Annotations for

Monthly Discharges for 2400 and Streams of the former [FSU]

v1.1, September, 2001

Byron A. Bodo [email protected] Toronto, Canada

Disclaimer

Users assume responsibility for errors in the and stream discharge data, associated metadata [river names, gauge names, drainage areas, & geographic coordinates], and the annotations contained herein.

No doubt errors and discrepancies remain in the metadata and discharge records. Anyone data set users who uncover further errors and other discrepancies are invited to report them to NCAR.

Acknowledgement

Most discharge records in this compilation originated from the State Hydrological Institute [SHI] in St. Petersburg, . Problems with some discharge records and metadata notwithstanding; this compilation could not have been created were it not for the efforts of SHI. The University of New Hampshire’s Global Hydrology Group is credited for making the SHI Basin data available.

Foreword

This document was prepared for on-screen viewing, not printing !!! Printed output can be very messy.

To ensure wide accessibility, this document was prepared as an MS Word 6 doc file. The www addresses are not active hyperlinks. They have to be copied and pasted into www browsers.

Clicking on a page number in the Table of Contents will jump the cursor to the beginning of that section of text [in the MS Word version, not the pdf file].

Distribution Files

Files in the distribution package are listed below:

Contents File name

short abstract abstract.txt ascii description of distribution files & formats readme site catalogue & summary statistics fsu.inv (ascii) fsu_inv.xls (spreadsheet) monthly discharge data fsu2400.q notes fsu2400.doc [this file]

2

Updates: September 2001

• 8 records added from Dettinger & Diaz (2000) GHCDN compilation • 2 records updated with 6-8 months missing data • minor metadata fixes

• the 8 new [yellow-shading ; green shading ] records and 2 updated [blue-shading] records are listed below:

ID # River Gauge km2

97065 Poronai Krasnyi Oktybr A 6,080 97066 Bolshaya Bira Birobidzhan A 7,560 97067 Pokrovka A 370,000 97068 Zeya Belogor'ye A 229,000

97069 Chagodoscha Megrino E 7,330 97070 Vetluga E 22,200 97071 Kyn E 10,400 97072 Iput' Ushcherp'ye E 8,100

91138 Solza (Nizhnyaya Solza ?) Sukhie Porogi E 1,190 96166 Pechenga Pechenga E 1,680

• all sites are in Russia • the new Asian records are most helpful

3

Table of Contents

Disclaimer...... 2 Acknowledgement ...... 2 Foreword...... 2 Distribution Files ...... 2 Updates: September 2001...... 3 Table of Contents...... 4 Part I — Data Set Synopsis ...... 6 Introduction...... 6 Data Set Summary ...... 6 Geographic Coverage ...... 9 Data Sources:...... 10 ds553.1 ...... 10 State Hydrological Institute [SHI]...... 10 ARC [R-Arcticnet] ...... 11 SHI-2 / SHI-3 ...... 12 SHI-2 ...... 12 SHI-3 ...... 12 SHI-4 ???? ...... 13 GHCDN ...... 13 Order of Precedence ...... 13 Caveats ...... 14 Part II — Analysis...... 15 Data and Metadata Quality...... 15 Metadata Quality...... 15 Gauge Identification Codes...... 15 FSU Gauge ID Codes ...... 15 Data Source Codes ...... 16 Country Codes...... 16 River and Gauge Names...... 17 Gauge Locations ...... 17 Drainage Areas ...... 17 Drainage Area Discrepancies — River Basin ...... 18 Drainage Area Discrepancies — Kyrgyzstan...... 18 @ Motravn (Motravi) ...... 19 Gauge Elevations ...... 19 Metadata Quality Checks...... 19 Results ...... 20 Discharge Data Quality ...... 20 Discharge Data Quality Checks ...... 21 A. Replicate Scans ...... 21 Kuzreka @ Kuzreka...... 21 B. Scans for Egregious Extremes...... 22 C. Inter-Data Set Matching ...... 22 D. Digital Accuracy Issues...... 23 E. Potential Problems ...... 23 Replicates...... 23 Akavret @ Sikhalidzeebi / Chirukhistskali @ Shuakhevi ...... 23 Bazarka @ Bazar ...... 24 Bolshoy Nimnyr @ Bolshoy Nimnyr...... 24 Chuksha @ ??? ...... 24 Dnestr @ Zaleschiki ...... 24

4

Kacha @ Bashtanovka / Kacha @ Suvorovo...... 24 Karakol @ mouth...... 25 RuCh.Protoka @ Demin Erik / RuCh.Protoka @ Slavyansk-Na-Kubani...... 25 Severnaya Un'ga @ Panfilovskoye / Yuzhnaya Un'ga @ Panfilovskoye ...... 25 Ulu-Uzen @ Solnechnogorskoye...... 26 Vozhega @ Nazarovskaya ...... 26 Appendix A — FSU Basin/Gauge Codes ...... 27

5

Part I — Data Set Synopsis

Introduction

River and stream discharge data for the republics of the former Soviet Union [FSU] are useful for many purposes. FSU data have gradually become available through various sources that are limited in geographic temporal coverage, and are in different formats that are not readily integrated. Available data sets also suffer certain numerical errata in discharge records and metadata. River and gauge names are given in unconventional English forms that are often radically inconsistent with English norms for FSU place names; and furthermore, are inconsistent between alternate data sources.

