<<

Recent temperature and precipitation increases in West and their association with the

Karen E. Frey & Laurence C. Smith

Surface air temperature and precipitation records for the years 1958–1999 from ten meteorological stations located throughout West Siberia are used to identify climatic trends and determine to what extent these trends are potentially attributable to the Arctic Oscillation (AO). Although recent changes in atmospheric variability are associated with broad Arctic change, West Siberia appears particularly susceptible to warming. Furthermore, unlike most of the Arctic, moisture transport in the is highly variable. The records show that West Siberia is expe- riencing signifi cant warming and notable increases in precipitation, likely driven, in part, by large-scale Arctic atmospheric variability. Because this region contains a large percentage of the ’s peatlands and con- tributes a signifi cant portion of the total terrestrial freshwater fl ux to the Arctic , these recent climatic trends may have globally signifi cant repercussions. The most robust patterns found are strong and prevalent springtime warming, winter precipitation increases, and strong asso- ciation of non-summer air temperatures with the AO. Warming rates for both spring (0.5 - 0.8 °C/decade) and annual (0.3 - 0.5 °C/decade) records are statistically signifi cant for nine of ten stations. On average, the AO is linearly congruent with 96 % (winter), 19 % (spring), 0 % (summer), 67 % (autumn) and 53 % (annual) of the warming found in this study. Signifi - cant trends in precipitation occur most commonly during winter, when four of ten stations exhibit signifi cant increases (4 - 13 %/decade). The AO may play a lesser role in precipitation variability and is linearly congruent with only 17 % (winter), 13 % (spring), 12 % (summer), 1 % (autumn) and 26 % (annual) of precipitation trends.

K. E. Frey & L. C. Smith, Dept. of Geography, 1255 Bunche Hall, University of California Los Angeles, Los Angeles, CA 90095-1524, USA, [email protected].

The sensitivity of Arctic climate is documented winter temperatures across much of the former in the instrumental record of recent decades and Soviet Union (FSU) since the early 1950s or mid- is evident in the large temperature increases seen 1960s. More recently (1979–1997), winter tem- over land areas from about peratures have warmed at an exceptionally high 40° N to 70° N (Nicholls et al. 1996). In particu- rate of 1 °C/decade in the Eurasian Arctic, which lar, sharp temperature increases from 1966–1995 is contrasted by winter cooling (–1 °C/decade) in are found in and northwest Amer- the (Rigor et al. 2000). ica, with pronounced warming during winter and Although Serreze et al. (2000) show relatively spring (Serreze et al. 2000). Similarly, Fallot large winter temperature increases over wide- et al. (1997) fi nd signifi cant positive trends in spread areas in Eurasia and western North Amer-

Frey & Smith 2003: Polar Research 22(2), xx-xx 1 Fig. 1. Trends in summer mean surface air temperature (°C per decade) from 40° N to 90° N for the years 1966−1995 (modi- fi ed from Serreze et al. 2000; printed with kind permission of Kluwer Academic Publishers, M. Serreze and J. Walsh). ica, warming during spring and summer is ampli- fi ed over West Siberia and the North Slope of Fig. 2. West Siberia, showing locations of the ten meteorologi- (Fig. 1). Northern Eurasia has also expe- cal stations used in this study. rienced slight increases in precipitation over the last 50 years (Groisman et al. 1991), concurrent with signifi cant increases in cyclone density since tude Northern Hemisphere. It is uncertain, how- the mid-1960s for north of 60° N (Serreze ever, whether this recent AO trend is refl ective et al. 1997) and large reductions in level pres- of natural climate variability and/or symptomat- sure over the central Arctic (Walsh et al. 1996; ic of anthropogenic forcing such as lower strato- Serreze et al. 2000). sphere (Volodin & Galin 1998) A primary mode of climate variability in the and emissions to the Arctic is the Arctic Oscillation (AO; Thompson (Fyfe et al. 1999; Shindell et al. 1999). & Wallace 1998). The AO index is determined by Although recent changes in atmospheric vari- the variation of the wintertime leading empirical ability are associated with broad Arctic climate orthogonal function (EOF) of sea level pressure change, West Siberia appears particularly suscep- northward of 20° N and is dependent on the fl uc- tible to warming (Fig. 1). Furthermore, the pre- tuation of atmospheric pressures between the cen- vailing direction (north or south) of moisture fl ux tral Arctic and two weaker centres at about 45° N in the region is highly unpredictable each year, over the Atlantic and Pacifi c basins. The AO is unlike most other parts of the Arctic (Rogers et highly correlated with atmospheric phenomena al. 2001). This regional temperature amplifi ca- throughout the Northern Hemisphere, account- tion and highly unstable direction of moisture ing for ca. 50 % of the winter warming observed fl ux are particularly relevant to the global over Eurasia and ca. 30 % of the winter warm- cycle, owing to the presence of the West Siberian ing seen over the whole Northern Hemisphere peatlands that occupy nearly 400 000 km2 in the for varying record lengths (Thompson & Wal- region (Zhulidov et al. 1997). Northern peatlands lace 1998; Thompson et al. 2000). AO trends are are a major pool of stored carbon in the form of also well correlated with variability in sea level undecomposed plant matter and are a signifi cant pressure, storm tracks and precipitation (Thomp- component of global and son & Wallace 2001). Since the late 1960s, the emission calculations. For example, total carbon AO has exhibited a pronounced positive phase stocks of northern peatlands are currently esti- (Thompson & Wallace 1998, 2001; Thompson mated at 455 Pg C (ca. one-third of the global et al. 2000), which may be partly responsible for pool of soil C; Gorham 1991). In addition, north- the recent found in the High lati- ern peatlands are currently a signifi cant source of

2 Recent temperature and precipitation increases in West Siberia global atmospheric (Roulet et al. 1992; triggering changes in the volume of North Atlan- Panikov 1999), which has been modelled at ca. 20 tic Deep Water (NADW) formation, salinity dis-

Tg CH4/year (Christensen et al. 1996). While cur- tribution and formation (Rahmstorf 1995; rently a net sink or small source of CO2, north- Vörösmarty et al. 2001). ern peatlands may become a signifi cant source In sum, the data of Serreze et al. (2000; Fig. 1) of atmospheric CO2 under a warming climate, suggest amplifi cation of warming in West Sibe- owing to associated reduction in wetness and aer- ria, a region of global signifi cance with respect obic decomposition of peat (Gorham 1991, 1994; to hydrology, carbon storage and greenhouse gas

Oechel et al. 1993). Emissions of CH4 are also exchange. Here, we present meteorological data expected to take dramatic shifts with changes in for the years 1958− 1999 from ten stations locat- wetness (e.g. Laine et al. 1996; Moore et al. 1998). ed throughout West Siberia (Fig. 2, Table 1). The The observed recent climatic trends in West Sibe- stations are geographically located within the ria may therefore have critical consequences for area of maximum warming in Eurasia (Fig. 1). global carbon cycle dynamics. Seasonal and annual time series of temperature West Siberia (Fig. 2) generates large volumes of and precipitation are investigated with the Mann- freshwater runoff to the . Therefore, Kendall test to determine the presence of long- any perturbations in temperature and/or precipi- term trends. The meteorological records are then tation across West Siberia may also affect global- compared with the AO index to determine the scale processes through hydrology. The ’ and components of the observed trends that are line- rivers of West Siberia alone account for arly congruent with the AO. about 35 % of the total terrestrial freshwater fl ux to the Arctic Ocean (Aagaard & Carmack 1989). Warming temperatures and increasing precip- Data and methods itation in West Siberia may alter the timing, volume and distribution of freshwater transport Meteorological and Arctic Oscillation data to the Arctic Ocean, thereby potentially modify- ing Arctic Ocean and global ocean circulation by Meteorological data for the years 1958–1999

Table 1. Meteorological station records used in this study. Average surface air temperature and precipitation over the record period are given for December–February (DJF), March–May (MAM), June–August (JJA), September–November (SON) and December–November (annual).

