NDP-039

DOI: 10.3334/CDIAC/cli.ndp039

Two Long-Term Instrumental Climatic Data Bases of the People's Republic of China (1997)

Download the Data and ASCII Documentation files of NDP-039

Download View, download, or print PDF documentation of NDP-039



Contributed by
Tao Shiyan, Fu Congbin, Zeng Zhaomei, Zhang Qingyun
Institute of Atmospheric Physics
Chinese Academy of Sciences
Beijing, China

Prepared by
D. P. Kaiser
(d9k@ornl.gov)
Carbon Dioxide Information Analysis Center
Oak Ridge National Laboratory
Oak Ridge, Tennessee

Environmental Sciences Division
Publication No. 4699

Date Published: September 1997

Prepared for the
Environmental Sciences Division
Office of Biological and Environmental Research
U.S. Department of Energy
Budget Activity Number KP 12 04 01 0

Prepared by the
Carbon Dioxide Information Analysis Center
OAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee 37831-6290
managed by
LOCKHEED MARTIN ENERGY RESEARCH CORP.
for the
U.S. DEPARTMENT OF ENERGY
under contract DE-AC05-96OR22464


CONTENTS

List of Figures
List of Tables
Abstract

1. Name of the Numeric Data Package
2. Principal Investigators
3. Keywords
4. Background Information
5. Description of the Database
6. CDIAC Quality Assurance Checks
7. Limitations of the Database
8. How to Obtain the Data and Documentation
9.File Descriptions

Documentation File for NDP-039 (ndp039.txt)
FORTRAN and SAS Data Retrieval Programs (.for and .sas files)
Station Inventory File for the PRC 65-Station Climate Data Set (sta65.inv)
Data Gaps File for Stations in the PRC 65-Station Climate Data Set (sta65.gap)
Station Inventory File for the PRC 205-Station Temperature and Precipitation Data Sets (sta205.inv)
65-Station Climate Data Set (sta65.dat)
Individual Station Data Files from the 65-Station Climate Data Set
205-Station Monthly Mean Temperature Date Set (temp205.dat)
205-Station Monthly Precipitation Data Set (prec205.dat)
10. References

APPENDIX A. Instrumentation and Data Collection Methods

of Stations in the PRC 65-Station Climate Database


LIST OF FIGURES

1. Locations of stations in the 65-station PRC climate database

2. Locations of stations in the PRC 205-station temperature and precipitation databases


LIST OF TABLES

1. Inventory of stations in the 65-station PRC climate database

2. Gaps in the periods of record of the 65-station PRC climate database

3. Inventory of stations in the PRC 205-station temperature and precipitation databases


ABSTRACT

Tao Shiyan, Fu Congbin, Zeng Zhaomei, and Zhang Qingyun. 1997. Two Long-Term Instrumental Climatic Data Bases of the People's Republic of China. ORNL/CDIAC-102, NDP-039. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tennessee. doi: 10.3334/CDIAC/cli.ndp039

Two long-term instrumental data bases containing meteorological observations from the People's Republic of China (PRC) are presented in this NDP . The first version of this database was made available in 1991 by the Carbon Dioxide Information Analysis Center (CDIAC) as CDIAC NDP-039. This update of the database includes data through 1993. These data sets were compiled in accordance with a joint research agreement signed by the U.S. Department of Energy and the PRC Chinese Academy of Sciences (CAS) on August 19, 1987. CAS has provided records from 267 stations, partitioned into two networks of 65 and 205 stations, with three stations common to both data bases.

The 65-station-network data contain monthly means, extremes, or totals of barometric pressure, air temperature, precipitation amount, relative humidity, sunshine duration, cloud amount, dominant wind direction and frequency, wind speed, and number of days with snow cover. Station histories are available from 59 of the 65 stations.

The 205-station-network data contain monthly mean temperatures and monthly precipitation totals; however, station histories are not currently available.

Sixteen stations from these data sets (13 from the 65-station, 3 from the 205-station) have temperature and/or precipitation records beginning before 1900, whereas the remaining stations began observing in the early to mid-1900s.


1. NAME OF THE NUMERIC DATA PACKAGE

Two Long-Term Instrumental Climatic Data Bases of the People's Republic of China


2. PRINCIPAL INVESTIGATORS

Tao Shiyan, Fu Congbin, Zeng Zhaomei, Zhang Qingyun
Institute of Atmospheric Physics
Chinese Academy of Sciences
Beijing, China


3. KEYWORDS

China; regional climate change; atmospheric pressure; temperature; precipitation; relative humidity; sunshine duration; cloud amount; snow cover; dominant wind direction; dominant wind frequency; wind speed; extreme temperature


4. BACKGROUND INFORMATION

The U.S. Department of Energy (DOE) and the People's Republic of China (PRC) Chinese Academy of Sciences (CAS) signed an agreement on August 19, 1987, to carry out a joint research program on possible CO2-induced climate changes. Concern over such climate changes stems from rising atmospheric concentrations of CO2 and other greenhouse gases, as well as the ever-increasing global CO2 emissions from fossil fuel use (Marland and Rotty 1984). The nature of these phenomena signals the need for international research efforts in the area of possible human-induced climate change. The intent of the DOE/CAS research agreement is to contribute to research efforts through four tasks: (1) analysis of general circulation models (GCMs), (2) preparation and analysis of proxy and instrumental data, (3) study of the relationship between large- and regional-scale climate, and (4) study of possible increases in atmospheric methane (Riches et al. 1992). Descriptions of these tasks and additional background information on each country's involvement in atmospheric CO2 research may also be found in Riches et al. (1992). This database contains and describes two long-term PRC instrumental climate data sets, which were exchanged along with PRC proxy data sets and U.S. instrumental data sets according to the data exchange protocol contained in the DOE/CAS agreement.


5. DESCRIPTION OF THE DATABASE

Two PRC climate data bases, derived mainly from instrument measurements, are presented in this package; one consists of monthly means, extremes, or totals of 14 meteorological variables observed at 65 stations; the other contains monthly mean temperatures and monthly precipitation totals from 205 stations. The two data bases have three stations in common. Each data base is described in detail in the following sections.

