Historical Soviet Daily Snow Depth Version 2 (HSDSD)

Summary

The Historical Soviet Daily Snow Depth Version 2 (HSDSD) product on CD-ROM updates and replaces HSDSD Version 1. The new version contains an additional ten years of data, improved data quality control, an HTML interface, and a Java tool for data browsing and extraction. Data now span from 1881 (for the earliest operational stations) through 1995. Other parameters include snow cover percent, snow characteristics, site characterization, and quality flags. Data are in ASCII format.

The HSDSD product is based on observations from 284 World Meteorological Organization (WMO) stations throughout Russia and the former Soviet Union. The area covered is 35 to 75 degrees north latitude and 20 to 180 degrees east longitude. The State Hydrometeorological Service in Obninsk, Russia, sent the data to NSIDC through the bilateral US-USSR Working Group Eight data exchange program.

The development of this data product was funded through NOAA's Environmental Services Data and Information Management (ESDIM) program, and through the support of the NOAA/NESDIS National Geophysical Data Center. Data are available on one CD-ROM. The CD-ROM contains a slightly different version of this documentation.

Citing These Data

Citations for use of these data should read:

Armstrong, R. 2001. Historical Soviet daily snow depth version 2 (HSDSD). Boulder, CO: National Snow and Ice Data Center. CD-ROM.

Overview Table

Category	Description
Data format	ASCII
Spatial coverage and resolution	35^oN to 75^oN, 20^oE to 180^oE Resolution varies; irregular sampling locations
Temporal coverage and resolution	1881 through 1995, daily measurements
Tools for accessing data	HTML and Java browse and extraction tools on CD-ROM
Parameter(s)	Snow depth
Procedures for obtaining data	To request a copy of the CD-ROM, complete the Data Order Form. Corrected mean snow depth climatology data are available via FTP.

1. Contacts and Acknowledgments
2. Detailed Data Description
3. Data Access and Tools
4. Data Acquisition and Processing
5. References and Related Publications
6. Document Information

1. Contacts and Acknowledgments

Investigator(s) Name and Title

Richard Armstrong
National Snow and Ice Data Center
CIRES, 449 UCB, University of Colorado, Boulder, CO, USA 80309-0449

Technical Contact

NSIDC User Services
National Snow and Ice Data Center
CIRES, 449 UCB
University of Colorado
Boulder, CO 80309-0449 USA
phone: +1 303.492.6199
fax: +1 303.492.2468
form: Contact NSIDC User Services
e-mail: nsidc@nsidc.org

2. Detailed Data Description

Data are stored in ASCII text files. See individual data sections below for descriptions of formats.

Daily Data
Seasonal Summary Data
Monthly Climatology Data
Monthly Summary Data
Synoptic Data

Daily Data

File names for daily data contain the letters "dat" and the WMO station number (for example, "dat22887.dat"). Data fields are fixed length and comma-delimited with a total record length of 26 characters. Each day appears as a separate record with the following parameters:

Date of observation - year, month, day
Snow cover and quality of this observation
Characteristic of site and quality of this observation
Snow depth (in cm) and quality of this observation

Sample Data Records for dat22887.dat:

1966, 1, 1,0,1,0,20, 23, 0

1966, 1, 2,0,1,0,20, 23, 0

1966, 1, 3,0,1,0,20, 28, 0

For the first record:

1966	:	Year
1	:	Month
1	:	Day
0	:	Snow cover percentage (SCP) These data were not subject to additional quality control at NSIDC; suspect values may exist.
1	:	Snow cover characteristic (SCC)
0	:	Snow cover quality (SCQ)
20	:	2 = Location characteristic (LC) 0 = Location quality (LQ for the quality of the site characterization)
23	:	Snow depth in cm (SD)
_0	:	_= Snow depth characteristic (SDC) In this case, the field is blank. 0 = Snow depth quality (SDQ) If only one zero exists in this field, the user should read it as zero for both fields.

