LONG-TERM TRENDS IN LAKE ICE COVER IN FINLAND - River Ice

Proceedings oh the 18th IAHR InternationalSymposium on Ice (2006)LONG­TERM TRENDS IN LAKE ICE COVER IN FINLANDJohanna KorhonenFinnish Environment Institute (SYKE), Helsinki, FinlandABSTRACTThe longest ice break­up series in Finland started in spring 1693 (river Tornionjoki) and freezing andbreak­up records are available at least since the mid­19th century from three lakes ­ Kallavesi, Näsijärviand Oulujärvi. The number of time series starting in the late 19 th century amounts to at least twenty. Alarge number of ice observations were analysed in order to identify long­term changes in the ice regime.The analysis clearly showed that there is a statistically significant change towards earlier ice break­up inFinland, from the late 19th century to the present time. There is also a significant trend towards laterfreezing in the longest series and thus also towards a shorter ice cover duration. The series of maximumthickness of ice showed both decreasing and increasing trends. These trends were statisticallysignificant for approximately half of the observation sites. Decreasing maximum ice thickness trendswere found in the southern part of the country and increasing trends in central and northern regions.KEY WORDS: lakes; break­up; freezing; ice thickness; trendsINTRODUCTIONWinter ice cover is a natural phenomenon in lakes and rivers in countries of high latitudes. Formation ofice and its disappearance are among the oldest hydrological observations in Finland. The long datasetsof ice break­up and freezing are an interesting subject of analysis because they are good indicators ofclimate change and variations. Ice break­up datasets of many lakes go further back in time than airtemperature records from the same areas. Moreover, Livingstone (1997) asserted that break­up dates areeven better climatic indicators than air temperature itself.The very first studies of cryophenological records in Finland (Hällström, 1841a,b) were reported byKajander (1989). Later, Johansson (1932) presented a number of river ice records and Simojoki (1940)introduced data from many Finnish lakes. In addition, statistics of ice conditions were published in theHydrological yearbooks starting a few years after foundation of the Hydrographical Bureau in Finlandin 1908. During recent decades and years, e.g. Palecki and Barry (1986), Kuusisto (1987; 1993),Magnuson et al. (2000) and Yoo and D'Ororico (2002) have analysed the trends of some Finnishbreak­up and freezing records. Kuusisto (1994) and Kuusisto and Elo (2000) also presented long­termchanges of maximum ice thickness for a few lake sites.-71-

Proceedings oh the 18th IAHR InternationalSymposium on Ice (2006)The aim of this study was to investigate long­term changes in the ice regime of several Finnish lakes. Alarge number of other lake sites are introduced in this study compared to the earlier analysis describedabove. In this study, the freezing and break­up records of almost ninety sites, and thickness records ofover thirty sites were analysed in order to study average ice regimes in Finland and to identify long­termtrends of ice cover. Most of the studied sites are lakes; only one river site (Tornionjoki) is included. Allthe sites studied have records at least since the 1960s. The data was analysed until year 2002. This paperis a summary of a more extensive study concerning ice conditions in Finland (Korhonen, 2005).DATA AND METHODSThe longest continuous ice break­up series from the river Tornionjoki dates back to the late 17 th century(Johansson, 1932; Kajander, 1995). The longest continuous freezing and break­up data series of lakesare from lakes Kallavesi, Näsijärvi and Oulujärvi. Break­up records are available from lake Kallavesisince the year 1822 and freezing data since 1833, from lake Näsijärvi since the winter of 1836­1837 andfrom lake Oulujärvi since 1853­1854. There are also some other lakes from which records exist sincethe late 19 th century. A large number of sites have records starting from the 1910s. Continuous icethickness observations were also started in the 1910s at some lake sites in central and southern Finland,but most of the ice thickness measurements began in the 1960s.The date of permanent freezing of the whole area visible from the observation site was taken as theofficial freezing date in this study. The break­up date was chosen as the date when ice was no longervisible from the observation site. If there were several records for a same year, the earlier one was used.The duration of