Temperature (°C) Precipitation (mm) Elev. No. Met. station Location Record (m) years DJF MAM JJA SON Annual DJF MAM JJA SON Annual

Mys Kamen- 68.47° N, 1958– 2 36 –24.1 –14.0 6.7 –5.7 –9.2 75.8 70.3 119.1 107.6 372.4 nyj 73.60° E 1 9 9 3 63.93° N, 1958– Berezovo 27 42 –20.5 –4.3 13.6 –3.3 –3.5 78.2 96.8 197.0 143.7 516.1 65.05° E 1999 64.92° N, 1958– Tarko-Sale 27 42 –23.6 –8.5 12.5 –5.5 –6.2 75.8 88.2 193.0 149.3 504.0 77.82° E 1999 65.78° N, 1960– Turuhansk 37 40 –24.8 –8.0 13.1 –6.5 –6.4 102.6 93.2 186.0 178.3 560.6 87.95° E 1999 - 60.97°N, 1958– 45 42 –18.4 –1.4 15.4 –1.4 –1.3 82.1 92.9 215.0 152.9 540.5 Mansiysk 69.07° E 1999 Aleksand- 60.43° N, 1958– 47 42 –19.6 –2.6 15.3 –2.1 –2.1 63.8 84.7 213.0 133.1 494.6 rovskoe 77.87° E 1999 61.60° N, 1959– Bor 62 41 –22.7 –3.6 15.2 –3.6 –3.6 109.9 103.6 194.1 183.2 591.9 90.00° E 1999 58.15° N, 1958– Tobol’sk 48.5 42 –16.7 0.9 16.5 0.3 0.3 62.8 74.7 192.4 119.7 449.7 68.18° E 1999 56.90° N, 1958– Tara 73 42 –16.9 0.9 16.7 0.4 0.4 58.1 75.9 187.7 109.6 433.0 74.38° E 1999 58.30° N, 1958– Kolpasev 80 42 –18.4 –1.0 16.0 –1.0 –1.0 71.2 89.2 198.0 134.7 493.1 82.90° E 1999

Frey & Smith 2003: Polar Research 22(2), xx-xx 3 from ten West Siberian stations (Mys Kamen- to precipitation records from the FSU in order nyj, Berezovo, Tarko-Sale, Turuhansk, Khanty- to homogenize the time series (e.g. Groisman et Mansiysk, Aleksandrovskoe, Bor, Tobol’sk, al. 1991; Fallot et al. 1997). K1 compensates for Tara and Kolpasev; Fig. 2, Table 1) are present- a change in gauge design implemented in ed. These stations are physiographically simi- the late 1940s to early 1950s. Because the time lar with the exception of Mys Kammenyj, which series used in this study start in 1958, this correc- may experience a greater maritime infl uence tion is not needed. K2 corrects for wind effects than other stations owing to its coastal location. on precipitation measurements, which is partic- Data for the years 1958–1995 were obtained from ularly important during winter months. K2 coef- the Daily temperature and precipitation data fi cients rely on annual mean wind speeds aver- for 223 USSR stations dataset (Razuvaev et al. aged over several years, introducing potentially 1993), available from the National Climatic Data large errors when applied to monthly time series. Center (NCDC). Daily temperature and precipi- K2 coeffi cients were therefore not applied in this tation data for the years 1996–99 were obtained study. K3 corrects for a change in FSU measure- from the NCDC Climate Visualization website ment protocol implemented in 1966, after which (http://lwf.ncdc.noaa.gov/oa/climate/onlineprod/ moisture adhering to gauge walls was includ- drought/xmgr.html/). Six-hourly surface data ed in the precipitation estimate. Therefore, this for Tarko-Sale were purchased directly from the K3 correction was applied to all precipitation NCDC as individual station data. Records after data in this study prior to 1966, using values for 1999 were not examined owing to the abrupt “raw” precipitation time series as recommended increase of missing data after this year. Month- by Groisman & Rankova (2001). The addition of ly AO indices (Thompson & Wallace 2000) were these recommended K3 coeffi cients can signifi - obtained from the Colorado State University cantly alter observed trends in precipitation and Annular Modes website (http://horizon.atmos.co can increase measured values by up to 6.4 % in lostate.edu/ao). Temperature and AO time series summer months and 9.6 % in winter months. were averaged and precipitation time series were Assessing potential errors in temperature and summed into winter (December–February, DJF), precipitation records associated with urbaniza- spring (March–May, MAM), summer (June– tion, land use change and data collection meth- August, JJA), autumn (September–November, ods is a challenging task. Despite industriali- SON) and annual (December–November) time zation that began in the early 1960s throughout series for the years 1958–1999. much of the region, the ten stations chosen for The physical relevance of the AO compared this study are located in relatively small cities, to the North Atlantic Oscillation (NAO; e.g. all with fewer than 100 000 inhabitants and most Hurrell 1995) has been subject to considera- with signifi cantly fewer than 50 000 inhabitants. ble debate (Deser 2000; Ambaum et al. 2001; Jones et al. (1990) found little evidence for signif- Rogers & McHugh 2002). The NAO describes icant urban infl uence on temperature from 1930 a large-scale fl uctuation of atmospheric mass in to 1987 in the western part of , including the North Atlantic region and is a major source West Siberia. Urbanization and land use change of seasonal to interdecadal variability in atmos- can also affect precipitation measurements, pheric circulation throughout the Northern Hem- although these effects are not normally apparent isphere (Hurrell 1995). Although the NAO and until ca. 20 - 40 km downwind of urban centres AO are highly correlated, the AO is found to be (e.g. Oke 1987). Precipitation collection meth- more strongly coupled to surface air tempera- ods are inherently more problematic than those ture fl uctuations over the Eurasian than for temperature. In 1965, precipitation measure- the NAO (Thompson & Wallace 1998). Thomp- ments within the FSU increased from two to four son & Wallace (1998) also suggest that the AO times daily, although this probably did not sig- captures more hemispheric variability in atmos- nifi cantly alter the total amount of precipitation pheric circulation, as opposed to more regional measured. Errors introduced by applying the gen- North Atlantic/European variability. In recogni- eralized K3 wetting correction coeffi cients or not tion of these observations, we focus this study on applying the wind-compensating K2 correction the AO. coeffi cients (as described above) are also a con- Empirical studies show that up to three correc- sideration. Precipitation in the form of may tion factors (K1, K2 and K3) should be applied also be more diffi cult to measure than rain, owing