This compilation is an easily accessible merged version of the readily available FSU discharge data bases. Discharge records are free of the most egregious redundancies and numerical errata. Many river and gauge names have been resolved to forms that are identical or sufficiently close to conventional English forms that gauge names and locations can be searched in gazetteers and other geographic data bases.

This compilation nominally contains records for 2,458 gauges. Some records (50- 100+) are of negligible use due to short records, missing metadata, etc. These have been retained in the event that future updates may resolve the omissions. Unwanted records are easily eliminated from the set as desired.

Data Set Summary

• nominally, the data set has records for: gauges — 2,458 records — 94,297 (station years) net years — 89,187.9 (total months / 12) mean record length — 36.3 yrs

• the table below summarizes the distribution of record lengths • 541 (23%) gauges have 50 or more years of data • 1,724 (70%) gauges have 25 or more years of data

† net yrs sites %

y < 5 75 3

6

5 ≤ y <10 183 7 10 ≤ y <25 476 19 25 ≤ y <50 1,183 48 50 ≤ y <100 527 21 100 ≤ y 14 1

Total 2,458 100

† net years = total months / 12

• data currency is summarized in the table below • 2,282 gauges have records ending in 1985 or later • only 326 gauges have record ending in 1990 or later; however, at least for Russia, there seems to be an appreciable lag before data become available after collection

End yr gauges %

1991 ≤ y < 1995 38 2 1986 ≤ y < 1990 871 35 1981 ≤ y < 1985 1,373 56 1976 ≤ y < 1980 94 4 1971 ≤ y < 1975 36 1 1966 ≤ y < 1970 19 1 1961 ≤ y < 1965 13 1 1936 ≤ y < 1960 14 1

2,458

• gross station year counts give another view of the data set • for each unique calendar year in the data set, the total months of data are for all sites are divided by 12 to give a measure of the temporal data density which is also a rough surrogate for the net active station density in any year

• as the plot below shows, the active station density peaks in 1981, declines modestly to 1985, and steeply thereafter • the steep decline post-1985 reflects primarily a lag in data processing and availability; however, there may have been a significant reduction in active stream gauging following the break-up of the FSU

• the gross active station density determined as the number of stations reporting at least one month of data in a calendar year is 5-7% higher than the net active station density through the most active gauging period of 1965–1985

7

2200 2000 Net Active Station Density by Year 1800 s

r 1600

yea 1400 n o

i 1200 t a t

s 1000 t

e 800 n 600 400 200 0 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000

• the cumulative distribution of gauges by drainage area is given below for 2,439 gauges that have drainage areas given in the metadata files • 640 gauges represent catchments ≥ 10,000 km2 • 1,055 (43%) gauges represent watersheds of 1,000 to 9,999 km2 • 632 (26%) gauges represent watersheds of 100 to 999 km2

2 km sites %

≥ 1,000 ,000 16 1 ≥ 100,000 139 6 ≥ 10,000 640 26 ≥ 1,000 1,695 69 ≥ 100 2,327 95 > 0 2,439 100

8

Geographic Coverage

Figure 1. Geographic distribution of river and stream gauges in the present compilation. Data sources: red dots — ARC (R-Arcticnet); magenta circles — SHI-3; magenta squares — NCAR ds553.0; bright green squares — UNESCO files. Data for 81 of 87 UNESCO sites have been superseded by data from the other sources. 8 sites added in v1.1 are not shown.

• gauge distribution is shown on Figure 1 • the approximate distribution by country is given below • country assignments are mainly as given in the source files • numerous obvious errata were corrected; more are likely present • re-assignment of countries with good national boundary files may change the tallies somewhat

Country ISO Gauges

Russia RU 1,849 KZ 175 UA 86 Kyrgyzstan KG 83 Georgia GE 60 Belarus BY 39 Armenia AM 28 TJ 27 Estonia EE 22

9

Azerbeijan AZ 21 UZ 20 Lithuania LT 19 Latvia LV 18 Turkmenistan TM 9 Moldova MD 2

Data Sources:

Data were compiled from the sources listed below:

1. ds553.1 316 FSU gauge records in NCAR ds553.1 which is a compilation derived from mainly from NCAR ds553.0 and ds552.0; for details see notes accompanying ds553.1 for [www.scd.ucar.edu/dss/datasets/ds553.1.html]

2. ARC / 1,494 gauges in the FSU Arctic prepared by Russia’s State Hydrological SHI-1 Institute [SHI] and available at R-Arcticnet [www.R-Arcticnet.sr.unh.edu/]