Codes for sample data records:

`Year`	`Mon`	`Day`	`SCP`	`SCC`	`SCQ`	`LC/` `LQ`	`SD`	`SDC/` `SDQ`
`1966`	`1`	`1`	`0,`	`1,`	`0,`	`20,`	`23,`	`0`
`1966`	`1`	`2`	`0,`	`1,`	`0,`	`20,`	`23,`	`0`
`1966`	`1`	`3`	`0,`	`1,`	`0,`	`20,`	`28,`	`0`

Detailed Field Description:

Year
Month
Day
Snow cover percentage (SCP): Percentage of ground covered by snow, determined by visual examination of exposed ground. Values range from 0 to 9, indicating 0% to 90% coverage. A value of "0" in this field and a "1" in the "Snow cover characteristic" field indicates 100% snow. This field contains a "9" if the value is invalid (indicated by a "3" or a "4" in the "Snow cover quality" field). Column SCP in Table 1 lists values for the snow cover state for given SCC and SCQ values.
Snow cover characteristic (SCC): Characteristics of the snow cover estimate. This field contains a "9" if the snow cover state measurement is invalid (indicated by a "3" or a "4" in the "Snow cover quality" field). See Column SCP in Table 1 for valid measurement values.
Snow cover quality (SCQ): Quality of snow cover measurements. A value of "0" in this field indicates that the snow cover estimate is reliable. See column SCQ in Table 1 for other possible values.
Location characteristic (LC): Indicates characteristics of the location where snow depth is measured. Values range from 1 to 7. See column LC in Table 2 and Table 3 for a list and description of possible values.
Location quality (LQ): Indicates the reliability of the location characteristic observation. See column LQ in Table 3 for a list and description of possible values.
Snow depth (SD): Depth measured in cm by permanent snow stakes placed at specially chosen sites. Column SD in Table 4 shows possible snow depth values for given values of SDC and SDQ.
Snow depth characteristic (SDC): Used to describe characteristics of the snow depth measurement. If the SDC field is blank, the value is "0". If the SDC field is "5", the value was originally derived from humidity data (not snow depth); therefore, SD is artificially set to "0." These data are left in the record to maintain a uniform time series and to indicate the type of original data; however, they can safely be ignored. See column SDC in Table 4 for possible values.
Snow depth quality (SDQ): Reliability of the snow depth measurement. See column SDQ in Table 4.

Table 1. Examples for fields SCP, SCC, and SCQ.
("N" is any number between 0 and 9.)

Situation	SCP	SCC	SCQ
Value is reliable
Snow cover is from 0 to 9 (0% to 90%)	N	0	0
Snow cover is 100%	0	1	0
Temporary snow melting	0	2	0
Value is suspect	N	0	2
Value is rejected	9	9	3
Observation was not made	9	9	4

Table 2. Meaning of values found in field LC.

Code - LC	Site Characteristic
1	Site is protected from strong wind
2	Site is open to wind
3	Observations are made at two sites
4	There is no information about the site's characteristic
5	Site is not protected and not always open
7	Type of site is not specified and therefore snow depth is not measured
9	Site characteristic value was rejected or observations were not made.

Table 3. Examples of values found in fields LC and LQ, and field LC values for given values of LQ.
("N" is any site characteristic code; any number between 1 and 9. See Table 2).

Situation	LC	LQ
Value is correct	N	0
Type of the site is not specified; therefore, snow depth is not measured	7	0
Continuous snow melting (summer)	0	0
Value is suspect	N	2
Value is rejected	9	3
Observations were not made	9	4

Table 4. Example snow depth values (column SD) for given values of SDC and SDQ, and snow depth and SDC values for given values of SDQ.
("NNN" indicates any snow depth measurement in cm.)