4 Recent temperature and precipitation increases in West Siberia to the potential for signifi cant problems of under- 1 – α/2 quantile of the standard normal distribu- catch of solid precipitation (e.g. Woo et al. 1983; tion. This study uses an exceedance probability of Yang & Ohata 2001). p = 0.90 (α = 0.2) to establish trend. A slope esti- mator is not used (e.g. Sen 1968), since the aim of Trend analysis and linear congruence with the this study is simply to establish the presence or AO absence of trend. To estimate the potential contribution of the Seasonal and annual time series of surface air AO to observed trends in temperature and precip- temperature and precipitation are analysed using itation, we apply trend analysis similar to Rigor the nonparametric Mann-Kendall test for monot- et al. (2000), Thompson et al. (2000), Kryjov onic trend (Mann 1945; Kendall 1975; Maidment (2002), Rigor et al. (2002) and Wallace & Solo- 1993), which makes no hypothesis about the value mon (2002). The analysis is as follows: 1) all time of a parameter in a statistical density function. series are linearly detrended; 2) for each result- The Mann-Kendall statistic S is given by: ing time series, values of temperature or precip- itation are regressed onto associated AO indices n – 1 n S = Σ Σ (1) (where AO indices are normalized by the series zk t’= 1 t = t’ + 1 standard deviation); 3) the resulting regression coeffi cient is then multiplied by the linear trend where the ranked series zk is generated by fi rst in the associated AO index time series (in units of considering the annual time series yt, t = 1,…, n standard deviations per decade); 4) this product is and comparing each value yt’ , t = 1, n – 1 with all the component of the decadal temperature or pre- subsequent values yt, t = t’ + 1, t’ + 2, …, n and cipitation trend that is “linearly congruent” with applying the following conditions: the AO. Linear congruence does not necessarily imply that the AO is driving the observed vari- > zk = 1 if yt yt’ ance in temperature or precipitation, but it does identify likely connections between them. zk = 0 if yt = y t’ (2) < zk = –1 if yt yt’ Results The S statistic therefore represents the number of positive differences minus the number of nega- Departures of surface air temperature, precipi- tive differences found in yt. For n > 40, the stand- tation and AO index from their associated long- ardized test statistic z is obtained using a normal term record means (Table 1) are shown in Fig. approximation: 3a. Positive values indicate above average anom- alies and negative values indicate below average z = ————S + m (3) anomalies. Both seasonal and annual temperature √Var (S) anomalies for the ten stations are relatively simi- where m = 1 if S < 0, m = 0 if S = 0, and m = –1 if lar to one another for a given year (Fig. 3a), indi- S > 0. Because the Mann-Kendall test is based cating a strong correlation in temperature vari- on ranks of the data only, a correction is needed ability over a relatively broad geographic area. for the effect of data ties on the variance of S. This pattern is much less evident in records of Data ties occur when adjacent entries have the precipitation departures, where little correlation same value or when two or more years of data exists between stations for a given year (Fig. 3b). are absent (missing values are replaced with the Interannual variability in temperature is greatest series mean). The correction is as follows: during DJF (ca. 18 °C) and least during JJA (ca. 4 °C). In contrast, interannual variability in pre- 1 n cipitation is greatest during JJA (ca. 200 mm) Var (S) = — [n(n – 1)(2n + 5) – Σti (ti – 1)(2ti + 5)] (4) 18 i=1 and least during DJF (ca. 100 mm). This largely exists owing to the signifi cantly greater amount where n is the number of tied groups and ti is the of precipitation falling in JJA as compared with number of data in the ith (tied) group. The null DJF (e.g. JJA precipitation averages more than hypothesis (no trend) is rejected at the α signif- twice that of DJF precipitation; Table 1). For this icance level if |z| > z(1 – α/2), where z(1 – α/2) is the reason, July precipitation variability can drive up

Frey & Smith 2003: Polar Research 22(2), xx-xx 5 (a)

Fig. 3. Departures of the Arctic Oscillation (AO) index; (a) surface air temperature; (b, opposite page) precipitation for ten West Siberian stations. Plots are shown for December–February (DJF), March–May (MAM), June–August (JJA), Septem- ber–November (SON) and December–November (annual). Departures are calculated from associated long-term record means.

6 Recent temperature and precipitation increases in West Siberia (b)

Frey & Smith 2003: Polar Research 22(2), xx-xx 7 to a 70 % difference in annual poleward moisture tions, calculated from ordinary least-squares fl ux in this region (Rogers et al. 2001). regression over the associated length of each Of the 50 seasonal and annual temperature record. Nine of ten stations exhibit signifi cant records analysed, 28 (56 %) display trends at sig- warming trends for both MAM and annual tem- nifi cance levels of p ≥ 0.90, all of which display perature records (Tables 2, 3), while statistically positive (warming) trends (Table 2). Of these signifi cant trends in DJF, JJA and SON tempera- 28 records, 14 records display trends at signifi - tures are much less common (Tables 2, 3). Trends cance levels of p = 0.90, 9 records at p = 0.95, and in annual time series range from 0.3 - 0.5 °C/ 5 records at p = 0.99. Table 3 presents correspond- decade and trends in seasonal time series range ing decadal temperature trends for these 28 sta- from 0.1 - 0.8 °C/decade. On average, the steep-

Table 2. Mann-Kendall statistics for seasonal and annual time series (ca. 1958–1999) of surface air temperature and precipitation for ten meteorological stations. Negative values indicate negative trends and positive values indicate positive trends. Signifi - cance levels: p = 0.90 (boldface), p = 0.95 (italicized boldface) and p = 0.99 (underlined, italicized boldface).

Temperature (°C) Precipitation (mm) Met. station DJF MAM JJA SON Annual DJF MAM JJA SON Annual Mys Kamennyj 0.37 1.62 2.47 0.57 1.76 –2.52 –2.41 –2.48 –2.85 –3.99 Berezovo 0.02 1.91 -0.07 0.41 1.02 –0.61 0.91 0.48 –1.13 –0.19 Tarko-Sale 1.01 2.58 1.98 1.33 2.31 2.25 1.06 1.52 –0.84 1.09 Turuhansk 1.02 1.57 0.80 0.97 1.50 1.77 1.70 –0.10 0.37 1.48 Khanty-Mansiysk 0.98 1.59 1.14 1.06 1.61 0.41 0.74 –1.38 –1.15 –1.12 Aleksandrovskoe 1.32 2.17 1.24 1.13 2.34 0.79 1.72 –1.13 –1.01 –0.80 Bor 1.64 1.52 0.33 0.51 1.69 0.03 –0.86 –0.42 0.62 0.15 Tobol’sk 1.08 0.91 1.13 1.41 1.50 0.64 0.52 –0.67 –0.41 0.07 Tara 1.30 1.37 1.80 1.65 2.43 2.49 –0.43 –0.72 0.00 0.12 Kolpasev 1.39 2.08 1.02 1.15 2.54 1.39 –1.00 –0.87 0.26 –0.56

Table 3. Estimated decadal trends in surface air temperature and precipitation (ca. 1958–1999) for those records displaying statistically signifi cant (p ≥ 0.90) Mann-Kendall statistics. Values in parentheses indicate the components of the trends that are linearly congruent with the AO index (as defi ned in the section “Trend analysis and linear congruence with the AO”).