3. SHI-2 a set of 547 FSU gauge records prepared by SHI [espejo..org.uy/index.html]

4. SHI-3 a set of 1200+ FSU gauge records prepared by SHI; not currently available on-line

5. GHCDN a global set containing old UNESCO records available in ds553.1 plus 83 records from another source

ds553.1

• 316 gauge records in this set were reviewed rigorously • 289 gauge records originated from the Russian hydrometric agency headquartered in Obninsk [ds553.0]; thus, most records are for gauges in the Russian Federation • 252 gauge records, that are common to one or more of the other three sources, are generally superior versions having have longer time series, more precise low discharges (<1 m3/s), more reliable metadata and corrections that have only been partially addressed in the data sets originating via SHI • these 252 gauge records have already incorporated post-1985 data that are available from ARC and SHI-3 • 64 records are not available in the SHI data sets; 58 of these came from ds553.0 and 6 originate from the old UNESCO files [ds552.0]

State Hydrological Institute [SHI]

10

• SHI is the State Hydrological Institute of Russia located in St. Petersburg • contrast of data originating from the Russian hydrological monitoring service headquartered in Obninsk with gauge records common to SHI data sets shows that some errata have crept into discharge records, likely during transfers from Obninsk to SHI • records from Obninsk often extended back further in time; while numerous SHI records had some more recent (post-1985) data not available in the data set from Obninsk

ARC [R-Arcticnet]

• A large set of discharge records for 1,486 unique gauge records in the Russian / FSU Arctic drainage basin is available on ARC [http://www.R-Arcticnet.sr.unh.edu/] courtesy of Professor Charles Vörösmarty and colleagues at The University of New Hampshire. These data originated via SHI.

• data at ARC often have up to 5 years or more of additional data after 1985 not available at gauges common to ds552.0 and ds553.0

• ARC data for some sites still have some errata identified in ds553.1

• ARC data report numerous missing values for Arctic streams that were identified as legitimate 0s according to the data qualifier codes supplied with ds553.0 which originated from the source agency at Obninsk; the false missing values can inflate long term mean annual discharge and specific runoff estimates, and lead to other distortions in time series modelling and analysis

• due to rounding, data in ARC have lower digital precision than data from ds553.0; this is only of limited consequence for low discharge streams with frequent monthly discharges < 1 m3/s

• ARC metadata have location coordinates that differ somewhat from those specified in other sources, including other data sets from SHI • Kola Peninsula and White drainage basin gauges have numerous bad location coordinates [these were reported to R-Arcticnet and may have been fixed]

• ARC gives many river and gauge names in unconventional English transliterations that bear little resemblance to the English names given in US or British sources; hence, it is often impossible to find these sites in international gazetteers and other geographical name data bases

• moreover, the romanizations used for river and gauge names in ARC differ from those used in other SHI data sets; hence, the only definitive way to determine which

11

site records match those in other data sets is to run matching algorithms using correlations and other similarity measures

• 118 gauges in ARC have no gauge name

SHI-2 / SHI-3

• SHI-3 is a superset of SHI-2 • at the time of writing, only SHI-2 was available on-line for anyone who wished to corroborate the data

SHI-2

• this set of 547 FSU gauge records is part of a global discharge data set accompanying an internet version [espejo.unesco.org.uy/index.html] of the SHI monograph “World water resources at the beginning of the XXI st century” prepared for UNESCO Division of Water Sciences in 1998.

• 234 SHI-2 gauge records are also available in ARC, and 135 are available in ds553.1

SHI-3

• this set of 1,206 gauge records was evidently an update of the SHI-2 global discharge set • SHI-3 embraces the 547 records in SHI-2, but some metadata, specifically river and gauge names differ slightly from those given in SHI-2 • 540 SHI-3 gauge records are not available in ARC, SHI-2, or ds553.1 • like SHI-2, discharges are more crudely rounded than FSU and ARC

• SHI-3 accompanied draft documents prepared for the World Water Forum held in The Hague, March 2000 • SHI-3 appeared on the World Water Vision www site from about December 1999 to February 2000 • the entire global discharge data set was removed pending corrections because of corruption affecting non-FSU gauge records; however, the promised replacement has yet to materialize • examination of the FSU gauge records in SHI-3 including cross-checks against the ca. 35% of gauge records common to the other FSU data sets showed that these data were not affected by the problems afflicting non-FSU gauge records

• sites common to SHI-2 and ARC generally have the same data, including the same errata found at numerous sites in the preparation of ds553.1

12

• metadata differ somewhat from ARC • record end dates suggest that these data were likely processed before ARC (1999) • many river and gauge names are given in English forms that are not easily recognizable, but closer to English norms for Russian names than ARC

• discharges are more crudely rounded than in ARC or data given by the source agency in Obninsk • at some sites, low discharges (<1 m3/s) are truncated at one digit to the right of the decimal, i.e., at tenths of m3/s • at low discharge sites or for low discharge months at other sites, monthly discharges could be perceptibly underestimated, e.g., if a discharge of 0.19 is truncated to 0.1 m3/s • however, on average, long term monthly means may only be 5% too low if discharges are predominantly <1 m3/s

SHI-4 ????

