIS Atlases 6-10, U.S. Department of Commerce, NOAA, NESDIS, 1994.

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

Smith, T.M., Reynolds, R.W., Livezey, R.E., and Stokes, D.C., Reconstruction of historical sea surface temperatures using empirical orthogonal functions, J. Climate, 9, 1,403-1,420, 1996.

Steurer, P.M., Reply: unexpected duplicate ship reports in the Comprehensive Ocean-Atmosphere Data Set (COADS), Trop. Ocean-Atmos. Newslett., No. 40, 12-14, 1987.

Trenberth, K.R., Christy, J.R., and Hurrell, J.W., Monitoring global monthly mean surface temperatures, J. Climate, 5, 1,405-1,423, 1992.

UK Meteorological Office, International Meteorological Code Adopted by the International Meteorological Organization, Washington, 1947, M.O. 509, H.M. Stationary Office, London, 39 pp., 1948.

UK Meteorological Office, Marine Observer's Handbook, 10th ed., H.M. Stationery Office, London, 157 pp., 1977.

U.S. Navy, Marine Climatic Atlas of the World. Vol. IX: World-wide Means and Standard Deviations, NAVAIR 50-1C-65, U.S. Government Printing Office, Washington, D.C., 169 pp., 1981.

Ward, M.N. and Hoskins, B.J., Near-surface wind over the global ocean 1949-88, J. Climate, 9, 1,877-1,895, 1996.

Wilkerson, J.C., and Earle, M.D., A study of differences between environmental reports by ships in the Voluntary Observing Program and measurements from NOAA buoys, J. Geophys. Res., 95, 3,373-3,385, 1990.

WMO (World Meteorological Organization), International List of Selected, Supplementary and Auxiliary Ships, WMO No. 47, Geneva, Switzerland, 1955-.

WMO (World Meteorological Organization), Fourth World Meteorological Congress Abridged Report with Resolutions, WMO No. 142, Geneva, Switzerland, 1963.

WMO (World Meteorological Organization), Reports on Marine Science Affairs, Report No. 3, The Beaufort Scale of Wind Force, WMO, Geneva, Switzerland, 22 pp., 1970.

Wolter, K., Trimming problems and remedies in COADS, J. Climate, 10, 1,980-1,997, 1997.

Wolter, K., Lubker, S.J., and Woodruff, S.D., Trimming-a potential error source in COADS, Trop. Ocean-Atmos. Newslett., No. 51, 4-7, 1989.

Woodruff, S.D., Preliminary comparison of COADS (US) and MDB (UK) ship reports, Observed Climate Variations and Change: Contributions in Support of Section 7 of the 1990 IPCC Scientific Assessment, Parker, D. (Ed.), WMO/UNEP publication, 1990.

Woodruff, S.D., COADS Project report I: update plans and unresolved issues, Proceedings of the International COADS Winds Workshop, Kiel, Germany, 31 May-2 June 1994, Diaz H.F. and Isemer, H.-J. (Eds.), NOAA Environmental Research Laboratories, Boulder, Colo., 12-28, 1995.

Woodruff, S.D., Slutz, R.J., Jenne, R.L., and Steurer, P.M., A comprehensive ocean-atmosphere data set, Bull. Amer. Meteor. Soc., 68, 1,239-1,250, 1987.

Woodruff, S.D., Lubker, S.J., Wolter, K., Worley, S.J., and Elms, J.D., Comprehensive Ocean-Atmosphere Data Set (COADS) Release 1a: 1980-92, Earth System Monitor, 4, No. 1, 1-8, 1993.

Wright, P.B., Problems in the use of ship observations for the study of interdecadal climate changes, Mon. Wea. Rev., 114, 1,028-1,034, 1986.

Wright, P.B., On the reality of climatic changes in wind over the Pacific, J. Climatol., 8, 521-527, 1988.