Temperature (°C / decade) Precipitation (% / decade) Met. station DJF MAM JJA SON Annual DJF MAM JJA SON Annual Mys Kamennyj 0.7 (0.2) 0.6 (0.0) 0.5 (0.2) –15 (–1) -10 (-1) –9 (–1) –13 (0) –12 (0) Berezovo 0.6 (0.1) Tarko-Sale 0.8 (0.1) 0.4 (0.0) 0.2 (0.1) 0.5 (0.3) 13 (4) 18 (2) Turuhansk 0.5 (0.1) 0.4 (0.2) 9 (2) 9 (2) 4 (2) Khanty- 0.5 (0.1) 0.3 (0.1) –6 (–1) Mansiysk Aleksand- 0.6 (0.6) 0.6 (0.1) 0.4 (0.2) 7 (1) rovskoe Bor 0.8 (0.8) 0.5 (0.1) 0.4 (0.3) Tobol’sk 0.1 (0.1) 0.3 (0.2) Tara 0.5 (0.5) 0.5 (0.1) 0.3 (0.0) 0.2 (0.1) 0.4 (0.2) 11 (2) Kolpasev 0.7 (0.6) 0.6 (0.1) 0.4 (0.2) 4 (1) Average trend1 0.7 0.6 0.4 0.2 0.4 4 2 1 –13 –4 Average trend2 986 4 Average % of trends congru- 96 % 19 % 0 % 67 % 53 % 17 % 13 % 12 % 1 % 26 % ent with the AO 1 Average incorporating all statistically signifi cant trends. 2 Average excluding the negative precipitation trends at Mys Kamennyj.

8 Recent temperature and precipitation increases in West Siberia est trends are found during DJF (0.7 °C/decade) Surface air temperatures in West Siberia have and the shallowest trends are found during SON increased strongly since 1958 (ca. 0.4 °C/decade (0.2 °C/decade). However, statistically signifi - annually; Table 3) as compared with global mean cant trends are most consistently found in MAM surface air temperature increases over the past (average 0.6 °C/decade) and annual records (aver- 40 years (0.050 - 0.075 °C/decade; Nicholls et al. age 0.4 °C/decade). 1996). West Siberia may therefore be considered Of the 50 seasonal and annual precipitation a region of amplifi ed warming. Annual tempera- records analysed, only 14 (28 %) display trends ture records show statistically signifi cant warm- at signifi cance levels of p ≥ 0.90 (Table 2). Of ing trends at nine of ten stations, ranging from 0.3 these 14 records, 4 records display trends at sig- to 0.5 °C/decade (Table 3). However, these may be nifi cance levels of p = 0.90, 4 records at p = 0.95, primarily driven by even stronger warming during and 6 records at p = 0.99. Statistically signifi cant MAM, the season exhibiting the most prevalent trends are positive for all station records (i.e. pre- warming trends (Table 2). Rates of MAM tem- cipitation increase) except for those at Mys Kam- perature increase average 0.6 °C/decade (ranging menyj, where negative trends are found for all from 0.5 to 0.8 °C/decade; Table 3). This fi nding seasonal and annual records (Table 3). Signifi - is consistent with a broader pattern of warm- cant trends in precipitation are most commonly ing Eurasian springs. Using Russian found during DJF (fi ve stations) and least com- drifting station records for the years 1961–1990, monly found during SON (one station). Season- Martin et al. (1997) fi nd statistically signifi cant al trends range from –15 to +18 %/decade (–21.2 increases in May temperatures of 0.89 °C/decade. to +25.6 mm/decade). Annual trends range from Krijov (2002) fi nds even stronger springtime –12 to 4 %/decade (–59.5 to +20.2 mm/decade) warming (up to ca. 1.4 °C/decade) from 1968– (Table 3). Statistically signifi cant trends in pre- 1997 in northern Russia. Rigor et al. (2000) also cipitation are most commonly found at higher lat- observe spring warming as high as 2 °C/decade in itudes, with all but two of the signifi cant trends the eastern Arctic. In fact, Groisman et al. (1994) occurring at stations north of 60° N. fi nd that Northern Hemisphere temperatures The components of observed trends in temper- have increased more during spring than any other ature and precipitation that are linearly congruent season. These springtime temperature increases with the AO are also presented in Table 3. Corre- have been shown to parallel the retreat of North- lations between the AO and temperature and pre- ern Hemisphere spring snow cover extent and cipitation time series are qualitatively apparent in are likely related with a strong positive feedback Fig. 3. The most robust pattern found is the high (Groisman et al. 1994; Brown 2000). Our fi nd- linear congruence of the AO with non-summer ing of strong springtime warming is corrobo- air temperatures. On average, the AO is linearly rated by time shifts (ca. 1 - 3 weeks) in river ice congruent with 96 % of DJF, 19 % of MAM, 67 % melt onset dates in western and eastern Siberia of SON, and 53 % of annual warming seen in this for the years 1917–1994 (Smith 2000). Addition- study (Table 3). In contrast, none of the summer ally, since the early 1960s, the annual amplitude warming found in this study can be attributed to of the seasonal CO2 cycle has increased by 40 % the AO. The potential contribution of the AO to in the Arctic, likely resulting from an increase precipitation trends is relatively weak (Table 3) in growing season length (Keeling et al. 1996). and is largest for DJF and annual time series, Randerson et al. (1999) suggest that these recent but smallest for SON time series. On average, the high latitude amplitude increases in the season-

AO is linearly congruent with 17 % of DJF, 13 % al CO2 cycle are responding to growing seasons of MAM, 12 % of JJA, 1 % of SON and 26 % of that are lengthened because of warmer springs, annual precipitation trends. not warmer autumns. Time series of the satellite- derived normalized difference vegetation index (NDVI) provide additional evidence of changes Discussion and conclusions in the magnitude and duration of growing seasons in the Arctic and across Eurasia, brought about by The most robust fi ndings of this study are strong warming spring temperatures (e.g. Myneni et al. and prevalent springtime warming, increas- 1997; Zhou et al. 2001). es in winter precipitation and strong association The most evident result from the precipitation of non-summer air temperatures with the AO. records examined is a general increase in winter