• another collection of SHI data has appeared here webworld.unesco.org/water/ihp/db/shiklomanov/part'4/index4_2.html • due to bad html links this is presently (Sep 2001) not readily accessible • however, data were obtained earlier and there is nothing new in this particular subset of FSU discharge records

GHCDN

• the Global Hydroclimatic Data Network [V1.0a, 8 Sept 2000] compilation by Dettinger & Diaz (2000) • includes FSU sites from UNESCO files plus 83 records from another source • only 8 sites are new, two others have 6-8 months missing in the v1.0 of this compilation

Dettinger, M.D., and Diaz, H.F. 2000. Global characteristics of stream flow seasonality and variability. J. Hydrometeor. 1(4):289–310.

Order of Precedence

• the general order of precedence for gauges with records in two or more sources was: ds553.1 > ARC > SHI-3

13

Caveats

• available discharge records and metadata for the FSU are not as reliable as those produced by leading national agencies such as the US Geological Survey [USGS]

• discharge records and metadata do contain scattered errata

• data quality analyses of the present data set were limited to low level scans to identify the most egregious numerical errata in discharge records and metadata, and to match narrative metadata (river and gauge names) to available geographic place / feature name data bases

• less severe numerical errata such as those uncovered by more exhaustive and rigorous data quality analyses of ds553.1 are likely present in the 2,134 gauge records drawn from other sources • the annotations for ds553.1 illustrate the kinds of errata likely to be present

• despite scattered errata that may remain, the overall quality of the data set is generally good (say B+ relative to the A+ rating that might be assigned to USGS discharge records)

14

Part II — Analysis

Data and Metadata Quality

Metadata Quality

• inaccurate, corrupted or unconventional FSU gauge metadata, specifically location coordinates and river and gauge names, pose more serious problems than discharge data quality • most metadata problems pertain to unconventional English transliterations of river and gauge names; and mis-specified gauge location coordinates

Gauge Identification Codes

• all gauges in the set were arbitrarily assigned ID codes in the 90000 series • these are used in the discharge data file to identify records

FSU Gauge ID Codes

• all but three sites in ds553.1 and all sites in ARC had the old 4-5 digit FSU gauge numbers • these numbers are generally unique and, broadly, by the 1st 1-2 digits give the hydrographic basin of the gauge [see Appendix A] • when available, these numbers facilitate matching gauges to other FSU data sets; e.g., cross-referencing against ARC

• there are scattered cases where the same data have appeared under two different codes; these appear to be cases where the gauge has been moved slightly and new codes assigned, but the same data have shown up under both codes • known cases are listed below:

code code river gauge

1492 1594 Amguema mouth of Shoumny Bk (Ust'e Shumny) U Mosta 174 km

3288 3629 Vost. Khandyga Zapa

3332 3579 Churkuo Lavinda

3552 3881 Alazeya Argakhtakh

9048 9049 Krasnoyarskaya Ges Divnogorsk

15

10251 10252 Tom Tomsk

11037 11667

11407 11408 Ishym Sergeievka Sergeievka Ges

75309 75308 Orel (Kostamarovo)

• SHI-2 / SHI-3 did not give the FSU ID codes, but FSU codes for ca. 470 gauges were available from assorted FSU gauge lists

• the metadata file includes the 5-digit FSU gauge codes for 1,715 gauges, plus three sites from ds553.1 that had been assigned the pseudo FSU ID codes below

code river gauge

75999 Yeltsy RU EU 84999 Kara Samur Luchek RU EU 85999 Vorotan Urut AZ EU

Data Source Codes

• the metadata spreadsheet file “FSU_inv.xls” has data source codes • 1,910 records can be traced to on-line sources for anyone that wishes to corroborate those data • 540 gauge records from SHI-3 cannot presently be accessed on-line

Country Codes

• the present metadata file includes 2-letter ISO country codes for each gauge • there were scattered discrepancies in the countries assigned to gauges between SHI-2 and SHI-3; and scattered errors in the country assignments of SHI-3

• if country is important, country assignments should be determined with GIS using good national boundary files

16

River and Gauge Names

• most metadata problems involve ad hoc, inconsistent English transliterations of river and gauge names to forms that often differ so much from conventional English norms for FSU place names that the sites cannot be found in English language gazetteers and geographic name data bases • moreover, English transliterations of river and gauge names are often inconsistent from one data set to another to the extent that certain sites are unrecognizable as being the same entities; and some discharge records have entirely different gauge names in different data releases