Situation	SD	SDC	SDQ
Value is correct (C3 = 0 or blank)	NNN	0	0
Snow cover is less than 0.5 cm	000	4	0
Snow cover absent at site; however, snow is in the vicinity and a coverage is specified.	000	3	0
Value suspected to be humidity - set to 000 or 999	000	5	0
Snow depth was identified as high by a factor of ten and was rescaled by 0.1	NNN	6	0
Snow depth was identified as low by a factor of ten and was rescaled by 10	NNN	7	0
Continuous snow melting (summer)	000	1	0
Temporary snow melting	000	2	0
Value is suspect	NNN	0	2
Value is rejected	999	9	3
No observations	999	9	4

Seasonal Summary Data

NSIDC computed the first and last dates of existing snow. All years (1881 through 1995) are included even if no data were reported. File names for seasonal summary data contain the letters "sea" and the WMO station number (for example, "sea22887.dat"). Data fields are fixed length and comma-delimited with a total record length of 34 characters. All data files are in the "seasonal_summary" directory and contain the following fields:

Hydrological year
Day of the first recorded snow depth
Month of the first recorded snow depth
Year of the first recorded snow depth
Day of year (DOY) of the first recorded snow depth
Day of the last recorded snow depth
Month of the last recorded snow depth
Year of the last recorded snow depth
Day of year (DOY) of the last recorded snow depth

Sample Data Records:

1965,29,10,1964,303, 3, 5,1965,123
1966,13,10,1965,286,30, 4,1966,120
1967,15,10,1966,288,20, 4,1967,110
1968,23,10,1967,296,30, 4,1968,121

A hydrological year runs from October through the following September. For example, the hydrological year 1978 runs from October 1977 through September 1978. All other years in the seasonal summary data refer to the standard calendar year.

Monthly Climatology Data

NSIDC generated monthly climatological data for a 30-year time series (1966-1995). The data are stored in fixed length, comma-delimited fields with a total record length of 46 characters. The first record contains the station number for the file. Records 2 through 11 contain the data fields. The monthly climatology file names contain the letters "clm" and the WMO station number (for example, "clm22887.dat"). The files are stored in the "climatology" directory. The following fields are included:

First record:

WMO Station Number

Second and continuing records:

Month
Mean depth - monthly mean depth in cm
Mean standard deviation depth - monthly mean standard deviation depth in cm
Number of valid days - number of valid days used in the above calculations
Mean maximum depth - mean maximum snow depth for this month in cm
Mean maximum standard deviation - monthly maximum standard deviation depth in cm
Number of mean maximum depth measurements
Mean minimum depth - mean minimum snow depth for this month in cm
Mean minimum standard deviation - monthly minimum standard deviation depth in cm
Number of mean minimum depth measurements

Sample Data Records for clm22887.dat

38987 
 1  8.457,  7.758,  269, 16.538,  11.079, 26,  1.130,  0.344,  23

The first record indicates data for station 38987. The second record contains the following fields:

1 - Month (January)
8.457 - Mean depth in cm for January
7.758 - Mean standard deviation depth in cm for January
269 - Number of valid days throughout the year
16.538 - Mean maximum depth in cm for January
11.079 - Mean maximum standard deviation depth in cm for January
26 - Number of mean maximum depth measurements throughout the year
1.130 - Mean minimum depth in cm for January
0.344 - Mean minimum standard deviation depth in cm for January
23 - Number of mean maximum depth measurements throughout the year

Data Manipulation:

The monthly climatological data are calculated using all observations from 1966 to 1995 for each individual month. The mean depth and standard deviation are generated with data from all valid days in a given month.

The mean maximum and minimum depth are calculated based on the number of months of valid data between 1966 and 1995. Because each month has only one maximum and minimum, a maximum of 30 values are averaged for these fields.

The number of valid and missing days inform the user of the number of days of data. Large numbers of missing days indicate that the station closed or that its data records were not included in the latest update (1985 through 1995).

NSIDC discovered errors in these snow depth climatology files. Corrected data are available via ftp.

Monthly Summary Data

These data include the number of days where snow depth was greater than 5 cm, 10 cm, or 50 cm. File names for monthly summary data contain the letters "sum" and the WMO station number (for example, "sum22887.dat"). All data fields are fixed length and comma-delimited. The total record length is 16 characters. The data files are in the "monthly" directory. A value of "-2" indicates missing data.