Frey & Smith 2003: Polar Research 22(2), xx-xx 9 precipitation throughout West Siberia. It is pos- annual temperature increase, which is consistent sible that since station gauges measure rain with with other studies (Thompson & Wallace 1998; more accuracy than snow, a portion of the pre- Rigor et al. 2000; Thompson et al. 2000). The cipitation increases found during DJF could be AO is also linearly congruent with 19 % of MAM due to a transition from snowfall events to more warming and may thus play a role in creating ear- rain events, thus leading to erroneous observa- lier dates of melt onset. In contrast, none of JJA tions of increasing winter precipitation. How- warming is attributable to the AO, an expected ever, even for stations exhibiting the strong- result since the AO is generally considered to be est increases in DJF precipitation, average air a wintertime phenomenon (Thompson & Wallace temperatures during these months are still well 1998). The AO is also linearly congruent with below freezing throughout the entire record, sug- 17 % of precipitation trends during winter, when gesting that a signifi cant increase in rain events is statistically signifi cant trends are more prevalent unlikely. Although the trends found in DJF pre- than any other season. cipitation may be small in terms of magnitude, Recent intensifi cation of the AO has estab- they are substantial in terms of percent increase. lished climatic conditions that are consistent with For instance, the observed ca. 8.3 mm/decade the strong warming and precipitation increas- increase in DJF precipitation at Tarko-Sale rep- es over West Siberia found in this study. In the resents a 55 % increase in DJF precipitation over past 30 years, Arctic atmospheric circulation pat- the 42-year record. Similar fi ndings of increased terns have shown large deviations from normal winter precipitation have been found over the past with a cyclonic pattern of circulation persisting 20 years in northern Eurasia (Serreze et al. 2000). over the polar region, resulting in unusually low In West Siberia, Aizen et al. (2001) fi nd a positive pressure, strong subpolar westerlies, and warm trend in annual precipitation over the past 100 high latitude temperatures over land (Serreze et years and Vinnikov et al. (1990) fi nd increases al. 2000). Although the strong anticyclone cir- of 14 %/100 years. Winter precipitation increases culation system known as the Siberian High is found in this study contradict the results of Fallot the most pronounced feature of atmospheric cir- et al. (1997), who fi nd no evidence of an increas- culation in the lower troposphere over continen- ing trend in cold season precipitation in the tal during winter, it has experienced pro- former Soviet Union over the last century. How- nounced weakening during the last ca. 20 years ever, time series used in their study end in 1984 (Gong & Ho 2002). Gong et al. (2001) show and thus do not capture the most recent positive that there is a signifi cant out-of-phase relation- anomalies in winter precipitation (Fig. 3b). Also, ship between the AO and Siberian High, where the majority of stations in their study fall south of the negative phase of the AO might dynamical- 60° N, where winter precipitation is not expected ly strengthen the Siberian High and vice versa. to increase in response to positive AO conditions During positive AO conditions, cyclone activi- (e.g. McCabe et al. 2001). In this study, signifi - ty in the Northern Hemisphere shifts poleward, cant precipitation trends are found most com- leading to anomalously strong subpolar wester- monly at higher latitudes, with 12 of the 14 signif- lies, the conditions of which are characterized icant trends occurring at stations north of 60° N. by westerly geostrophic surface winds along Signifi cant precipitation trends are positive for ca. 55° N and strong zonal fl ow extending into all stations with the sole exception of Mys Kam- , resulting in abnormally high surface air menyj, where all records display negative trends. temperatures as as Siberia (Rogers 1997; Mys Kammenyj also tends to be much drier than Clark et al. 1999; Thompson & Wallace 2001). any other station (Table 1), which is likely due to The position of the increased cyclone activity in the infl uence of relatively cold and dry Arctic Siberia favors stronger and more frequent warm, maritime air masses that have not yet gained southerly winds, which are consistent with recent moisture through evaporation enhanced by for- temperature anomalies and early melt (Serreze et ests and wetlands farther south on the continent al. 1995). This northward shift in cyclone activ- (Shahgedanova 2003). ity is important as cyclones are one of the main Strong association of West Siberian air temper- factors setting the variance of temperature, pres- atures with the AO suggests that the recent dom- sure, and moisture in the troposphere on times- inance of the positive AO phase may potential- cales of 2.5 - 10 days (Paciorek et al. 2002). The ly contribute to about half (53 %; Table 3) of the warm winter temperatures in Siberia have in

10 Recent temperature and precipitation increases in West Siberia fact been linked to stronger westerly fl ow and of convective cloudiness is most frequent (Sun stronger intrusions of cyclone warm sectors into et al. 2001). Factors other than the AO should this region (Rogers & Mosley-Thompson 1995; be driving observed summer warming trends, Rogers 1997). A recent increase in cyclone activ- when linear congruence with the AO is zero. June ity has been observed over high latitude Northern through August is termed an “inactive” season Hemisphere regions and Eurasia, coincident with by Thompson & Wallace (2000), when the AO is the recent increase in the AO and northward shift much weaker and the subpolar zonal wind anom- in Northern Hemisphere storm tracks. Paciorek et alies are displaced poleward of their wintertime al. (2002) fi nd an increase in winter cyclone inten- position. Therefore, congruence between the AO sity over Eurasia over the past ca. 50 years. Ser- and climate phenomena during summer months reze et al. (1997) and McCabe et al. (2001) fi nd a is not expected here. signifi cant increase in high latitude cyclone activ- The strong springtime warming and winter ity in the Northern Hemisphere, although only for precipitation increases of the past few decades regions northward of 60° N. This is relatively con- may have important implications for the carbon sistent with the fi ndings in this study, where 12 of storage, fl ux of greenhouse gases, and hydrology the 14 statistically signifi cant precipitation trends in West Siberia. The ca. 40-year warming trends occur at stations north of 60° N. in springtime temperatures have undoubtedly led The recent persistence of Arctic cyclone activ- to an earlier occurrence of spring thaw and an ity has been linked to relatively large reductions increase in growing season length. The effect of (particularly along the Siberian sector) in North- spring warming on melt season length has been ern Hemisphere sea ice cover, which may in part quantifi ed by Rigor et al. (2000), who fi nd that occur because enhanced southerly winds advect spring warming is associated with a lengthening the ice poleward away from the coasts (Serreze et of the melt season by 2.6 days/decade over the al. 1995; Maslanik et al. 1996). Rigor et al. (2002) eastern Arctic Ocean. The timing of melt onset, suggest that these changes in sea ice may in turn and thus the duration of both the growing and the be responsible for recent observed trends in sur- dormant seasons for plants, are strongly linked face air temperatures, by way of increased latent to the carbon balance of Boreal and Arctic sys- heat released during formation of new ice in tems (Goulden et al. 1997) and may be impor- diverging leads and increased heat fl ux through tant for peat accumulation potentials (Nicholson thinner ice. Maslanik et al. (1996) additionally et al. 1996). As there is a large range in interan- suggest that because an increase in cyclone activ- nual variability (up to 6 weeks or more) in the ity favors divergence and shear within the sea timing of freeze and thaw at a given location (e.g. ice pack, resulting open water areas are increas- Frolking et al. 1996), these events not only serve ingly heated owing to decreased . These as climate indicators, but they are crucial for the phenomena are probably most important in the understanding of possible feedback mechanisms winter and spring warming of coastal areas, such in the carbon cycle. as at Mys Kammenyj. Greenhouse gas exchange in peatlands is Factors other than the AO are important in directly related to temperature and wetness con- driving the observed temperature and precipita- ditions. Small changes in water table, tempera- tion trends. The strong interannual and spatial ture, or timing of thaw and senescence all have variability in precipitation may in part be due to the potential to facilitate modifi cation of CO2 the contribution of convection. Convective weath- dynamics in peatlands (Bubier et al. 1998). One er patterns are much more spatially variable than probable effect of climate change is that warmer synoptic-scale patterns and are much less likely spring temperatures and an earlier start to the to correlate well with the AO. Sun et al. (2001) growing season will result in increased plant pro-

fi nd a statistically signifi cant increase in the fre- duction and uptake of CO2 (Moore et al. 1998). quency of convective cloudiness for all seasons A warming climate and associated reduction in over the FSU, which could contribute to the por- wetness could also cause peatlands to become tion of precipitation trends not congruent with the a signifi cant source of CO2 to the atmosphere AO. Increased convection could also contribute (Gorham 1991, 1994; Oechel et al. 1993; Botch et to observed warming through enhanced down- al. 1995; Alm et al. 1999). In contrast, CH4 emis- ward longwave radiation (e.g. Stone 1997), par- sions are expected to decrease with soil drying ticularly during summer when the occurrence (Roulet et al. 1993; Laine et al. 1996), as its pro-