• more such problems can be expected as the 14 non-Russian FSU republics release gauge lists with anglicized versions of river and gauge names as known in the national languages that may bear little resemblance to the anglicized Russian names used for these entities in the FSU era • this is evident in short lists of selected gauges obtained for the Baltic republics (Estonia, Latvia, Lithuania) and Kazakhstan

• 118 gauges from ARC have no gauge name

Gauge Locations

• gauge location coordinates given for FSU gauges are usually within 30 km of the true location; most are likely within 10 km of true location • there are often discrepancies, usually small, between location coordinates of nominally identical gauges given by different data sets and gauge lists • metadata usually have a small scattering of more extreme errata • because the river and gauge names are frequently in unconventional forms, it can be difficult or impossible to corroborate the locations of gauges that do not plot on the river network or otherwise appear to be mis-located

Drainage Areas

• the quality of these data is not known, but appears reasonable for most gauges • drainage areas were generally accepted except for the cases below from SHI-3: a) certain large discrepancies in drainage areas cited for some Ob River basin gauges [see below] b) 22 mis-scaled drainage areas for gauges in Kyrgyzstan c) miscellaneous errata at 13 other sites mostly in Russia d) small discrepancies for 24 gauges that were resolved by averaging alternative areas from different gauge lists [see sheet “area_discrepancies” in spreadsheet “fsu_inv.xls”]

17

• if high resolution digital data, i.e., DEMs and Digital Line Graphs (DLGs) of the river channel networks, are available, the drainage areas should be checked; the FSU has vast areas of arid and swampy terrain where the drainage areas given in the present metadata may not be accurate

Drainage Area Discrepancies — Ob River Basin

• large discrepancies exist between certain main branch and large gauges in the Ob River basin

• the differences of up to 530,000 km2 in drainage areas for the total Ob basin and the main downstream gauge Ob @ Salekhard are apparently due to inclusion/exclusion of endorheic drainage areas in the Irtysh tributary system (including the subcatchment), which includes a sizable area of internal closed drainage and is bounded on the south by closed drainage areas of

• for the present metadata, the lower drainage areas (mainly from ARC) have been accepted

River Gauge area km2

accepted alternates min max

Ob Salekhard 2,430,000 2,550,000 2,950,000

Irtysh 321,000 769,000

Irtysh Ust- 564,000 1,060,000

Irtysh 969,000 1,490,000

Tobol Kurgan 98,800 159,000

Tobol Yalutorovsk 177,000 241,000

Tobol Lipovka / Lipovskoye 359,000 423,000

Drainage Area Discrepancies — Kyrgyzstan

• at 20 of 84 Kyrgyzstan gauges, drainage areas were scaled down by 100-fold and rounded to the nearest integer • at 2 others, drainage areas were scaled down by 10-fold and rounded to the nearest integer • these were identifiable by having mean annual specific runoff >5 m well in excess of the typical range: 100–1,100 mm

18

• drainage areas given in independent sources further confirm that SHI-3 drainage areas are too low by 100-fold for 7 gauges

• the areas of the following 6 gauges can only be determined to one digit accuracy; hence, the given areas may be somewhat in error

River Gauge km2 potential range

Shankol Shankol 60 55–64 Karakol Koschan 100 50–149 Zerger Kischl.Tassay 200 150–249 Kekdzherty Aktala 200 150–249 Dzhergetal Dzhergetal 300 250–349 Karakol mouth 500 450–549

Yazgulem @ Motravn (Motravi)

2 ID FSU ID River Gauge km mm

90340 Yazgulem Motravn (Motravi) TJ 194 5,887

• the drainage area given for this site in Tajikistan was changed to 1,940 km2 • specific runoff at neighbouring sites ranged from 85–1,320 mm • the drainage area should be corroborated

Gauge Elevations

• gauge elevations are given for 1,131 gauges • these data come from assorted FSU gauge lists and SHI-3 • data quality is unknown; use with caution • N.B. negative gauge elevations are legitimate in the vicinity of the which is below m.s.l.

Metadata Quality Checks

• location coordinates and river and gauge names for the 2,134 sites not in ds553.1 were checked against assorted FSU gauge catalogue files and geographic feature name / location data sets for Russia and several FSU republics • some sites were manually checked against US NIMA’s (National Imagery and Mapping Agency) GEOnet name-server / on-line gazetteer at http://164.214.2.59/gns/html/

• there were cases involving ca. 5 cases for which two, and three in one case, gauges had been assigned the same location coordinates

19

• to prevent 2 gauges from plotting over top each other, the coordinates were jittered slightly (by ca. 1-3 km ) so that the gauges plot as distinct points on clickable maps

Results

• most river and stream names should be identical or near enough to forms used in GEOnet or other English language geographic name data bases

• most location coordinates for the 316 gauges derived from ds553.1 should be within 15 km of true location

• for gauges derived from ARC and SHI-3 that could be verified, most location coordinates for gauges on larger rivers and streams should be within 30 km of true location (most are likely within 15 km of true location)