ice cover in days was defined as the difference between the break­up and the freezingdates. Ice thickness has been measured in the wintertime on the 10th, 20th and 30th of the month sincethe late 1970s. Before that, ice thickness was measured every 15 th and 30 th of the month. A manualmeasuring rod and drill are used to measure ice thickness and the average value from three differentdrillholes is defined as the official measurement. Since the late 1970s snow ice thickness has also beenmeasured in addition to the total thickness of the ice cover.Long­term changes were defined by linear regression, moving averages were also used. Trends weretested in order to determine whether changes were statistically significant. The 2­tailed t­test was used todetermine the level of significance. Significances of trends were categorized to levels of 5 % (p

Proceedings oh the 18th IAHR InternationalSymposium on Ice (2006)north. The average standard deviation of the freezing date is approximately two weeks, which is abouttwice as high as that of the break­up.The mean maximum ice thickness in southern Finland is below 50 cm, in central Finland 50 to 60 cm,in the southern part of northern Finland 60 to 70 cm and in Lapland mainly 70 to 80 cm, but even 90 cmin north­western Lapland. The date of the maximum thickness occurs in mid­March in the south and inApril in Lapland ­ in southern Lapland at the beginning of April but in north­western Lapland as late asat end of the month.The break­up of ice occurs in southern Finland typically at the end of April, in central Finland by themiddle of May, in the southern part of northern Finland after the middle of May, in southern and centralLapland by the end of May and in northern Lapland at the beginning of June. In north­western Laplandthe ice breaks up normally as late as after mid­June. Ice breaks up simultaneously within a certainregion almost regardless of the size of the lake. There are no major differences between the south andnorth of Finland concerning standard deviation of the break­up date.The mean duration of ice cover in the south is normally about 140 days, in central Finland mainly 150 to180 days, in the southern part of northern Finland 180 to 200 days and in Lapland 200 to 220 days. Theduration of the ice cover can be estimated rather well by the freezing date. When freezing occurs early,the ice cover period is very likely to be longer than the average value, and the ice thicker than the meanthickness. A thicker ice cover obviously needs more heat to melt than a thinner one. Thus in the years ofthick ice cover the break­up will occur later than in average years.Connection between ice conditions and air temperaturesGeographical factors have a considerable effect on the ice conditions. Latitude is quite important as itdefines the amount of incoming solar radiation, which affects the air temperatures and also the iceconditions. Elevation of a lake is also of importance because the air temperature decreases as a functionof elevation. The dates of both freezing and break­up of given year can be estimated rather well usingair temperatures.The correlation between Nov­Dec air temperatures and freezing dates and Apr­May air temperaturesand break­up dates were calculated for the longest observation series. The coefficients were calculatedfor the longest observation periods available and for the period 1961­2000. Also moving correlationcoefficients were used. The correlations for break­up dates were better than for freezing dates, becausethe depth of a lake affects the freezing date but not the break­up date. The squares of correlationcoefficients for the period of 1961­2000 varied mainly from 0.40 to 0.65. There were a rather largevariation between different periods which can be seen from the moving correlation coefficients. Forsome periods correlation is remarkably lower or higher than the average.Trends of freezing, break­up and duration of ice coverLess than half of the ice cover series studied had some statistically significant trends for the wholeobservation period before the year 2002. Most of the trends were found for the break­up date. Thebreak­up date has become earlier and the freezing date later at many observation sites during the lastfew decades. The trends were statistically significant mainly in those sites which have records at leastsince the late 19 th century. Trends of the series that started in the 20 th century were in most cases not-73-