Frey & Smith 2003: Polar Research 22(2), xx-xx 11 duction is optimal under anaerobic waterlogged mate patterns, likely resulting from associated conditions (Valentine et al. 1994; Bubier 1995; warming temperatures and an increase in cyclone Bubier et al. 1995). Under climate change sce- abundance and intensity. Although the predict- narios, CH4 emissions from northern peatlands ed timing is uncertain, sensitivity experiments are more sensitive to changes in wetness and have identifi ed a threshold of freshwater import water table position than in temperature (Roulet to the Arctic Ocean at which point NADW will et al. 1992), although climate-induced changes in cease to form (e.g. Clark et al. 2002; Rahmstorf

CH4 emissions may be dependent on the initial 2002). The increases in winter precipitation over position of the water table (e.g. Heikkinen et al. West Siberia found in this study should affect the 2002) or additional factors such as peatland type volume of freshwater input to the Arctic Ocean (Nykänen et al. 1998). Given similar wetness and may therefore have consequences of global regimes, however, CH4 emissions are enhanced importance. by warmer temperatures (Christensen et al. 1999) and in regions, greater active layer depths (Christensen & Cox 1995; Heyer et Acknowledgements.—Funding for this research was provided al. 2002). CH4 production is also expected to be by the Russian–American Initiative on Shelf–Land Environ- enhanced by permafrost degradation, owing to ments in the Arctic (RAISE) of the National Science Founda- resulting warm, saturated conditions and shifts tion’s Arctic System Science programme (grant number OPP- in vegetation (Bubier et al. 1995). However, from 9818496). Additional funding was provided by a National Aeronautics and Space Administration System Science a greenhouse forcing point of view, the radia- Graduate Student Fellowship (grant number NGT5-30338). tive effect of continued warming of West Siberi- We thank two anonymous reviewers for valuable comments an peatlands is diffi cult to predict. Because CO2 on an earlier version of this manuscript. We also thank J. Walsh and M. Serreze for permission to use Fig. 1. Finally, K. and CH4 dynamics are negatively coupled in wet- lands (e.g. Whiting & Chanton 2001), one might Frey wishes to acknowledge M. Raphael, without whom this study might never have initiated. expect the radiative effects of CH4 to be temper- ed by those of CO2 and vice versa. For example, earlier warming may enhance sedge production, simultaneously enhancing CO2 uptake and CH4 References emissions (Bubier et al. 1995). Should drier con- ditions enhance peat decomposition and CO2 pro- Aagaard, K. & Carmack, E. C. 1989: The role of sea ice and duction, Laine et al. (1996) suggest that this new other fresh water in the Arctic circulation. J. Geophys. Res. source of CO would be completely compensated 94(C10), 14 485−14 498. 2 Aizen, E. M., Aizen, V. B., Melack, J. M., Nakamura, T. & for by decreasing CH4 emissions and increasing Ohata, T. 2001: Precipitation and atmospheric circulation tree-stand biomass storage. Although uncertain- patterns at mid-latitudes of Asia. Int. J. Climatol. 21, ties in the specifi c response of northern peatlands 535−556. to climate change may exist, there is no question Alm, J., Schulman, L., Walden, J., Nykänen, H., Martikain- en P. J. & Silvola J. 1999: Carbon balance of a Boreal bog that perturbations in temperature and precipita- during a year with an exceptionally dry summer. tion have the potential to profoundly affect the 80, 161–174. exchange of atmospheric CO2 and CH4. Ambaum, M. H. P, Hoskins, B. J. & Stephenson, D. B. 2001: Hydrologically, an earlier transition from Arctic Oscillation or North Atlantic Oscillation? J. Clim. frozen to thawed conditions leads to the earlier 14, 3495–3507. Botch, M. S., Kobak, K. I., Vinson, T. S. & Kolchugina, T. P. occurrence of ice breakup and fl ooding in rivers, 1995: Carbon pools and accumulation in peatlands of the thus altering the timing of freshwater delivery former Soviet Union. Glob. Biogeochem. Cycles 9, 37–46. to the Arctic Ocean. Potentially of more impor- Brown, R. D. 2000: Northern Hemisphere snow cover varia- tance, however, is that an increase in winter pre- bility and change, 1915–97. J. Clim. 13, 2339–2355. Bubier, J. L. 1995: The relationship of vegetation to methane cipitation may expand the snow pack water equiv- emission and hydrochemical gradients in northern peat- alence, thus increasing river discharge during the lands. J. Ecol. 83, 403−420. following spring and summer months. Whether Bubier, J. L., Crill, P. M., Moore, T. R., Savage, K. & Varner, evaporation losses from increased temperature R. K. 1998: Seasonal patterns and controls on net ecosys- tem CO exchange in a Boreal peatland complex. Glob. Bio- will offset this increase in discharge is unclear. 2 geochem. Cycles 12, 703−714. Peterson et al. (2002) found a 7 % increase in Eur- Bubier, J. L., Moore, T. R., Bellisario, L. & Comer, N. T. 1995: asian river discharge over the past 70 years that Ecological controls on methane emissions from a northern is consistent with large-scale hemispheric cli- peatland complex in the zone of discontinuous permafrost,

12 Recent temperature and precipitation increases in West Siberia , . Glob. Biogeochem. Cycles 9, 455−470. and during the thawing period. Tellus 54B, 231–249. Christensen, T. R. & Cox, P. 1995: Response of methane Hurrell, J. W. 1995: Decadal trends in the North Atlantic emission from Arctic to climatic change: results Oscil lation: regional temperatures and precipitation. Sci- from a model simulation. Tellus 47B, 301−309. ence 269, 676–679. Christensen, T. R., Jonasson, S., Callaghan, T. V., Havstrom, Jones, P. D., Groisman, P. Y., Coughlan, M., Plummer, N., M. & Livens, F. R. 1999: Carbon cycling and methane Wang, W.-C. & Karl, T. R. 1990: Assessment of urbaniza- exchange in Eurasian tundra ecosystems. Ambio 28, tion effects in time series of surface air temperature over 239−244. land. Nature 347, 169–172. Christensen, T. R., Prentice, I. C., Kaplan, J., Haxeltine, A. & Keeling, C. D., Chin, J. F. S. & Whorf, T. P. 1996: Increased Sitch, S. 1996: Methane fl ux from northern wetlands and activity of northern vegetation inferred from atmospheric