65-Station Network

As previously noted, the data from these 65 stations consist of monthly means, extremes, or totals of the following 14 meteorological variables:

Mean Station Pressure (mb)
Mean Temperature (°C)
Mean Maximum Temperature (°C)
Mean Minimum Temperature (°C)
Total Precipitation (mm)
Sunshine Duration (hours)
Mean Cloud Amount (percentage of sky cover)
Mean Relative Humidity (percent)
Snow Days (days with snow cover)
Dominant Wind Direction (degrees)
Mean Wind Speed (m/s)
Dominant Wind Frequency (percent)
Extreme Maximum Temperature (°C)
Extreme Minimum Temperature (°C)

Each record contains one month's data: the WMO station number, year, and month; followed by the data variables in the same order as previously listed. Each data value is given in tenths of its specified unit of measure (i.e., actual values are multiplied by 10 to yield an integer value). For example, a value of "10132" for mean station pressure means the pressure is 1013.2 mb). Missing data values are indicated by "-9999".

Four basic criteria were used in selecting stations to make up the 65-station network: (1) the station should be representative of a particular climate region of China, (2) the station's data should be of relatively high quality, (3) the period of record of the station should be reasonably long, and (4) the resulting station network should have a relatively uniform spatial distribution. Station locations are depicted on the map shown in Figure 1, with an inventory of the stations given in Table 1. Data are available from all 65 stations over the period 1954-1993. Thirteen stations began observing before 1900. Beijing's record is the longest, with temperature and precipitation data dating back to 1841. All 65 stations have temperature and/or precipitation data available from the beginning of their periods of record, with other types of observations beginning more randomly with time throughout the network. Observations of monthly extreme maximum and minimum temperature data are available for each station only since 1984. Nearly one-half of the 65 stations have some type of data for at least 70 years, but only six have data for 100 years or more. Periods of whole years without data are omitted from the data set, and these gaps are detailed in Table 2. Station numbers plotted on the map in Figure 1 correspond to the station reference numbers given at the left in Tables 1 and 2. Details of data collection methods used at the 65 stations (e.g., type, orientation, and calibration of instruments; daily observing times; and observational techniques) have been made available by CAS and are contained in Appendix A. Station histories extending through 1983 are available for 59 of the 65 stations via Table 1.

205-Station Network

Data from the 205-station network contain two variables: monthly mean temperature and total monthly precipitation. Station locations are depicted on the map shown in Figure 2, with an inventory of the stations given in Table 3. The data are stored in two files, one with temperatures (in tenths of degrees Celsius) and the other with precipitation totals (in tenths of millimeters). Each record of a file contains data from one year: the WMO station number and year followed by twelve monthly data values. Missing values are represented by "-9999". The temperature data date back as far as 1907 (Jingzhou, Station No. 57476) and the precipitation data as far back as 1880 (Wuhu, No. 58334). Temperature records extend through 1993 or 1994 depending on the station; precipitation records end in 1993. Records from Geershiquanhe (No. 55228), Hekou (No. 56989), and Lingling (No. 57866) have gaps extending from the early 1980s through 1988. Since there are relatively few years with missing data, such records have been filled with missing data indicators and remain in the data files. For both temperature and precipitation data, the most typical length of data record ranges between 35 and 45 years, with a relatively small share of stations [~15% for temperature; ~12% for precipitation] having records for 50 or more years. The 65-station data set offers more truly long-term records and, as previously noted, contains temperature and/or precipitation data from the beginning of each station record. The first year for which all 205 stations have temperature and precipitation data is 1961. Unfortunately, station histories are not currently available for any of the stations in the 205-station network; therefore, details regarding instrumentation, collection methods, changes in station location or observing times, and official data sources are not known.


6. CDIAC QUALITY ASSURANCE CHECKS

An important part of the numeric data package process at CDIAC is the quality assurance (QA) of data before distribution. Data received at CDIAC are rarely in perfect condition for immediate distribution, regardless of the source. To guarantee data of the highest possible quality, CDIAC conducts extensive QA reviews, which involve examining the data for completeness, reasonableness, and accuracy. Although these reviews have common objectives, they are tailored to each data set, often requiring extensive programming efforts. Although time-consuming, the QA process is an important component in the value- added concept of ensuring accurate, usable data for researchers. Given the large amount and many types of data contained in the PRC data sets, it is not surprising that CDIAC found some problems with the values contained in the data files. Many of these problems were resolved with the help of CAS, but in the interest of timely dissemination of the data, other questionable data are flagged so users may treat these data as they see fit. Flag codes and their definitions for the data in both the 65-station and 205-station data sets are provided in Sect. 9. Users are strongly advised to check data values for associated flags that may have been assigned. Not doing so may to lead to spurious analytical results. Descriptions of the QA reviews performed for each of the PRC data sets are presented in the following.

65-STATION CLIMATE DATA SET

The data from these stations were provided by CAS as three files; one with data extending through 1983 (received in 1988), one with data over the period 1984-1988 (received in 1990), and one containing data from 1989-1993 (received in 1996). The second and third data sets, in addition to the twelve variables found in the first, contained values of monthly extreme maximum and minimum temperature. These three data sets were interleaved according to station number, and the missing indicator "-9999". was inserted in the record position of extreme maximum and minimum temperature for years through 1983. The resulting data set was then examined thoroughly using software routines developed to check for (1) trends or jumps in the data not indicative of normal seasonal cycles or documented station moves; (2) statistical outliers; (3) identical, non-zero values reported over consecutive months for variables such as total precipitation and sunshine duration; (4) proper relationships between monthly temperature variables; and (5) impossible values (e.g., number of days with snow cover exceeding the number of days in a particular month). The results of the QA checks (by data type), along with a brief assessment of station histories, are given in the following sections.

Station Histories

Station histories are available (via Table 1) for 59 of the stations contained in the original release of NDP-039 (Tao et al. 1991). Histories are not presently available from the 5 additional stations whose data were provided by CAS in 1996 (augmenting the former 60-station data set to 65 stations). These histories contain much of the information needed to determine if a particular station's data can be used confidently by researchers interested in making assessments of the homogeneity of observations. This includes information regarding station locations (e.g., city, suburb, or countryside; elevation), station moves, and observing times. However, some potentially useful station-specific details are not present. For example, the switch from the Jordan Photographic Sunshine recorder to the Campbell-Stokes Sunshine recorder, noted in Appendix A as occurring in 1954, may not have taken place at all stations simultaneously as implied. [Sunshine recorder changes took many years to implement over the entire U.S. sunshine recording network (Karl and Steurer 1990)]. Users should be aware that the 59 station histories from the original 60-station database extend only through 1983. Also, it was found that for three stations (53463, 58367, and 58633), the data record begins before the period documented by the station history. For three other stations (50745, 54527, and 56294), data are not available until several years after the first year documented in the station histories. Finally, barometric pressure data suggest that three station histories (for stations 57127, 57461, and 57516) lack documentation of station moves (discussed in the following section).