The data fields are as follows:

Year
Month
Number of days with greater than or equal to 5 cm of snow depth
Number of days with greater than or equal to 10 cm of snow depth
Number of days with greater than or equal to 50 cm of snow depth

Sample Data Record:

1966, 1,31,31, 0
1966, 2,28,28,10
1966, 3,31,31,28

Synoptic Data

The synoptic data product includes data for single days from every station from 1966 through 1995. These data are intended for validation of remote sensing or other projects that require data for a single day over a large spatial extent. All data files are in the "synoptic" directory. The file name contains the year, month, and day of the data contained in each file (for example, "19831226.dat"). Each field is fixed length and comma-delimited with a total record length of 37 characters. The data in each record include:

WMO station number
Latitude, longitude, and elevation
Snow cover and quality of this observation
Characteristic of site and quality of this observation
Snow depth and the quality of this observation

Sample Data Records for 19891221.dat:

20674 73.5  80.4 0047 0,1,0,20, 12, 0
20891 72.0 102.5 0033 0,1,0,20, 19, 0
21946 70.6 147.9 0048 0,1,0,20, 18, 0

For the first record:

20674	:	WMO station number
73.5	:	Station latitude
80.4	:	Station longitude
0047	:	Station elevation
0	:	Snow cover percentage (SCP)
1	:	Snow cover characteristic (SCC)
0	:	Snow cover quality (SCQ)
20	:	2 = Location characteristic (LC) 0 = Location quality (LQ for the quality of the site characterization)
12	:	Snow depth in cm (SD)
_0	:	_ = Snow depth characteristic (SDC). In this case the field is blank. 0 = Snow depth quality (SDQ). If only one zero exists in this field, the user should read it as zero for both fields.

Example codes for sample data records:

`WMO` `#`	`Lat`	`Lon`	`Alt`	`SCP`	`SCC`	`SCQ`	`LC/` `LQ`	`SD`	`SDC/` `SDQ`
`20674`	`73.5`	`80.4`	`0047`	`0,`	`1,`	`0,`	`20,`	`12,`	`0`
`20891`	`72.0`	`102.5`	`0033`	`0,`	`1,`	`0,`	`20,`	`19,`	`0`
`21946`	`70.6`	`147.9`	`0048`	`0,`	`1,`	`0,`	`20,`	`18,`	`0`

Detailed Field Description:

WMO station number
Station latitude in degrees North
Station longitude in degrees East
Station elevation in m above sea level
Snow cover percentage (SCP): Percentage of ground covered by snow, determined by visual examination of exposed ground. Values range from 0 to 9, indicating 0% to 90% coverage. A value of "0" in this field and a "1" in the "Snow cover characteristic" field indicates 100% snow. This field contains a "9" if the value is invalid (indicated by a "3" or a "4" in the "Snow cover quality" field). Column SCP in Table 1 lists values for snow cover for given SCC and SCQ values.
Snow cover characteristic (SCC): Characteristics of the snow cover estimate. This field contains a "9" if the snow cover measurement is invalid (indicated by a "3" or a "4" in the "Snow cover quality" field). See Column SCP in Table 1 for valid measurement values.
Snow cover quality (SCQ): Quality of snow cover measurements. A value of "0" in this field indicates that the snow cover estimate is reliable. See column SCQ in Table 1 for other possible values.
Location characteristic (LC): Indicates characteristics of the location where snow depth is measured. Values range from 1 to 7. See column LC in Table 2 and Table 3 for a list and description of possible values.
Location quality (LQ): Indicates the reliability of the location characteristic observation. See column LQ in Table 3 for a list and description of possible values.
Snow depth (SD): Depth measured in cm by permanent snow stakes placed at specially chosen sites. Column SD in Table 4 shows possible snow depth values for given values of SDC and SDQ.
Snow depth characteristic (SDC): Used to describe characteristics of the snow depth measurement. If the SDC field is blank, the value is "0". If the SDC field is "5", the value was originally derived from humidity data (not snow depth); therefore, SD is artificially set to "0." These data are left in the record to maintain a uniform time series and to indicate the type of original data; however, they can safely be ignored. See column SDC in Table 4 for possible values.
Snow depth quality (SDQ): Reliability of the snow depth measurement. See column SDQ in Table 4.