tundra. Tellus 48B, 652−661. CO2 measurements. Nature 382, 146–149. Clark, M. P., Serreze, M. C. & Robinson, D. A. 1999: Atmos- Kendall, M. G. 1975: Rank correlation methods. 4th edition. pheric controls on Eurasian snow extent. Int. J. Climatol. London: Charles Griffi n. 19, 27–40. Kryjov, V. N. 2002: The infl uence of the winter Arctic Oscil- Clark, P. U., Pisias, N. G., Stocker, T. F. & Weaver, A. J. 2002: lation on the northern Russia spring temperature. Int. J. The role of the thermohaline circulation in abrupt climate Climatol. 22, 779–785. change. Nature 415, 863–869. Laine, J., Silvola, J., Tolonen, K., Alm, J., Nykanen, H., Van- Deser, C. 2000: On the teleconnectivity of the “Arctic Oscil- sander, H., Sallantaus, T., Savolainen, I., Sinisalo, J. & lation.” Geophys. Res. Lett. 27, 779–782. Martikainen, P. J. 1996: Effect of water-level drawdown Fallot, J. M., Barry, R. G. & Hoogstrate, D. 1997: Variations on global climatic warming: northern peatlands. Ambio of mean cold season temperature, precipitation and snow 25, 179−184. depths during the last 100 years in the former Soviet Union Maidment, D. R. (ed.) 1993: Handbook of hydrology. New (FSU). Hydrol. Sci. J. 42, 301–327. York: McGraw-. Frolking, S., Goulde, M. L., Wofsy, S. C., Fan, S. M., Sutton, Mann, H. B. 1945: Non-parametric test against trend. Econo- D. J., Munger, J. W., Bazzaz, A. M., Daube, B. C., Crill, P. metrica 13, 245–259. M., Aber, J. D., Band, L. E., Wang, X., Savage, K., Moore, Martin, S., Munoz, E. & Dreucker, R. 1997: Recent observa- T. & Harriss, R. C. 1996: Modeling temporal variability in tions of a spring–summer warming over the Arctic Ocean. the carbon balance of a spruce/ Boreal forest. Glob. Geophys. Res. Lett. 24, 1259–1262. Change Biol. 2, 343–366. Maslanik, J. A., Serreze, M. C. & Barry, R. G. 1996: Recent Fyfe, J. C., Boer, G. J. & Flato, G. M. 1999: The Arctic and decreases in Arctic summer ice cover and linkages to Oscillations under global warming. Geophys. atmospheric circulation anomalies. Geophys. Res. Lett. Res. Lett. 26, 1601–1604. 23, 1677–1680. Gong, D.-Y. & Ho, C.-H. 2002: The Siberian High and climate McCabe, G. J., Clark, M. P. & Serreze, M. C. 2001: Trends in change over middle to high latitude Asia. Theor. Appl. Cli- Northern Hemisphere surface cyclone frequency and inten- matol. 72, 1–9. sity. J. Clim. 14, 2763–2768. Gong, D.-Y., Wang, S.-W. & Zhu, J.-H. 2001: East Asian Moore, T. R., Roulet, N. T. & Waddington, J. M. 1998: Uncer- winter monsoon and Arctic Oscillation. Geophys. Res. tainty in predicting the effect of climatic change on the Lett. 28, 2073–2076. carbon cycling of Canadian peatlands. Clim. Change 40, Gorham, E. 1991: Northern peatlands: role in the carbon 229–245. cycle and probable response to global warming. Ecol. Appl. Myneni, R. B., Keeling, C. D., Tucker, C. J., Asrar, G. & 1, 182–195. Nemani, R. R. 1997: Increased plant growth in the northern Gorham, E. 1994: The future of research in Canadian peat- high latitudes from 1981 to 1991. Nature 386, 698–702. lands: a brief survey with particular reference to global Nicholls, N., Gruza, G. V., Jouzel, J., Karl, T. R., Ogallo, change. Wetlands 14, 206–215. L. A. & Parker, D. E. 1996: Observed climate variability Goulden, M. L., Daube, B. C., Fan, S.-M., Sutton, D. J., and change. In J. T. Houghton et al. (eds.): Climate change Bazzaz, A., Munger, J. W. & Wofsy, S. C. 1997: Physiolog- 1995: the science of climate change. Contribution of Work- ical responses of a black spruce forest to weather. J. Geo- ing Group I to the Second Assessment of the Intergovern- phys. Res. 102(D24), 28987–28996. mental Panel on Climate Change. Pp 137–192. Cambridge: Groisman, P. Y., Karl, T. R. & Knight, R. W. 1994: Observed Cambridge University Press. impact of snow cover on the heat balance and the rise of Nicholson, B. J., Gignac, L. D. & Bayley, D. E. 1996: Peatland continental spring temperatures. Science 263, 198–200. distribution along a north–south transect in the Mackenzie Groisman, P. Y., Koknaeva, V. V., Belokrylova, T. A. & Karl, River Basin in relation to climatic and environmental gra- T. R. 1991: Overcoming biases of precipitation measure- dients. Vegetatio 126, 119–133. ment: a history of the USSR experience. Bull. Am. Meteor- Nykänen, H., Alm, J., Silvola, J., Tolonen, K. & Martikain- ol. Soc. 72, 1725–1733. en P. J. 1998: Methane fl uxes on Boreal peatlands of differ- Groisman, P. Y. & Rankova, E. Y. 2001: Precipitation trends ent fertility and the effect of long-term experimental low- over the Russian permafrost-free zone: removing the arti- ering of the water table on fl ux rates. Glob. Biogeochem. facts of pre-processing. Int. J. Climatol. 21, 657–678. Cycles 12, 53–69. Heikkinen, J. E. P., Maljanen, M., Aurela, M., Hargreaves, Oechel, W. C., Hastings, S. J., Vourlitis, G., Jenkins, M., K. J. & Martikainen, P. J. 2002: and meth- Riechers, G. & Grulke, N. 1993: Recent change of Arctic ane dynamics in a sub-Arctic peatland in northern . tundra ecosystems from a net carbon dioxide sink to a Polar Res. 21, 49–62. source. Nature 361, 520–523. Heyer, J., Berger, U., Kuzin, I. L. & Yakovlev, O. N. 2002: Oke, T. R. 1987: Boundary layer . 2nd edition. Lon- Methane emissions from different ecosystem structures of don: Routledge. the tundra in western Siberia during midsummer Paciorek, C. J., Risbey, J. S., Ventura, V. & Rosen, R. D. 2002:

Frey & Smith 2003: Polar Research 22(2), xx-xx 13 Multiple indices of Northern Hemisphere cyclone activity, northern Eurasia. Oxford: Oxford University Press. winters 1949–99. J. Clim. 15, 1573–1590. Shindell, D. T., Miller, R. L., Schmidt, G. & Pandolfo, L.