Monthly Mean Barometric Pressure

The data seem to show an occasional departure from reporting values to the nearest tenth of a millibar, as evidenced by occurrences of several (in one case, six) consecutive monthly values containing a zero in the tenths place. Software designed to find deviations from the otherwise characteristic seasonal cycle of monthly mean barometric pressure uncovered evidence of undocumented station moves at stations 57127, 57461, and 57516, each one having one or more years of pressure data with a markedly different range as compared to prior or subsequent years (assumably due to a change in station elevation). Other single-month values that departed significantly from typical seasonal trends were flagged if not corroborated by neighboring stations.

Temperature

Checks for proper relationships between the various monthly mean/extreme temperatures have been made, and the relatively small number of errors found have been corrected through consultation with CAS. Another simple check involved looking for statistical outliers by calculating the mean and standard deviation of the respective monthly values for all temperature variables from each station. Values lying 3.5 or more standard deviations away from the mean of the monthly values over the station's period of record were flagged if not corroborated by temperatures at neighboring stations. Checking for departures from the characteristic seasonal cycle of monthly mean temperature produced several values that, while not statistical outliers, were also flagged as suspect if they could not be corroborated by monthly temperature trends from neighboring stations. The user is referred to the discussion of flag codes for the 65-station data set in Sect. 9 for additional details.

Total Precipitation

QA checks basically focused on flagging very high totals and non-zero repeated totals. Totals exceeding 500 mm (~20 in.) were found in 176 cases, and three of these totals exceeded 1 m. Practically all of these totals were considered valid, as they occurred during the summer monsoon season. However, the extremely high totals found in the original CAS data set for LaSa (No. 55591) for May- September of 1936 (486.5, 517.7, 2049.6, 1313.2, and 619.2 mm, respectively) were considered suspect and replaced by totals for LaSa obtained from Eischeid et al. (1991). These totals are flagged in the file included with this package and are 48.5, 52.7, 205.0, 131.0, and 62.0 mm for May-September, respectively. Several high totals for other stations, while not edited, were also flagged, as were several non-zero identical totals over consecutive months (Sect. 9).

Sunshine Duration

These data were checked for identical values over consecutive months. This type of occurrence should be relatively rare since the monthly totals are rounded to the nearest tenth of an hour. Forty-nine pairs of months were found where this did occur. These values have been flagged (Sect. 9) so users may decide if they appear plausible (perhaps by considering cloud amounts or other data over these months). Documentation provided by CAS (Appendix A) points out that a significant number of estimates (not identified by CAS) are included in these data.

Monthly maximum possible sunshine duration was calculated for each station using the station's latitude and assuming a smooth, spherical earth where the sun's rays would not be blocked by natural or man-made obstructions. Using the maximum durations, monthly values of percentage of possible sunshine were calculated for each station and plotted versus corresponding monthly cloud amounts. It was determined that for many stations - mainly northern stations, and exclusively during astronomical winter (primarily from November through January) - monthly sunshine duration is often unrealistically low. For most stations, this phenomenon became more prevalent through time, being at least partly responsible for most stations showing significant decreasing trends in sunshine duration over their periods of record. This may relate to increasing atmospheric turbidity (air pollution), which would decrease recorded sunshine most during the low sun season due to longer atmospheric path lengths. Details of this analysis may be found in Kaiser (1993), copies of which are available from the author (d9k@ornl.gov).

Because of the large number of suspect low monthly sunshine durations, these values have not been individually flagged in the 65-station data set, and users are cautioned in their use of these sunshine data.

Monthly Mean Cloud Amount

Monthly cloud amounts were checked via scatter plots of cloud amount vs. percentage of possible sunshine. Several suspect values (seemingly too low) were noted and flagged in the data file (see discussion of flags in Sect 10). Documentation provided by CAS (Appendix A) points out that a significant number of estimates (not identified by CAS) are included in these data.

Monthly Mean Relative Humidity

These data were simply checked for negative values or values greater than 100%. None were found; the data appeared reasonable upon cursory inspection.

Total Number of Days with Snow Cover

Monthly variations for stations reporting any days with snow cover seem reasonable, and no values exceeding the number of days in a particular month were found.

Monthly Dominant Wind Direction and Dominant Wind Frequency

Inconsistencies have been found in these data. The dominant wind direction is defined in Appendix A as the most frequent wind direction observed for the month, measured in 22.5° increments of azimuth clockwise from north. A direction of 0° is used to denote calm winds, while 360° indicates a north wind. The data have been checked to make sure all dominant wind directions (if not set to the missing code "-9999") are 0° or some multiple of 22.5° up to 360°. Pre-1981 data show a large number of dominant wind directions with values of 0°, the extent of which varies greatly by individual station. This probably makes good physical sense, as diurnal variations alone in many areas could often be associated with lengthy periods of calm at night. However, all dominant wind directions for 1981-1983 are missing, and there are no values of 0° for any station during the period 1984-1988. This most likely implies a change in the expression of dominant wind direction for the period 1984-1988, while the reason for missing data for 1981-1983 is unknown. Data prior to 1981 have reasonable values for dominant wind frequency, usually ranging from 10 to 50%. Frequencies for 1981-1983 are missing due to missing dominant wind direction, as are those for 1984-1988, even though values of unknown character are present for dominant wind direction. Taking these factors into consideration, the pre-1981 data are considered reasonable, but the nature of these data thereafter is unknown.

Monthly Mean Wind Speed

These data appear reasonable; values range from 0 to 9.4 m/s. Only one value of 0 was found (for December 1952 at Yi Ning, No. 51431), and since the dominant wind direction is given as 0°(calm) the value has not been researched. (The dominant wind frequency is missing for this month).

205-STATION NETWORK

CAS provided data from these stations as separate temperature and precipitation files. These have been slightly reformatted to allow for the insertion of flags. QA checks and results for each data set are described in the following sections.

Temperature

Means and standard deviations of each station's respective monthly values were calculated. Values lying 3.5 or more standard deviations away from their respective means were flagged if not corroborated by temperatures at neighboring stations. Checking for departures from the characteristic seasonal cycle of monthly mean temperature produced several values that, while not statistical outliers, were deemed suspect if they could not be corroborated by monthly temperature trends from neighboring stations. For additional details of flag codes for the 205-station temperature data set see Sect. 9.