Spatial Coverage

WMO stations are primarily located in inhabited areas within the middle latitudes of Eurasia: 35 to 75 degrees north latitude and and 20 to 180 degrees east longitude.

Spatial Resolution

Resolution varies; data were collected from irregular sampling locations. See a list of weather stations.

Temporal Coverage

Data span from 1881 (for the earliest operational stations) through 1995.

Temporal Resolution

Data were collected daily.

Quality Assessment

Systematic data gaps (caused by breaks in station function or punch card deterioration) are set to default values. To estimate the quality of the data, each snow depth value in the archive is followed by characteristic (C) and quality (Q) flags set to one of the following values:

CQ
00: valid value
02: suspect value
93: rejected value
94: observation was not made

Field positions are filled with "9" for quality flag values of "3" or "4". NSIDC added additional flags in the characteristic field to indicate data with errors that were fixed during processing. Data flagged in this manner have a quality value of "0", indicating valid data. The numeric flags are as follows:

50 - Humidity data
60 - Snow depth high by a factor of ten
70 - Snow depth low by a factor of ten

The quality flag is set to "99" for invalid dates such as 30 February.

NSIDC checked the locations of the meteorological stations against the WMO station directory for close correlation.

NSIDC developed a quality control routine to correct gross errors within the data. This routine differs from that used in Version 1 of the Historical Soviet Daily Snow Depth; therefore, Version 2 data supersede all Version 1 data.

NSIDC developed an improved quality control routine for HSDSD Version 2, based on input from several users of HSDSD Version 1. Results were compared to manually-corrected Version 1 raw data independently developed by Dr. David Robinson at Rutgers University.

The new algorithm checks for data entries outside of two standard deviations from the monthly mean and examines these entries for the following known errors:

Factor of ten high
Factor of ten low
Missing humidity flag
Misplaced summer melt flag

Corrections to Factor of Ten Errors

NSIDC noted anomalously large snow depth fluctuations -- often more than a normal snowfall -- and determined that the data changed by a factor of ten, for example, from 11 cm one day to 110 cm the next.

The Russian and USSR data collection instructions required snow depth in cm and precipitation in mm. NSIDC discovered errors in transcribing or reporting the snow depth, for example, confusing the units of snow depth and precipitation in the original data; therefore, if a change in snow depth was greater than or equal to two standard deviations from the mean, a factor of ten error was investigated. If the value had changed by approximately a factor of ten, NSIDC rescaled the data by 0.10.

Other data appeared to be a factor of ten lower than data from the surrounding days. If the value had changed to a single digit and could be explained by a factor of ten decrease, NSIDC rescaled the value by ten.

For both types of factor of ten errors, NSIDC checked the corrected value for plausibility by evaluating the difference between the mean daily change for the previous day and that of the corrected current day.

Humidity Flag Corrections

The completed HSDSD data product contains no humidity values; however, the raw data uses the same field for both snow depth and humidity data, depending on the season. A data quality flag of "50" indicates that humidity data were recorded. This flag was occasionally missed during digitization, resulting in an invalid snow value. If a value deviated more than ten times from the mean for a given month, the value was assumed to represent humidity data and was flagged accordingly. NSIDC set the value of these data records to 000. Any record with a characteristic and quality code combination of 50 is suspect and should be ignored.

Misplaced Summer Melt Flag

A characteristic and quality flag combination of "10" indicates summer melting. The value "00010" represents a valid snow measurement during the summer months; however, errors in the digitization process might yield an entry of "00100" that would indicate a valid 1 cm snow depth measurement. NSIDC fixed this problem by taking the indication of summer as truth; therefore, if a day indicated summer melting ("10") then all values less than 1 cm were corrected to 0 cm depth with a characteristic and quality flag combination of 02 to indicate a suspect value.