Proceedings oh the 18th IAHR InternationalSymposium on Ice (2006)statistically significant but some trends were also found for these shorter series, especially in southernFinland. In the series that started in the late 19 th century, the ice break­up has moved generally 6 to 9days earlier over one hundred years.The freezing has been delayed since the late 19th century, in most cases by 5 to 8 days per hundredyears. The duration of the ice cover has also significantly decreased at sites with the longest records, byabout two weeks (12 to 16 days) per hundred years. Only some of the freezing and duration of ice covertrends were statistically significant. Some trends towards an earlier freezing date were also found for afew shorter series. In the late 1920s and early 1930s, freezing occurred extremely late. On the otherhand, in autumn 2002 freezing occurred in small lakes in southern Finland extremely early. This resultsin a statistically significant trend towards earlier freezing in some small lakes in southern Finland, forexample in lake Kuivajärvi (Fig. 1). The extremely late freezing in the winter 1929­1930 and very earlyfreezing in 2002 make trend statistically significant.The trends of break­up were stronger than those of the freezing date. The wide variation of freezingdates hides the trends of freezing, especially in lakes with only short periods of records. The trends of icecover period were also more statistically significant than those of the freezing date. There were nostatistically significant trends of break­up, freezing or duration of ice cover in northern Lapland, but thetime series from this part of the country are still relatively short. Generally, trends were mainly strongerin the southern part of the country. The magnitudes of trends in this study were rather similar to thosepresented by Magnuson et al. (2000).When examining the subsets of long time series, some short­term trends can be found. The freezingseries of lakes Näsijärvi, Kallavesi and Oulujärvi show that no dramatic changes occurred from themid­1800s to the late19th century. From 1885 a number of other series can also be examined. At almostall of the sites freezing was delayed from the late 19 th century to the 1930s, because of the very latefreezing dates in some winters in the 1920s and the 1930s. After that, from the 1930s until the late 1960sfreezing has moved earlier in most places. From the 1960s to the present, freezing has again beenslightly delayed.The break­up series of the river Tornionjoki has many subset trends. From the late 17 th century until the1740s no remarkable changes can be found, but from the 1740s to the early 1800s break­up variedconsiderably. Break­up of lake Kallavesi was clearly delayed during the period from 1822 to the 1870s,and similar trends can be seen in the time series from lake Näsijärvi and the river Tornionjoki. For manyseries the late 19 th century was a turning point, after which the break­up started to move earlier.Simojoki (1940) determined the turning point for lake Kallavesi to be 1886. On the other hand, Eklund(1999) presented that a change to earlier break­up occurred in the early 19 th century in lake Mälaren,Sweden. From the late 1880s, the break­up dates have been moving earlier rather consistently, althoughthere is some variation between the years. A similar trend can be seen in the spring (March­April­May)air temperatures in Finland (Tuomenvirta, 2004). In the 1960s and the 1970s variation between theyears was rather small and extreme break­up dates did not occur.The duration of ice cover has decreased from the late 19 th century to the 1920s. Of course, there isconsiderable variation from year to year. From the 1930s to the early 1950s ice cover periods weregenerally shorter than in the 1920s or late 1950s, especially in southern and central Finland. From thelate 1950s to the present time ice cover period have decreased slowly. The decreasing trend has beenstronger in the south. This examination of subsets of also shows clearly that by choosing differentperiods in the time series, different trends can be found.-74-