Total Precipitation

QA checks focused on researching very high totals and flagging non-zero totals repeated over consecutive months. Several hundred totals exceeding 500 mm are present in the data, with three June totals from Heyuan (No. 59293) exceeding 1 m. These Heyuan totals and several other especially large totals were researched by CAS and confirmed. All of these confirmed large amounts were received during the summer monsoon season. A few unrealistically high totals that could not be confirmed by CAS were flagged in the precipitation data file, but the values themselves were not edited. Many instances of identical non-zero totals over consecutive months were found in the data. These have been flagged as described in Sect. 9.


7. LIMITATIONS OF THE DATABASE

Few station records included in the PRC data sets can be considered truly homogeneous. Even the best stations were subject to minor relocations or changes in observing times, and many have undoubtedly experienced large increases in urbanization. Fortunately, for 59 of the stations in the 65-station network, station histories (see Table 1) are available to assist in proper interpretation of trends or jumps in the data; however, station histories for the 205-station network are not available. In addition, examination of the data from the 65-station data set has uncovered evidence of several undocumented station moves (Sects. 6 and 10). Users should therefore exercise caution when using the data. Inhomogeneities probably exist in the sunshine data because of increases in atmospheric turbidity over time (see Sect. 6) and changes in sunshine recorder instrumentation.

Only a portion of the questionable values found in the data have been thoroughly researched and edited via communication with CAS. Most questionable values have been left intact and flagged (Sects. 6 and 9) so that the user may determine how to treat them. Users are strongly advised to check data values for associated flags that may have been assigned. Not doing so may to lead to spurious analytical results.


8. HOW TO OBTAIN THE DATA AND DOCUMENTATION

This database is available in machine-readable form from CDIAC without charge. The database may also be downloaded from CDIAC's anonymous File Transfer Protocol (FTP) site, by using the address, login instructions, and FTP commands listed below. In addition, the database is available on 9-track magnetic tape, 8-mm tape, or IBM- or Macintosh-formatted floppy diskettes. For tape requests, please include preferred tape specifications (i.e., 1600 or 6250 BPI for 9-track tapes and 8200 or 8500 format for 8-mm tapes, labeled or nonlabeled, ASCII or EBCDIC characters, variable or fixed-record lengths). Requests without tape specifications will be filled on 9-track, 6250 BPI, nonlabeled tapes.

This documentation is available only from CDIAC. Printed copies of the documentation that accompanied the original version of this database (Tao et al. 1991) may be obtained from CDIAC; however, a printed document for the current version of the database (NDP-039) has not been produced. Requests for printed copies of Tao et al. (1991) may be obtained using the information below.

Carbon Dioxide Information Analysis Center
World Data Center-A for Atmospheric Trace Gases
Oak Ridge National Laboratory
Post Office Box 2008
Oak Ridge, TN 37831-6290, USA


The tapes, diskettes, and documentation may also be ordered by telephone, facsimile, or electronic mail:

Telephone: (865) 574-3645 or (865) 574-0390
Fax: (865) 574-2232
Electronic mail: cdiac@ornl.gov

FTP access: ftp cdiac.esd.ornl.gov (or 128.219.24.36) Enter anonymous at the userid prompt
Enter your e-mail address as the password
Change the directory to /pub/ndp039 (i.e., ftp> cd /pub/ndp039)
Retrieve all files (i.e., ftp> mget *)

NOTE: When using these climate data in a presentation or publication, PLEASE acknowledge the principal investigators, Shiyan Tao, Congbin Fu, Zhaomei Zeng, Qingyun Zhang, and the Chinese Academy of Sciences.


9. FILE DESCRIPTIONS

This section describes the 82 files contained in this database. In addition to the actual data files, the 82 files include associated documentation and FORTRAN and SAS codes for reading the data files.

DOCUMENTATION FILE FOR NDP-039R1 (ndp039.txt)

This file is an ASCII version of the documentation that may be found for NDP-039 on CDIAC's World Wide Web site. It exists for the benefit of those downloading the database directly from CDIAC's file transfer protocol (FTP) area without the use of a web browser.

FORTRAN AND SAS DATA RETRIEVAL PROGRAMS (.for and .sas files)

These files are provided for the benefit of users with FORTRAN or SAS on their systems, enabling them to read any of the 5 data files in this database using these software packages. The program files are:

INV65.FOR -
FORTRAN program for reading the inventory file for the 65-station database (sta65.inv)

GAPS65.FOR -
FORTRAN program for reading the data gaps file for the 65-station database (sta65.gap)

INV205.FOR -
FORTRAN program for reading the inventory file for the 205-station database (sta205.inv)

DAT65.FOR -
FORTRAN program for reading the climate data file for the 65-station database (sta65.dat)

DAT205.FOR -
FORTRAN program for reading the temperature and precipitation data files for the 205-station database (temp205.dat and prec205.dat)

INV65.SAS -
SAS program for reading the inventory file for the 65-station database (sta65.inv)

GAPS65.SAS -
SAS program for reading the data gaps file for the 65-station database (sta65.gap)

INV205.SAS -
SAS program for reading the inventory file for the 205-station database (sta205.inv

DAT65.SAS -
SAS program for reading the climate data file for the 65-station database (sta65.dat)>

DAT205.SAS -
SAS program for reading the temperature and precipitation data files for the 205-station database (temp205.dat and prec205.dat)


STATION INVENTORY FILE FOR THE PRC 65-STATION CLIMATE DATA SET (sta65.inv)

The station inventory file for the 65-station climate data set is sorted by sequence number (1-65) and 5-digit station number, with one record per station containing station name, station number, latitude, longitude, elevation above sea level, and the beginning year of the station's record. The contents of the station inventory file are shown in Table 1.