Monthly Products Quality Control

Valid snow depths with a characteristic and quality flag combination of 50 were used for the development of the monthly products. These flags indicate a non-zero and valid snow depth measurement.

Errors in Snow Depth Climatology

In October 2001 NSIDC discovered errors in mean and standard deviation depth values for climatology data files. Mean snow depth values were computed using erroneous data. Mean maximum and minimum depth were calculated correctly.

NSIDC computed a corrected mean snow depth climatology, available via ftp. The statistics in the reprocessed files were calculated without using zero snow depth values; therefore, they represent mean snow depth only when snow was present.

3. Data Access and Tools

Data Access

To request a copy of the CD-ROM, complete the Data Order Form. Corrected mean snow depth climatology data are available via FTP.

Volume

File sizes vary by data type:

Daily data: 1.1 MB per file
Seasonal summary data: 4 KB per file
Monthly climatology data: 1 KB per file
Monthly summary data: 23 KB per file
Synoptic data: 22 KB per file

Software and Tools

The Historical Soviet Daily Snow Depth Version 2 CD-ROM contains an HTML interface and Java tool for extracting and browsing data.

Related Data Collections

4. Data Acquisition and Processing

Data Acquisition Methods

Observers took daily measurements of snow depth from three snow measuring rods placed in a meteorological enclosure. In individual cases when the enclosure was not representative of the surroundings with respect to snow cover formation, they placed the snow measuring rods near the station on a specially selected plot.

Observations were made on each rod to 1 cm precision. The observers remained 2 m to 3 m away from the rod while making measurements. The measured snow depth was indicated by the line on the rod that was level with the snow surface. The observers also read the layer of ice or water on the rod that often formed after snow melted near the rod.

If the reading was less than 0.5 cm, they recorded it as 0 cm. If the depth was greater than 0.5 cm but less than 1.0 cm, they recorded it as 1 cm.

The observers calculated the daily mean by averaging the readings on the three rods, rounded to the nearest whole cm. If the mean was less than 0.5 cm, they recorded it as 0 cm; if the mean was greater than 0.5 cm but less than 1 cm, they recorded it as 1 cm.

Data Source

NSIDC derived the HSDSD product from the Soviet meteorological archive, Sytochny Temperatur, Osadki, Sneg (STOS), which contains daily temperature, precipitation, and snow depth data from Russian World Meteorological Organization (WMO) stations. The State Hydrometeorological Service in Obninsk, Russia, sent the data to NSIDC through the bilateral US-USSR Working Group Eight data exchange program.

The original data set was a compilation of the following three meteorological archives:

TM1-Day archive with daily meteorological observations for the USSR between 1874 and 1965
Day-76 meteorological archive for the entire USSR between 1966 and 1984
Day-76 meteorological archive for the Taiga region between 1966 and 1984

The latest update from 1985 through 1995 comes from the Day-76 meteorological archives and includes 222 stations.

NSIDC checked the WMO station latitude and longitude of the original data with those listed in the WMO Handbook to determine the station name and elevation. The WMO station position was used where slight differences between the two existed. Six stations had WMO numbers that did not match any station within the WMO directory. The original data may have contained typographic errors such as 24759 for station 24959. NSIDC corrected these station numbers to match the WMO station listing based on the station name, latitude, longitude, and elevation. Corrected station numbers are identified by an asterisk (*). The list of stations summarizes station locations in the data set, ordered by WMO number.

ASCII	American Standard Code for Information Interchange
ESDIM	Environmental Services Data and Information Management
ftp	File Transfer Protocol
HSDSD	Historical Soviet Daily Snow Depth
HTML	Hypertext Markup Language
NESDIS	National Environmental Satellite, Data, and Information Service
NOAA	National Oceanic and Atmospheric Administration
NSIDC	National Snow and Ice Data Center
WMO	World Meteorological Organization