Panikov, N. S. 1999: Fluxes of CO2 and CH4 in high latitude 1999: Simulation of recent northern winter climate trends wetlands: measuring, modeling, and predicting response to by greenhouse-gas forcing. Nature 399, 452–455. climate change. Polar Res. 18, 237–244. Smith, L. C. 2000: Trends in Russian Arctic river-ice forma- Peterson, B. J., Holmes, R. M., McClelland, J. W., Vörös- tion and breakup, 1917 to 1994. Phys. Geogr. 21, 46–56. marty, C. J., Lammers, R. B., Shiklomanov A. I., Shiklo- Stone, R. S. 1997: Variations in western Arctic temperatures manov, I. A & Rahmstorf, S. 2002: Increasing river dis- in response to cloud radiative and synoptic-scale infl uenc- charge to the Arctic Ocean. Science 298, 2171–2173. es. J. Geophys. Res. 102(D18), 21769−21776. Rahmstorf, S. 1995: Bifurcations of the Atlantic thermoha- Sun, B., Groisman, P. Y. & Mokhov, I. I. 2001: Recent changes line circulation in response to changes in the hydrological in cloud-type frequency and inferred increases in convec- cycle. Nature 378, 145–149. tion over the and the former USSR. J. Clim. Rahmstorf, S. 2002: Ocean circulation and climate during the 14, 1864−1880. past 120,000 years. Nature 419, 207–214. Thompson, D. W. J. & Wallace, J. M. 1998: The Arctic Oscil- Randerson, J. T., Field, C. B., Fung, I. Y. & Tans, P. P. 1999: lation signature in the wintertime geopotential height and Increases in early season ecosystem uptake explain recent temperature fi elds. Geophys. Res. Lett. 25, 1297–1300.

changes in the seasonal cycle of atmospheric CO2 at high Thompson, D. W. J. & Wallace, J. M. 2000: Annular modes in northern latitudes. Geophys. Res. Lett. 26, 2765–2768. the extratropical circulation. Part I: month-to-month vari- Razuvaev, V. N., Apasova, E. G., Martuganov R. A., Vose, R. ability. J. Clim. 13, 1000–1016. S. & Steurer P. M. 1993: Daily temperature and precipita- Thompson, D. W. J. & Wallace, J. M. 2001: Regional climate tion data for 223 USSR stations, ORNL/CDIAC NDP-040. impacts of the Northern Hemisphere annular mode. Sci- Oak Ridge National Laboratory, Oak Ridge, TN: Carbon ence 293, 85–89. Dioxide Information Analysis Center. Thompson, D. W. J., Wallace, J. M. & Hegerl, G. C. 2000: Rigor, I. G., Colony, R. L. & Martin, S. 2000: Variations in Annular modes in the extratropical circulation. Part II: surface air temperature observations in the Arctic, 1979– trends. J. Clim. 13, 1018–1036. 97. J. Clim. 13, 896–914. Valentine, D. W., Holland, E. A. & Schimel, D. S. 1994: Eco- Rigor, I. G., Wallace, J. M. & Colony, R. L. 2002: Response of system and physiological controls over methane production sea ice to the Arctic Oscillation. J. Clim. 15, 2648–2663. in northern wetlands. J. Geophys. Res. 99(D1), 1563−1571. Rogers, A. N., Bromwich, D. H. & Sinclair, E. N. 2001: The Vinnikov, K. Y., Groisman, P. Y. & Lugina, K. M. 1990: atmospheric hydrologic cycle over the Arctic Basin from Empirical data on contemporary global climate changes reanalyses. Part II: interannual variability. J. Clim. 14, (temperature and precipitation). J. Clim. 3, 662−677. 2414–2429. Volodin, E. M. & Galin, V. Y. 1998: Sensitivity of midlatitude Rogers, J. C. 1997: North Atlantic storm track variability and Northern Hemisphere winter circulation to ozone deple- its association to the North Atlantic Oscillation and climate tion in the lower stratosphere. Russ. Meteorol. Hydrol. 8, variability of . J. Clim. 10, 1635–1647. 23–32. Rogers, J. C. & McHugh, M. J. 2002: On the separability of Vörösmarty, C. J., Hinzman, L. D., Peterson, B. J., Brom- the North Atlantic Oscillation and Arctic Oscillation. Clim. wich, D. H., Hamilton, L. C., Morison, J., Romanovsky, V. Dyn. 19, 599–608. E., Sturm, M. & Webb, R. S. 2001: The hydrologic cycle Rogers, J. C. & Mosley-Thompson, E. 1995: Atlantic Arctic and its role in Arctic and global environmental change: cyclones and the mild Siberian winters of the 1980s. Geo- a rationale and strategy for synthesis study. Fairbanks, phys. Res. Lett. 22, 799–802. Alaska: Arctic Research Consortium of the U.S. Roulet, N. T., Ash R. & Quinton, W. 1993: Methane fl ux Wallace, W. J. & Solomon, S. 2002: Interpretation of recent from drained northern peatlands: effect of a persistent climate change. Science 296, 895– water table lowering on fl ux. Glob. Biogeochem. Cycles 899. 7, 749−769. Walsh, J. E., Chapman, W. L. & Shy, T. L. 1996: Recent Roulet N., Moore, T., Bubier, J. & Lafl eur, P. 1992: North- decrease of sea level pressure in the central Arctic. J. Clim. ern fens: methane fl ux and climatic change. Tellus 44B, 9, 480–486. 100–105. Whiting, G. J. & Chanton, J. P. 2001: Greenhouse carbon bal- Sen, P. K. 1968: Estimates of the regression coeffi cient based ance of wetlands: methane emission versus carbon seques- on Kendall’s Tau. J. Am. Stat. Assoc. 63, 1379–1389. tration. Tellus 53B, 521–528. Serreze, M. C., Carse, F., Barry, R. G. & Rogers, J. C. 1997: Woo, M., Heron, R., Marsh, P. & Steer, P. 1983: Comparison Icelandic low cyclone activity: climatological features, of weather station snowfall with winter snow accumulation linkages with the NAO, and relationships with recent in High Arctic basins. Atmosphere–Ocean 21, 312–325. changes in the Northern Hemisphere circulation. J. Clim. Yang, D. Q. & Ohata, T. 2001: A bias-corrected Siberian reg- 10, 453–464. ional precipitation climatology. J. Hydrometeorol. 2, 122– Serreze, M. C., Maslanik, J. A., Key, J. R. & Kokaly R. F. 139. 1995: Diagnosis of the record minimum in Arctic sea ice Zhou, L., Tucker, C. J., Kaufmann R. K., Slayback, D., Shab- area during 1990 and associated snow cover extremes. anov, N. V. & Myneni, R. B. 2001: Variations in northern Geophys. Res. Lett. 22, 2183–2186. vegetation activity inferred from satellite data of vegeta- Serreze, M. C., Walsh, J. E., Chapin III, F. S., Osterkamp, T., tion index during 1981 to 1999. J. Geophys. Res. 106(D17), Dyurgerov, M., Romanovsky, V., Oechel, W. C., Morison, 20 069–20 083. J., Zhang, T. & Barry, R. G. 2000: Observational evidence Zhulidov, A. V., Headley, J. V., Robarts, R. D., Nikano- of recent change in the northern high-latitude environment. rov, A. M. & Ischenko, A. A. 1997: Atlas of Russian wet- Clim. Change 46, 159–207. lands. Saskatoon: National Hydrology Institute, Environ- Shahgedanova, M. (ed.) 2003: The physical geography of ment Canada.

14 Recent temperature and precipitation increases in West Siberia