Proceedings oh the 18th IAHR InternationalSymposium on Ice (2006)25.12.5.12.Oulujärvi Vaala (N) 1854­2002y = 0.0432x + 31928.6.18.6.8.6.Oulujärvi Vaala (N) 1854­2002y = ­0.067x + 150.5315.11.29.5.26.10.19.5.9.5.6.10.1854 1879 1904 1929 1954 197929.4.1854 1879 1904 1929 1954 1979Freezing date3.2.14.1.25.12.5.12.15.11.Kallavesi Kuopio (C) 1833­2002y = 0.0667x + 329.66Break­up date18.6.8.6.29.5.19.5.9.5.29.4.Kallavesi Kuopio (C) 1822­2002y = ­0.0599x + 142.7726.10.1833 1858 1883 1908 1933 1958 198319.4.1822 1847 1872 1897 1922 1947 1972 19973.2.14.1.Näsijärvi Tampere (S) 1836­2002y = 0.0594x + 342.6618.6.8.6.Näsijärvi Tampere (S) 1836­2002y = ­0.0905x + 138.0725.12.29.5.5.12.15.11.19.5.9.5.29.4.26.10.19.4.6.10.1836 1861 1886 1911 1936 1961 19869.4.1836 1861 1886 1911 1936 1961 19863.2.14.1.Kuivajärvi Saari (S) 1929­2002y = ­0.1865x + 334.8818.6.8.6.Tornionjoki Tornio (N) 1693­2002y = ­0.0431x + 142.2425.12.29.5.5.12.19.5.15.11.9.5.26.10.29.4.6.10.1929 1939 1949 1959 1969 1979 1989 199919.4.1693 1743 1793 1843 1893 1943 1993Figure 1. Time series of break­up and freezing at different sites. The letters S, C and N stand forsouthern, central and northern Finland.-75-

Proceedings oh the 18th IAHR InternationalSymposium on Ice (2006)The fact that trends of break­up dates were stronger than those of freezing dates also agrees with theobserved long­term changes in air temperatures in Finland (Tuomenvirta, 2004), northern Sweden(Klingbjer and Moberg, 2003) and north­western Russia (Filatov et al., 2005). These studies haveshown that air temperatures have increased more in spring than in autumn or in winter. The study ofFilatov et al. (2005) showed that the trend of spring temperatures was stronger in the southern part ofnorth­western Russia than in the northern part. This result dovetails with the finding of strongerbreak­up trends in the southern part of Finland in this work.10080Inari Nellim (N) 1950­2002y = 0.2892x + 65.266120100Kilpisjärvi (N) 1965­20026040806040202001950 1960 1970 1980 1990 200001965 1970 1975 1980 1985 1990 1995 2000Maximum ice thickness (cm)Muurasjärvi (C) 1910­1999100y = 0.1072x + 51.2328060402001910 1920 1930 1940 1950 1960 1970 1980 1990Iisvesi (C) 1912­20021008060402001912 1922 1932 1942 1952 1962 1972 1982 1992 200210080Päijänne Tehi (S) 1961­2002y = ­0.297x + 57.36110080Sääksjärvi (S) 1912­199960604040202001961 1966 1971 1976 1981 1986 1991 1996 200101912 1922 1932 1942 1952 1962 1972 1982 1992Figure 2. Time series of maximum ice thickness at different sites. The letters S, C and N stand forsouthern, central and northern Finland.-76-

Proceedings oh the 18th IAHR InternationalSymposium on Ice (2006)Trends of maximum ice thicknessBoth decreasing and increasing trends were found in the maximum ice thickness time series. At abouthalf of the sites the trends were statistically significant. The maximum thicknesses have generallyincreased in central and northern Finland and decreased in the south. The series with records since the1960s had significant trends of 2 to 3 cm per ten years, whereas for longer series the trend was mainly 1to 2 cm per ten years (Fig. 2).The time series of snow ice and snow thickness are rather short, about twenty years, and therefore nolong­term analysis can be made. In the 1980s there was much snow and snow ice in southern andcentral Finland. Thus the snow ice thickness has in most cases decreased from the 1980s to the year2000 in southern and central Finland. In northern Finland, the snow ice thickness has increased in someplaces during the twenty year period. Trends of maximum ice thickness are rather similar to the trendsof snow water equivalent (Hyvärinen, 2004). Snow water equivalent has increased in northern andeastern Finland and decreased in southern and western regions. Accordingly, snow ice thicknessprobably explains the reason for maximum ice thickness changes. The effects of climate change on icethickness are rather complex, because the ice thickness depends both on temperature conditions and onsnowfall. Even very similar frost sums and snowfalls in different winters can result in different icethicknesses, as was discussed earlier by Kuusisto (1994) and Kuusisto and Elo (2000).CONCLUSIONSThis study consisted of investigating the average ice regime in Finnish lakes and looking fortrends of ice cover time series. Almost ninety break­up and freezing series and over thirty icethickness series were analysed. The freezing of lakes occurs usually from late October to earlyDecember when progressing from Lapland to southern Finland. The average year­to­yearvariation of freezing date is about two weeks. The variation of the break­up date is about half ofthe variation of the freezing date. Lake ice break­up occurs generally in late April in southernFinland and in early June in the north. The mean duration of ice cover in the south is about 140days, in central Finland mainly 150 to 180 days and in northern Finland 180 to 220 days. Theaverage maximum ice thickness from the south to the north is from 50 cm to 80 cm.The analysis showed a statistically significant change towards earlier ice break­up