The file may be read using the following FORTRAN format:

      INTEGER SEQNUM,STANUM,LATDEG,LATMIN,LONDEG,LONMIN,BEGYR
      REAL ELEV
      CHARACTER*10 STNAME
      READ(5,100,END=99)SEQNUM,STNAME,STANUM,LAT,
    + LON,ELEV,BEGYR
100   FORMAT(I2,1X,A10,1X,I5,1X,F5.2,1X,F6.2,1X,F6.1,1X,I4)

or by using the SAS format:

 INPUT SEQNUM 1-2 STNAME $ 4-13 STANUM 15-19 LAT 21-25
       LON 27-32 ELEV 34-39 BEGYR 41-44

Stated in tabular form the contents include the following:

                      Variable           Variable         Starting          Ending
Variable                type              width            column           column

SEQNUM                 Numeric               2                1                2
STNAME                 Character            10                4               13
STANUM                 Numeric               5               15               19
LAT                    Numeric               5               21               25
LON                    Numeric               6               27               32
ELEV                   Numeric               6               34               39
BEGYR                  Numeric               4               41               44


where

SEQNUM               is the station's relative position in the file (1-65);

STNAME               is the name of the station;

STANUM               is the WMO station number;

LAT                  is the latitude of the station in decimal degrees North;

LON                  is the longitude of the station in decimal degrees East;

ELEV                 is the elevation of the station above sea level (m); and

BEGYR                is the beginning year of the station's record.

DATA GAPS FILE FOR STATIONS IN THE PRC 65-STATION CLIMATE DATA SET (sta65.gap)

This file lists gaps of one or more years for each station in the 65-station network. The contents of the data gaps file are shown in Table 2. The file may be read using the following FORTRAN format:
      INTEGER SEQNUM,STANUM
      CHARACTER*10 STNAME
      CHARACTER*51 GAPS
      READ(5,100,END=99)SEQNUM,STNAME,STANUM,GAPS
100   FORMAT(I2,1X,A10,2X,A51)

or by using the SAS format:

 INPUT SEQNUM 1-2 STNAME $ 4-13 STANUM 19-23 $ GAPS 26-76

Stated in tabular form the contents include the following:

                      Variable           Variable         Starting          Ending
Variable                type              width            column           column

SEQNUM                 Numeric               2                1                2
STNAME                 Character            10                4               13
STANUM                 Numeric               5               19               23
GAPS                   Character            51               26               76

where

SEQNUM               is the station's relative position in the file (1-65);

STNAME               is the name of the station;

STANUM               is the WMO station number;

GAPS                 is a list of gaps in the record (individual years or ranges of years;
                     Beijing (54511) has 3 records in this file due to its long observation
                     history and numerous gaps).

STATION INVENTORY FILE FOR THE PRC 205-STATION TEMPERATURE AND PRECIPITATION DATA SETS (sta205.inv)

The station inventory file for the 205-station data sets is sorted by 5-digit station number, with one record per station containing a sequence number (1-205), station name, station number, latitude, longitude, elevation above sea level, and the beginning years of the station's temperature and precipitation records. The contents of the station inventory file are shown in Table 3.

The file may be read using the following FORTRAN format:

      INTEGER SEQNUM,STANUM,LAT,LON,TBEGYR,PBEGYR
      REAL ELEV
      CHARACTER*20 STNAME
      READ(5,100,END=99)SEQNUM,STNAME,STANUM,LAT,LON,ELEV,TBEGYR,PBEGYR
100   FORMAT(I3,1X,A20,1X,I5,1X,F5.2,1X,F6.2,1X,F6.1,2(1X,I4))

or by using the SAS format:

 INPUT SEQNUM 1-3 STNAME $ 5-24 STANUM 26-30 LAT 32-36
       LON 38-43 ELEV 45-50 TBEGYR 52-55 PBEGYR 57-60

Stated in tabular form, the contents include the following:

                        Variable            Variable       Starting        Ending
Variable                  type               width          column         column

SEQNUM                  Numeric                3               1              3
STNAME                  Character             20               5             24
STANUM                  Numeric                5              26             30
LAT                     Numeric                5              32             36
LON                     Numeric                6              38             43
ELEV                    Numeric                6              45             50
TBEGYR                  Numeric                4              52             55
PBEGYR                  Numeric                4              57             60

where

SEQNUM               is the station's relative position in the file (1-205);

STNAME               is the name of the station;

STANUM               is the WMO station number;

LAT                  is the latitude of the station in decimal degrees North;

LON                  is the longitude of the station in decimal degrees East;

ELEV                 is the elevation of the station above sea level (m);

TBEGYR               is the beginning year of the station's temperature record; and

PBEGYR               is the beginning year of the station's precipitation record.

65-STATION CLIMATE DATA SET (sta65.dat)

The climate data set from the 65-station network contains monthly measurements of 14 climatological variables. The data are sorted by station number, with each record containing data for one month, including station number, year, month, mean station pressure, mean temperature, mean maximum and minimum temperatures, total precipitation, sunshine duration, mean cloud amount, mean relative humidity, total days with snow cover, dominant wind direction, mean wind speed, dominant wind frequency, and extreme maximum and minimum temperatures for the month (1984-1988 only). The file may be read using the following FORTRAN format:

      INTEGER STANUM,YEAR,MON,PRES,MEANT,MEANMX,MEANMN,
    + PRECIP,SUN,CLD,RH,SNOW,WDIR,WSPD,WFREQ,EXMX,EXMN

      CHARACTER*1 RECF,PRESF,MEANTF,MXF,MNF,PRECF,SUNF,CLDF,
    + RHF,SNOWF,WDIRF,WSPDF,WFREQF,EXMXF,EXMNF

      READ(5,100,END=99)STANUM,YEAR,MON,RECF,PRES,PRESF,MEANT,
    + MEANTF,MEANMX,MXF,MEANMN,MNF,PRECIP,PRECF,SUN,SUNF,
    + CLD,CLDF,RH,RHF,SNOW,SNOWF,WDIR,WDIRF,WSPD,WSPDF,
    + WFREQ,WFREQF,EXMX,EXMXF,EXMN,EXMNF
100   FORMAT(1X,I5,1X,I4,1X,I2,1X,A1,14(I6,1X,A1))


or by using the SAS format:

 INPUT STANUM 2-6 YEAR 8-11 MON 13-14 RECF $ 16 PRES 18-22
       PRESF $ 24 MEANT 26-30 MEANTF $ 32 MEANMX 34-38 MXF $ 40
       MEANMN 42-46 MNF $ 48 PRECIP 50-54 PRECF $ 56
       SUN 58-62 SUNF $ 64 CLD 66-70 CLDF $ 72
       RH 74-78 RHF $ 80 SNOW 82-86 SNOWF $ 88
       WDIR 90-94 WDIRF $ 96 WSPD 98-102 WSPDF $104
       WFREQ 106-110 WFREQF $ 112
       EXMX 114-118 EXMXF $ 120 EXMN 122-126 EXMNF $ 128;

Stated in tabular form, the contents include the following.