• ARC had 19 gauges in the Kola Peninsula / drainage basins and ca. 30 other gauges that were perceptibly mis-located • SHI-3 had ca. 25 gauges that were significantly mis-located (>30 km from true location), and ca. 50 that were modestly mis-located (10-30 km from true location)

• most (roughly 70-75%) gauge locations could not be corroborated because available geographic name data bases lacked good coverage of smaller river and streams; and are less complete and accurate for the trans- and central Asian republics [Armenia, Azerbeijan, Georgia, Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan, Uzbekistan] • gauges representing <5,000 km2 drainage area comprise about 63% of the sites in ARC and SHI-3 • gauges representing the trans-Caucasus and central Asian republics comprise about 44% of the unique sites contributed by SHI-3

• if the same proportions hold as for gauges that could be corroborated, 3-6% of the un-verified gauges may be mis-located by 30 km or more, and another 3-6% may be mis-located by 10–30 km.

Discharge Data Quality

• the available hydrologic data for the FSU are not as reliable as those produced by leading national agencies such as the US Geological Survey [USGS] • at the level of primary data collection, FSU discharge data quality likely rivals that of other leading national agencies • however, experience with the data sets comprising the ds553.1 compilation and the present data sets, indicates that FSU discharge records may be corrupted during data entry, during intra- and inter-agency transfers, or data base manipulations

20

• despite scattered problems uncovered in FSU discharge records, data quality is generally good • the over all frequency of errata is relatively low • nonetheless, records may contain scattered aberrant entries, muddled years and other forms of corruption that may affect sensitive analyses; hence, additional precautions such as robust statistical methods, sensitivity analyses, and consistency checks with adjacent gauges are advised when using these data for advanced analyses

Discharge Data Quality Checks

A. Replicate Scans

• separate preliminary scans of ARC and SHI-3 identified several cases of (a) the same site record appearing under different metadata (code numbers or names) and (b) lesser replicated records, generally some years repeated at the same or other sites in the set

• after preliminary scans, obvious cases were resolved:

• ARC had 8 cases of gauge records appearing under two different metadata entries [i.e., 1,494 gauge records before reduction to 1,486 unique records] • SHI-3 had 23 cases of gauge records appearing under two different metadata entries [i.e., 1,229 gauge records before reduction to 1,206 unique records]

• scattered, easily resolvable, cases of the same data appearing at the same site in multiple years or at other sites were corrected in both sets

• a final scan was run on the merged master data file after other quality checks listed below • replicated data that could not be resolved in the final set are listed in a separate section below

Kuzreka @ Kuzreka

• ARC had two records with distinctly different location coordinates labelled “Kuzreka @ Kuzreka” [also in SHI-3] • the 1st (“Kuzreka @ Kuzreka” below) is labelled and located correctly • the 2nd site (“unknown @ unknown” below) had significantly higher discharges from 1950–1969, and identical discharges to “Kuzreka @ Kuzreka” from 1979–1988 • 1950–1969 data are definitely from the southeastern corner of the Kola Peninsula as suggested by the longitude and the sequence of the FSU ID numbers

21

• 1950-1969 data correlate very strongly with the gauge labelled “Strel’na @ Strel’na”; and most likely represent a gauge in the upper Strel’na watershed or adjacent interior watershed with a drainage area of ca. 1,200–1,250 km2 • the 1950–1969 data area retained as “unknown (Strel’na ?) @ unknown”

2 3 ID FSU ID River Gauge Lat Lon km km mm yrs 1st last

91044 71192 Kuzreka Kuzreka 66.63 34.80 250 0.092 368.8 44.0 1946 1990

96191 71182 unknown unknown 66.62 38.00 -9 0.447 -9.0 19.2 1950 1969

91141 71183 Strel'na Strel'na 66.07 38.73 2,770 0.994 359.0 43.5 1935 1978

B. Scans for Egregious Extremes

• ARC and SHI-3 were scanned separately for egregious extrema • scans found some obvious decimal shift typos and other aberrant entries • the latter were likely extraneous numerical junk that had inadvertently filled missing months in the time series at some stage of data processing • egregious extrema were more frequent in SHI-3 records; evidently, ARC which comprises mainly Russian data has been more tightly edited than data for the other FSU republics

• scans used here were is not as rigorous as methods used to validate records in ds553.1; less extreme errata remain undetected • inspection of replicates and other records flagged in data and metadata quality checks revealed that less severe numerical errata similar to those identified in the annotations for ds553.1 were scattered through the records • similar error frequencies are expected for 2,134 gauge records derived from other sources

C. Inter-Data Set Matching

• ARC and SHI-3 were scanned against ds553.1 to identify sites common to ds553.1 • ds553.1 already included updated records from ARC and SHI-3 • the common sites were removed from working files