in Finland fromthe late 19th century to the present time. The break­up has moved mainly 6 to 9 days earlier over aperiod of hundred years. The greatest change towards earlier break­ups occurred in the late 19 thcentury. There was also a significant trend towards later freezing in the longest series and thus alsotowards a shorter ice cover duration. The delay of the freezing date was about 5 to 8 days in onehundred years for the longest time series. The duration of ice cover has decreased by about twoweeks (12 to 16 days) per one hundred years. For those series which started in the 20th century,the trends were mainly not significant. However, some of the shorter break­up series did havestatistically significant trends. The fact that trends of break­up were stronger than the trends offreezing agrees with observed long­term changes in air temperatures in Finland, Sweden andnorth­west Russia.The series of maximum thickness of ice showed both decreasing and increasing trends. The trendswere statistically significant for approximately half of the observation sites. Decreasing trendswere found in the southern parts of the country, increasing trends in the central and northern parts.-77-

Proceedings oh the 18th IAHR InternationalSymposium on Ice (2006)REFERENCESEklund, A. (1999), “Islägging och is lossning på sjöar”, Sveriges meteorologiska och hydrologiskainstitut, Nörrköping, SMHI Hydrologi Nr 81.Filatov, N., Salo, Y. and Nazarova, L. (2005), “Effect of climate variability on natural waterbodies in Northwest Russia”, Proceedings of the 15th Northern Research Basins InternationalSymposium and Workshop, Luleå to Kvikkjokk, Sweden, pp. 31­40.Hyvärinen, V. (2003), “Trends and Characteristics of Hydrological Time Series in Finland”,Nordic Hydrology 34 (1/2), 71­90.Hällström, G.G. (1841a), “Specimina mutanti currente seculo temporis, quo glacies fluminunannuae dissolutae sunt”, Acta Soc. Sci. Fenn. 1, 129–149.Hällström, G.G. (1841b), “De tempore regelationis et congelationis aquarum fluminis Kyro”, ActaSoc. Sci. Fenn. 1, 387–394.Johansson, O. V. (1932), “Isförhållandena vid Uleåborg och Torne älv”, Bidrag till kännedom afFinlands natur och folk, Utgifna af Finska Vetenskap­Societen H. 84 Nro 3, Helsingfors.Kajander, J. (1989), “150 years since G.G. Hällström’s studies on ice break­up dates as climaticindicators”, Publ. Acad. Finland 9, 329–338.Kajander, J. (1995), “Cryophenological records from Tornio”, Mimeograph Series of the nationalBoard of Waters and the Environment 552, Helsinki.Klingbjer, P. and Moberg, A. (2003), “A composite monthly temperature record from Tornedalenin Northern Sweden, 1802­2002”, International Journal of Climatology 23(12), 1465­1494.Korhonen,. J. (2005), “Suomen vesistöjen jääolot”, Finnish Environment 751, FinnishEnvironment Institute, Helsinki.Kuusisto, E. (1987), “An analysis of the longest ice observation series made on Finnish lakes”,Aqua Fennica 17 (2), 123­132.Kuusisto, E. (1993), “Lake ice observations in Finland in the 19th and 20th century: any messagefor the 21st?”, Snow watch '92: detection strategies for snow and ice, World Data Center A forGlaciology, Boulder. Report GD­25, pp. 57­65Kuusisto, E. (1994), “The thickness and volume of lake ice in Finland in 1961­90”, Publications ofthe Water and Environment Research Institute 17, Helsinki, pp. 27­36.Kuusisto, E. and Elo, A­R. (2000), “Lake and river ice variables as climate indicators in NorthernEurope”, Verh. Internat. Verein. Limnol. 27(5), 2761­2764.Livingstone, D. M. (1997), “Break­up dates of alpine lakes as proxy data for local and regionalmean surface air temperatures”, Climatic Change 37(2), 407­439.Magnuson, J. J., Robertson, D. M., Benson, B. J., Wynne, R. H., Livingstone, D. M., Arai, T.,Assel, R. A., Barry, R. G., Card, V., Kuusisto, E., Granin, N., Prowse, T. D., Stewart, K. M.and Vuglinski, V. S. (2000), “Historical trends in lake and river ice cover in the NorthernHemisphere”, Science 289(5485), 1743­1746.Palecki, M. A. and Barry, R. G. (1986), “Freeze­up and break­up of lakes as an index oftemperature changes during the transition seasons: a case study in Finland”, Journal of Climateand Applied Meteorology 25(7), 893­902.Tuomenvirta, H. (2004), “Reliable estimation of climatic variations in Finland”, FinnishMeteorological Institute Contributions 43, Finnish Meteorological Institute, Helsinki.Yoo, J. and D'Odorico, P. (2002), “Trends and fluctuations in the dates of ice break­up of lakes inNorthern Europe: the effect of the North Atlantic Oscillation”, Journal of Hydrology 268(1),100­112.-78-