                      Variable          Variable           Starting           Ending
Variable                type             width              column            column


STANUM                 Numeric             5                   2                 6
YEAR                   Numeric             4                   8                11
MON                    Numeric             2                  13                14
RECF                   Character           1                  16                16
PRES                   Numeric             5                  18                22
PRESF                  Character           1                  24                24
MEANT                  Numeric             5                  26                30
MEANTF                 Character           1                  32                32
MEANMX                 Numeric             5                  34                38
MXF                    Character           1                  40                40
MEANMN                 Numeric             5                  42                46
MNF                    Character           1                  48                48
PRECIP                 Numeric             5                  50                54
PRECF                  Character           1                  56                56
SUN                    Numeric             5                  58                62
SUNF                   Character           1                  64                64
CLD                    Numeric             5                  66                70
CLDF                   Character           1                  72                72
RH                     Numeric             5                  74                78
RHF                    Character           1                  80                80
SNOW                   Numeric             5                  82                86
SNOWF                  Character           1                  88                88
WDIR                   Numeric             5                  90                94
WDIRF                  Character           1                  96                96
WSPD                   Numeric             5                  98               102
WSPDF                  Character           1                 104               104
WFREQ                  Numeric             5                 106               110
WFREQF                 Character           1                 112               112
EXMX                   Numeric             5                 114               118
EXMXF                  Character           1                 120               120
EXMN                   Numeric             5                 122               126
EXMNF                  Character           1                 128               128


where

STANUM               is the WMO station number;

YEAR                 is the year of the data;

MON                  is the month of the data;

PRES                 is the monthly mean station pressure (mb);

MEANT                is the monthly mean temperature (°C);

MEANMX               is the monthly mean of the daily maximum temperatures (°C);

MEANMN               is the monthly mean of the daily minimum temperatures (°C);

PRECIP               is the total precipitation  (liquid and equivalent liquid of frozen precipitation)
                     for the month (mm);

SUN                  is the total sunshine duration for the month (hr);

CLD                  is the mean cloud amount for the month (percent of sky cover);

RH                   is the mean relative humidity for the month (percent);

SNOW                 is the number of days in the month with measurable snow cover;

WDIR                 is the dominant (most frequent) wind direction observed over the month,
                     measured in 22.5° increments of azimuth clockwise from north (0 indicates
                     calm winds, while 360 indicates a north wind);

WSPD                 is the mean wind speed for the month (m/s);

WFREQ                is the dominant wind frequency for the month (percent);

EXMX                 is the extreme maximum temperature observed for the month (°C); and

EXMN                 is the extreme minimum temperature observed for the month (°C).

All of the above meteorological variables are expressed in tenths of their specified
units of measure (i.e., actual values are multiplied by 10 to yield an integer value). For example,
a value of "10132" for mean station pressure means the pressure is 1013.2 mb).
Missing data values are indicated by "-9999".


Flag Codes for the Data

RECF is a flag denoting whether any observations in the record have been corrected using other sources or flagged as being suspect. The following codes indicate which values have been flagged: P = pressure; T = one or more temperature; R = precipitation total; S = sunshine duration; M = multiple parameters; and blank = record contains no data flags. PRESF is the pressure value flag. The codes are as follows: T = the pressure departs from general monthly trends; and E = pressure values for this year suggest a prior or subsequent change in barometer elevation not noted in the station history. MEANTF is a flag denoting suspect values of mean monthly temperature. The codes are as follows: A = outlier value*; not consistent with other reported mean temperature variables for this month; B = outlier value; repeated the following month; C = outlier value; D = value is identical to previous month's value; E = value does not fit typical monthly trends and is inconsistent with other reported mean temperature variables for the month; F = value is the highest ever reported for this month at this station and is repeated the following month; G = value is identical to previous month's value, which has been flagged as being suspect; and H = all three mean temperature variables have identical values over May and June of the current year. X = this temperature does not satisfy the relationship: MEANMN < MEANT < MEANMX, whereas the other two mean temperature entries are considered valid. Y = this temperature and at least one other mean temperature entry does not satisfy the relationship: MEANMN < MEANT < MEANMX. MXF is a flag denoting suspect values of monthly mean maximum temperature. The codes include A, D, E, H, and X from the aforementioned MEANTF flag, in addition to the following: I = outlier value; likely caused by an error in sign. MNF is a flag denoting suspect values of monthly mean minimum temperature. The codes are A, D, E, H, X, and I from the aforementioned MEANTF and MXF flags. PRECF is the precipitation total flag. The codes are as follow: R = total is identical to the previous or following month's total; H = total is especially high for this station and is considered suspect; and; E = original total was considered suspect due to being especially high for the station. It has been replaced by data from Eischeid et al. (1991). SUNF is the sunshine duration flag. The code used is: R = total is identical to the previous or following month's total. (See Sect.6 for further information on questionable sunshine duration values.) CLDF is the cloud amount flag. The code used is: L = cloud amount is likely too low, as determined by a scatter plot of monthly sunshine duration vs. cloud amount. EXMXF is the monthly extreme maximum temperature flag. The code used is: L = monthly extreme maximum temperature is lower than the monthly mean maximum temperature, which is considered valid. EXMNF is the monthly extreme minimum temperature flag. The code used is: H = monthly extreme minimum temperature is higher than the monthly mean minimum temperature, which is considered valid. (*Defined as lying 3.5 or more standard deviations away from the mean value of the variable.)
The data flags RHF, SNOWF, WDIRF, WSPDF, and WFREQF are not presently used in the 65-station data set (their record locations contain only blanks). Relative humidity and number of days with snow cover data contain no obvious outliers. However, the dominant wind direction data were found to contain values of a questionable nature. Pre-1981 data contain a considerable number of dominant wind directions of 0 (calm winds), whereas data from the period 1984-1988 contain none. (Dominant wind direction data for the period 1981-1983 are missing.)

INDIVIDUAL STATION DATA FILES FROM THE 65-STATION CLIMATE DATA SET

For the convenience of the user, the 65-station data file has been broken up by individual station. File names are of the form prcXXXXX.dat, where "XXXXX" is the station's WMO number. Station data may be downloaded using the links in Table 1, retrieved individually from the ndp039 directory on CDIAC's anonymous ftp server, or ordered from CDIAC on a variety of magnetic media. See the above information on the complete 65-station data file for format and flag information.