• the reduced ARC and SHI-3 files were scanned against each other to identify sites common to both • the reduced files have 2,134 site records not found in ds553.1 • these occur as: a) 1,157 sites unique to ARC b) 783 sites unique to SHI-3 c) 194 sites common to ARC and SHI-3

22

D. Digital Accuracy Issues

• low discharge data (<1 m3/s) are variably affected by different rounding practices applied to different contributing data sets • data for Russian gauges obtained from Obninsk have the best precision • ARC is rounded to two digits to the right of the decimal • SHI-3 is nominally rounded to two digits to the right of the decimal; however, some low discharge data were found to have been truncated to one digit to the right of the decimal; in extreme cases, a discharge of 0.19 m3/s could have been given as 0.10 m3/s

• generally, the low precision of some low discharge data may only be of concern for analyses that focus on low discharge streams or low discharge conditions of smaller rivers and streams

E. Potential Problems

• there may be cases of Siberian Arctic and arid gauge records having incorrectly interpreted 0s as missing values, and vice versa • contrast of common Russian records obtained via SHI and the hydrometric agency in Obninsk, shows that SHI has not correctly handled 0s and missing values at some gauges; i.e., there were false missing values that should have been given as 0s, and there may have been some false 0s that should have been given as missing values • there may be other similarly affected records in the set of 2,100+ gauge records obtained via SHI that cannot presently be corroborated against source agency data

• discharge records from ARC and SHI-3 have not been scanned rigorously for errata as ds553.1 and may contain typos and other data entry errors that could not be detected by the low level scans applied to these records

• some replicates (below) remain in the data set

Replicates

• the cases below involve replicated data that remain in the present set because the correct location or year cannot be readily determined

Akavret @ Sikhalidzeebi / Chirukhistskali @ Shuakhevi

ID FSU ID River Gauge

90053 Akavret Sikhalidzeebi GE

23

90067 Chirukhistskali Shuakhevi GE

• 1984 is same at both sites • Akavret @ Sikhalidzeebi is more likely the source of the 1984 data, but evidence is weak

Bazarka @ Bazar

ID FSU ID River Gauge

90929 11085 Bazarka Bazar KZ

• 1981 and 1985 are identical for 11 months

Bolshoy Nimnyr @ Bolshoy Nimnyr

ID FSU ID River Gauge

95140 3234 Bolshoy Nimnyr Bolshoy Nimnyr R U

• 1986 and 1987 have the same data

Chuksha @ ???

ID FSU ID River Gauge

95333 8323 Chuksha ??? R U

• 1984 and 1986 have the same data

Dnestr @ Zaleschiki

ID FSU ID River Gauge

90847 Dnestr Zaleschiki UA

• 1971 and 1972 are identical

Kacha @ Bashtanovka / Kacha @ Suvorovo

ID FSU ID River Gauge

24

90861 Kacha Bashtanovka UA 90862 Kacha Suvorovo UA

• these have the same 1985 data

Karakol @ mouth

ID FSU ID River Gauge

90358 Karakol mouth KG

• except for a typo, 1937 and 1938 have the same data

RuCh.Protoka @ Demin Erik / RuCh.Protoka @ Slavyansk-Na-Kubani

ID FSU ID River Gauge

90710 RuCh.Protoka Demin Erik R U 90712 RuCh.Protoka Slavyansk-Na-Kubani R U

90711 RuCh.Protoka Grivenskaya R U

90603 83801 Tikhovskiy R U

• most data (30 yrs) for gauges 90710 and 90712 are identical • only 9 years are distinctly different

• gauges labelled RuCh.Protoka are located on drainage or irrigation canals in the Kuban delta • without further information about the delta canal network, these gauge records are of practically no use

• “Kuban @ Tikhovskiy” is the last gauge above the delta

Severnaya Un'ga @ Panfilovskoye / Yuzhnaya Un'ga @ Panfilovskoye

ID FSU ID River Gauge

95545 Severnaya Un'ga Panfilovskoye R U 95546 Yuzhnaya Un'ga Panfilovskoye R U

25

• these have identical data for: • Jun–Dec 1961 • Jan–Dec 1969

Ulu-Uzen @ Solnechnogorskoye

ID FSU ID River Gauge

90782 Ulu-Uzen Solnechnogorskoye UA

• 1968 and 1969 are identical

Vozhega @ Nazarovskaya

ID FSU ID River Gauge

96012 Vozhega Nazarovskaya R U

• 1956 and 1965 are identical

26

Appendix A — FSU Basin/Gauge Codes

• FSU basin/gauge codes are summarized in the table below • this is not the most hydrologically rational system that might have been devised

• generally, the basin / hydrologic unit number is the integer left after dividing the gauge code by 1000; e.g., gauge number 77123 is in hydrographic basin 77 • by the FSU system, units 1–17 are roughly considered Asia, and units 18–85 are considered Europe