205-STATION MONTHLY MEAN TEMPERATURE DATA SET (temp205.dat)

The temperature data set from the 205-station network contains monthly mean temperatures. The data are sorted by station number, with each record containing data for one year, including station number, year, and mean temperatures for January through December. Temperatures are given in tenths of degrees Celsius.

The file may be read using the following FORTRAN format:

       INTEGER STANUM, YEAR, TEMP(12)
       CHARACTER*1 RFLAG, TFLAG(12)
       READ (5,100,END=99) STANUM, YEAR, RFLAG,
    +  (TEMP(I),TFLAG(I), I=1,12)
100    FORMAT(2I6,1X,A1,12(I6,1X,A1))

or by using the SAS format:

 INPUT STANUM 2-6 YEAR 9-12 RFLAG $ 14
 JAN 16-20 JANFL $ 22 FEB 24-28 FEBFL $ 30 MAR 32-36 MARFL $ 38
 APR 40-44 APRFL $ 46 MAY 48-52 MAYFL $ 54 JUN 56-60 JUNFL $ 62
 JUL 64-68 JULFL $ 70 AUG 72-76 AUGFL $ 78 SEP 80-84 SEPFL $ 86
 OCT 88-92 OCTFL $ 94 NOV 96-100 NOVFL $ 102 DEC 104-108 DECFL $ 110


Stated in tabular form, using variable names from the SAS format, the contents include the
following:

                     Variable              Variable          Starting          Ending
Variable               type                 width             column           column

STANUM               Numeric                  5                  2                6
YEAR                 Numeric                  4                  9               12
RFLAG                Character                1                 14               14
JAN                  Numeric                  5                 16               20
JANFL                Character                1                 22               22
FEB                  Numeric                  5                 24               28
FEBFL                Character                1                 30               30
MAR                  Numeric                  5                 32               36
MARFL                Character                1                 38               38
APR                  Numeric                  5                 40               44
APRFL                Character                1                 46               46
MAY                  Numeric                  5                 48               52
MAYFL                Character                1                 54               54
JUN                  Numeric                  5                 56               60
JUNFL                Character                1                 62               62
JUL                  Numeric                  5                 64               68
JULFL                Character                1                 70               70
AUG                  Numeric                  5                 72               76
AUGFL                Character                1                 78               78
SEP                  Numeric                  5                 80               84
SEPFL                Character                1                 86               86
OCT                  Numeric                  5                 88               92
OCTFL                Character                1                 94               94
NOV                  Numeric                  5                 96              100
NOVFL                Character                1                102              102
DEC                  Numeric                  5                104              108
DECFL                Character                1                110              110

where

STANUM           is the WMO station number;

YEAR             is the year of the data; and

JAN-DEC          are the monthly mean temperatures, given in tenths of degrees Celsius.


Flag Codes for the Data

RFLAG is a flag denoting whether any observations in the record are considered suspect and/or have been flagged. The codes are as follows: X = one or more temperatures have been flagged; Z = the record is not recommended for use due to a spurious series of at least three consecutive monthly values; and blank = the record contains no data flags. JANFL-DECFL are flags denoting suspect temperature values. The codes are as follows: A = outlier value*; departs from typical monthly trends; B = outlier value; likely caused by an error in sign; C = outlier value; repeated the following month; D = outlier value; identical to previous month's value; E = outlier value; F = value is identical to the previous month's value; G = this station (No. 54852) shows identical July and August temperatures over the period 1973-1975; H = this station (No. 57411) shows two pairs of repeated temperatures over the period May-August 1969. The values are not statistical outliers, but their pattern makes them suspect; and I = this station (No. 58734) has repeated temperatures for October-December 1945; the December value is an outlier. (*Defined as lying 3.5 or more standard deviations away from the mean value of the variable.)

205-STATION MONTHLY PRECIPITATION DATA SET (prec205.dat)

The precipitation data set from the 205-station network contains monthly precipitation totals. The data are sorted by station number, with each record containing data for one year, including station number, year, and precipitation totals for January-December (in tenths of millimeters).
The file may be read using the following FORTRAN format:
     INTEGER STANUM,YEAR,P(12)
     CHARACTER*1 RFLAG, PFLAG(12)
     READ (5,100,END=99) STANUM, YEAR, RFLAG, (P(I),PFLAG(I), I=1,12)
 100 FORMAT(2I6,1X,A1,12(I6,1X,A1))

or by using the SAS format:

 INPUT STANUM 2-6 YEAR 9-12 RFLAG $ 14
 JAN 16-20 JANFL $ 22 FEB 24-28 FEBFL $ 30 MAR 32-36 MARFL $ 38
 APR 40-44 APRFL $ 46 MAY 48-52 MAYFL $ 54 JUN 56-60 JUNFL $ 62
 JUL 64-68 JULFL $ 70 AUG 72-76 AUGFL $ 78 SEP 80-84 SEPFL $ 86
 OCT 88-92 OCTFL $ 94 NOV 96-100 NOVFL $ 102 DEC 104-108 DECFL $ 110

Stated in tabular form, using variable names from the SAS format, the contents include the
following:

                   Variable             Variable           Starting         Ending
Variable             type                width              column          column

STANUM             Numeric                 5                  2                6
YEAR               Numeric                 4                  9               12
RFLAG              Character               1                 14               14
JAN                Numeric                 5                 16               20
JANFL              Character               1                 22               22
FEB                Numeric                 5                 24               28
FEBFL              Character               1                 30               30
MAR                Numeric                 5                 32               36
MARFL              Character               1                 38               38
APR                Numeric                 5                 40               44
APRFL              Character               1                 46               46
MAY                Numeric                 5                 48               52
MAYFL              Character               1                 54               54
JUN                Numeric                 5                 56               60
JUNFL              Character               1                 62               62
JUL                Numeric                 5                 64               68
JULFL              Character               1                 70               70
AUG                Numeric                 5                 72               76
AUGFL              Character               1                 78               78
SEP                Numeric                 5                 80               84
SEPFL              Character               1                 86               86
OCT                Numeric                 5                 88               92
OCTFL              Character               1                 94               94
NOV                Numeric                 5                 96              100
NOVFL              Character               1                102              102
DEC                Numeric                 5                104              108
DECFL              Character               1                110              110


where

STANUM          is the WMO station number;

YEAR            is the year of the data; and

JAN-DEC         are the monthly precipitation totals, given in tenths of millimeters.