• the “Rivers” entries are the larger ones found in ds553.1

Unit Watershed Rivers Receiving Waters Code

1 NE , Anadyr, Amguema Arctic , , , ,

2 Kamchatka Kamchatka, Penzhina Pacific Ocean, Sea of Okhotsk,

3 NE Siberia , , Olenek, Omoloy, , E Siberian Sea, Yana, Anabar, Alazeya, Khatanga

4 Island Tym Sea of Okhotsk

5 lower Amur & Amur, , Amgun, Nimelen, Japan Sea, Pacific Ocean independent drainage Razdol'naya to Japan Sea from Primorsky Kray

6 upper Amur Amur, Shilka, Ingoda, Selemdzha, Pacific Ocean , Zeya

7 Baikal , Barguzin, Chikoi, Khilok Baikal > > Yenesei > > Arctic Ocean

8 Angara d/s Baikal Irkut, Iya, Biryusa Angara > Yenesei > Kara Sea > Arctic Ocean

9 Yenesei excluding Kan, Tuba, Norilka, Abakan, Yenesei > Kara Sea > Arctic Angara–Baikal, may Podkamennaya-Tunguska, Ocean include independent Nizhnyaya-Tunguska; may include Kara Sea drainage Pyasina, Taymra & other independent Kara Sea drainage

10 Ob - u/s of Irtysh Tom, Tym, Bia, Bolshoi Yugan, lower Ob > Obskaya Gulf > , presumably Charysh, Chulym, Kiya Kara Sea > Arctic Ocean to Ob source

27

11 Obskaya Gulf lower Ob, Severnaya Sos'va, Irtysh lower Ob > Obskaya Gulf > (except Tobol), Ishym, Konda, , Kara Sea > Arctic Ocean plus Nadym, Taz, Pur & likely other small Obskaya Gulf drainage independent of the Ob proper

12 Tobol trib of Irtysh trib Tobol, Lobva, , , Uy, Irtysh > lower Ob > Obskaya of Ob , Gulf > Kara Sea > Arctic Ocean

13 Turgai & L Tengiz Turgai, Kara Turgai, Lake Tengiz; Kazakhstan closed basins

14 Lake Balkash Sharyn, , , Lake Balkash; Kazakhstan closed basin

15 Lake Kul Kul; Kyrgyzstan closed basin

16 Syr Darya, , closed basin

17 Amu Darya, , , Aral Sea closed basin

19 Ural, Maly Uzen, Caspian Sea, closed basin

41 Pyarnu Pyarnu Gulf of Riga > Baltic >

48 Lake ? / may be > > a local closed basin > Baltic > Atlantic

49 Gulf of Onega Kem, Kovda, Vyg Gulf of Onega > White Sea > > Arctic Ocean

70 White Sea — Onega, Severnaya Dvina, Mezen, White Sea, Sea Pechora, Barents Sea > Arctic Ocean

71 Kola Peninsula Ura, Pecha, Kola, Kitsa, Ponoy, White Sea, Barents Sea > Umba, Kolvitsa Arctic Ocean

72 Gulf of Neva, Lake Ladoga (excluding Lake > Baltic > Onega), Narva, Luga Atlantic

73 Gulf of Riga excluding Gauja (Gauya), Daugava Gulf of Riga > Baltic > Pyarnu Atlantic Ocean

74 Baltic Proper Neman, small tribs, likely Pregolya Baltic > Atlantic

75 upper Volga; roughly Klyaz'ma, Moksha, , Oka, lower Volga > Caspian Sea, above / , Ugra, , Vetluga closed basin Kuybyshev

76 upper Volga – NE tribs , , , , lower Volga > Caspian Sea, affluent to Kuybyshev Chepsta, Ai, , Dema, closed basin reservoir

77 mid & lower Volga; Samura, Bolshoi Kinel & mostly lower Volga > Caspian Sea, roughly from small tribs closed basin to Caspian Sea 28

roughly from small tribs closed basin Kuybyshev reservoir to Caspian Sea

78 Don, Khoper, Severskiy , Sea of > > Medveditsa > Atlantic

79 upper Dnepr; u/s Kiev Dnepr, Pripyat, Vyazma Black Sea > Mediterranean reservoir, mostly in Sea > Atlantic Belarus & Russia

80 lower Dnepr; d/s top of Dnepr, , Seym Black Sea > Mediterranean Kiev reservoir Sea > Atlantic

81 Black Sea northwest Dnestr, Yuzhny Bug Black Sea > Mediterranean coast Sea > Atlantic

82 Black Sea east coast Rioni, likely other tribs entering Black Sea > Mediterranean from Georgia Sea > Atlantic

83 Kuban Kuban > Black Sea > Mediterranean Sea > Atlantic

84 Caspian Sea northwest drainage north of Caucasus divide Caspian Sea, closed basin including Kalaus, Terek, Samur

85 Caspian Sea southwest drainage south of Caucasus divide Caspian Sea, closed basin including Kura, Aras (Arax), L Yerevan, from lands of Azerbaijan, Armenia, Iran, Turkey

29