Flag Codes for the Data

RFLAG is a flag denoting whether any observations in the record have been flagged as being suspect. The codes are as follows: X = one or more precipitation totals have been flagged; and blank = the record contains no data flags. JANFL-DECFL are flags denoting suspect precipitation values. The codes used are: R = total is repeated over consecutive months. Totals less than or equal to 1.0 mm repeated over two or more consecutive months were not flagged. If all entries in records containing repeated totals contained zeros in their tenths place, the repeated totals were only flagged if they were greater than or equal to 10.0 mm. All non-zero totals repeated over three consecutive months were flagged, regardless of magnitude. H = total has been deemed unrealistically high.

10. REFERENCES

Eischeid, J. K, H. F. Diaz, R. S. Bradley, and P. D. Jones. 1991. A comprehensive precipitation data set for global land areas. DOE/ER-69017T-H1. Carbon Dioxide Research Division, U.S. Department of Energy, Washington, D.C.

Karl, T. R., and P. M. Steurer. 1990. Increased cloudiness in the United States during the first half of the twentieth century: Fact or fiction? Geophysical Research Letters 17(11):1925-28.

Kaiser, D. P. 1993. Cloud amount and sunshine duration in the People's Republic of China, 1954-88. Eighth Conference on Applied Climatology, 17-22 January 1993, Anaheim, CA. American Meteorological Society.

Riches, M. R., Zhao Jianping, W.-C. Wang, and Tao Shiyan. 1992. The U.S. Department of Energy and the People's Republic of China's Academy of Sciences Joint Research on the Greenhouse Effect: 1985-1991 Research Progress. Bulletin of the American Meteorological Society 73: 585-594.

Tao Shiyan, Congbin Fu, Zhaomei Zeng, and Qingyun Zhang. 1991. Two long-term instrumental climatic data bases of the People's Republic of China. ORNL/CDIAC-47, NDP-039. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, Tennessee.


APPENDIX A

INSTRUMENTATION AND DATA COLLECTION METHODS
OF STATIONS IN THE PRC 65-STATION CLIMATE DATABASE

BAROMETRIC PRESSURE

Fortin and Kew-pattern barometers and an aneroid barograph are employed. They are positioned in observation rooms to decrease temperature variability. The two instruments are positioned near each other, away from doors and windows, and are calibrated at regular intervals. Station pressures are recorded in millibars, with monthly means being calculated from daily means and reported to the nearest tenth of a millibar. Mercury column heights, expressed in millimeters, had been reported before 1953 for some stations, but they have been converted to millibars. Since 1950, readings have been corrected and reduced for temperature effects, instrumental error, and gravity error (including latitude and altitude effects). Readings were not corrected for gravity effects prior to 1950.

TEMPERATURE

Dry- and wet-bulb thermometers are fixed together in an instrument shelter. The standard bulb heights used prior to 1950 are not known. For the period 1954-1960 the standard height was 2 m. The standard height used since 1961 is 1.5 m, as it was for the period 1950-1953. Observing times have varied over the years. The observing times for each station prior to 1950, along with details on which observations were used in calculating the daily mean, may be found in the accompanying station histories (see Table 1). Since 1950, observations have been made using Local Standard Time (LST) (1951-1953), Local Mean Solar Time (LMST) (January 1954-July 1960), and Beijing Time (BT) (August 1960-present). Maximum and minimum thermometers are laid horizontally in the same shelter, 3 and 2 cm higher, respectively, than the dry- and wet-bulb thermometers. Prior to 1951, observations and adjustments were performed at 2000 BT, but since 1951 have taken place at 2000 LST.

RELATIVE HUMIDITY

A ventilated psychrometer and hair hygrometer (models not given) are housed in separate shelters. Readings from each instrument are translated into relative humidities via their respective tables or charts. Temperature observations are made simultaneously. Circa 1953, certain stations (not identified) used a different humidity table which resulted in errors of 1-2%.

PRECIPITATION

A Dines tilting-siphon rain gauge and a siphon rainfall recorder are used. Observation times prior to 1950 are not known. Since then, observations have been taken at midnight LST (1951-1953), 1900 LMST (January 1954 - July 1960), and 2000 BT (August 1960 - present). Observed totals may include liquid precipitation, liquid equivalent of solid precipitation, or frost, dew, or fog. Amounts less than 0.05 mm are recorded as zero.

SUNSHINE DURATION

CAS states that prior to 1954, a Jordan Photographic Sunshine Recorder was used, and since 1954 a Campbell-Stokes Sunshine Recorder has been used. Also, some monthly sunshine durations have been estimated, but these entries have not been flagged or documented.

CLOUD AMOUNT

Cloud amount has been estimated in tenths of sky cover by observers. Zero indicates clear skies; 10 indicates overcast skies. Monthly mean cloud amounts are obtained from averaging mean daily amounts for the month, multiplying by 10 to obtain a percentage, and rounding this value to the nearest whole percent. CAS states the data contain many estimates, but, as with the sunshine data, these have not been identified via flagging or documentation.

WIND DATA

Since 1954, an EL electric wind direction and speed device and a Dines wind direction and speed recorder have been used. Receivers for wind instruments are mounted on a pole 10-12 m above ground level. The type of instrument used before 1954 is unclear, but at the following stations the Beaufort Scale was employed:

ZhangYe (52652)
LanZhou (52889)
YanTai (54765)
LaSa (55591)
TengChong (56739)
TianShui (57006)
ZhengZhou (57083)
YiChang (57461)
ChangSha (57679)
GuiYang (57816)
AnQing (58424)
WenZhou (58659)
FuZhou (58847)
XiaMen (59134)
WuZhou (59265)
ShanTou (59316)
HaiKou (59758)

The Beaufort Scale entries from these stations have been converted to m/s. Observation times used in calculating the daily mean are unclear for the period prior to 1951. Several different time groups were used over the period 1951-1953 (see station histories via Table 1). For the period January 1954-July 1960, observations from 0100, 0700, 1300 and 1900 were used (assumably LMST), and since August 1960, observations from 0200, 0800, 1400, and 2000 have been used (assumably BT). Wind direction is recorded in 22.5° increments of azimuth clockwise from north (sixteen directions), with 0° indicating calm winds and 360° indicating a north wind. The monthly dominant wind direction is simply the most frequent wind direction observed during the month, and its frequency (percentage of the time it was observed) is the dominant wind frequency.