Spatio-temporal variation of daily extreme temperatures over Turkey

Authors

  • Hüseyin Toros

    Corresponding author
    1. Department of Meteorology, Istanbul Technical University, Maslak, Istanbul 34469, Turkey
    • Department of Meteorology, Istanbul Technical University, Maslak, Istanbul 34469, Turkey.
    Search for more papers by this author

Abstract

Extreme (maximum and minimum) temperature records are analysed for search of spatio-temporal variation patterns in regional climate change throughout Turkey based on 165 stations, as many stations as possible, and as far back as possible, in the period 1961–2008. The main purpose is to better understand the possibility of annual trends during the warm and cold periods. Homogeneity and Mann-Kendall trend test methodologies are applied in order to identify statistically significant trends at each station. It is observed that there is a significant warming in both annual maximum and minimum temperature records. Increasing trends in warm period are comparatively stronger than cold periods so far as the maximum temperatures are concerned. There are also increasing trends in annual minimum temperature during warm periods with almost no change in minimum temperatures of cold periods. The observations indicate that the temperature began to increase in the beginning of the 1980s. Nine stations are chosen from different geographical regions in order to compare monthly average temperature values of 1961–1984 with 1985–2008 and a marked increase was observed on extreme temperatures in these stations. Copyright © 2011 Royal Meteorological Society

1. Introduction

Energy and fresh water are the two major commodities that furnish the fundamentals of every human activity for a reasonable and sustainable quality of life. Solar energy is the most ancient source and the root for almost all fossil and renewable types (Şen, 2008). Climate phenomena are closely interactive with solar energy. Climate is one of the most vigorously debated topics all over the world due to climate change. Temperature is the most important issue of climate change and extreme (maximum and minimum) temperatures affect our lives the most. The greenhouse effect is unquestionably real and helps to regulate the temperature of the earth. Without a natural greenhouse effect, the temperature of the earth would be about − 18 °C instead of its present 14 °C. The Intergovernmental Panel on Climate Change (IPCC) concluded in 2007 that warming of the climate system was then ‘unequivocal’, based on observations of increases in global average air and ocean temperatures, widespread melting of snow and ice, and rising global average sea level (IPCC, 2007). Temperature fluctuation has both natural and anthropogenic origins, but the IPCC has concluded that most of the observed warming in global average surface temperature that has occurred since the mid-twentieth century is very likely a result of human activities (IPCC, 2007). During the first half of the last century, there was likely less human impact on the observed warming, and natural variations, such as changes in the amount of radiation received from the sun, which played a more significant role. Climatic variability from the past few million years to the present has been studied by many scientist owing to the fact that climate affects our lives, communities, health, welfare, agriculture, and economy. The natural and the artifical sources that effects the climate makes it difficult to understand. Climate oscillations and change are the results of many factors including the dynamic processes of the earth itself, human-induced effects, and sun activities. Some variations of the climate will be due to sun and earth interactions. The IPCC projects that, without further action to reduce greenhouse gas emissions, the global average surface temperature is likely to rise 1.8–4.0 °C during this century, and can increase up to 6.4 °C in the worst-case scenario. Even the lower end of this range would take the temperature increase since pre-industrial times above 2 °C, which is the threshold beyond which irreversible and possibly catastrophic changes become far more likely (IPCC, 2007). Nevertheless, this change has marked temporal and spatial, i.e. spatio-temporal differences. According to 4th Assessment Report of IPCC (2007), climate change is expected to result in significant climate variability in Turkey, which may lead to increasing temperatures and decreasing precipitation amounts.

Assessing the impacts of urbanisation and land use change on mean surface temperature is a challenging task. Influences of urbanisation on atmospheric properties are already investigated by many researchers. In these studies, they attempted to quantify the effect of increasing levels of urbanisation on the temperature record of major cities. These studies demonstrate the importance of isolating and removing the anomalous warmth produced by major cities. However, most of these studies suffer from the difficulty in quantifying urbanisation. Considerable amount of research work has been published in the last two decades (Karl et al., 1988; Karaca et al., 1995; Hughes and Balling, 1996; Englehart and Douglas, 2005; Şen, 2008). In order to understand the variations in climate parameters in the Mediterranean and Southeastern Europe, regional climate studies have been done by using meteorological mean annual, seasonal, cold, and warm temperature periods (Davi et al., 2002; Yihui and Zunya, 2008; Hachigonta et al., 2008; Silva et al., 2008). In Turkey, there are several studies on Turkey's climate and climate change. Some of them on climate change, climatic variability, and some on urbanisation (e.g. Toros, 1993; Karaca et al., 1995, 2000; Tayanç and Toros, 1997; Kadıoǧlu, 2000; Kömüşcü, 2001; Ezber et al., 2006; Türkeş et al., 1996; Freiwana and Kadıoǧlu, 2008; Tayanç et al., 2009). Tayanç et al. (1997) concluded that the southeastern regions of Turkey experience significant warming. The significant increases in the temperatures of the southern parts of the country is believed to be a result of desertification and the increasing frequency of the Africa and Middle East-originated heat waves for the last half century.

This study addresses climatic fluctuations in Turkey for the period 1961–2008. On the basis of data collected at 165 meteorological stations, yearly warm and cold period series of maximum and minimum temperature are analysed. Spatial and temporal analysis of these data gives more information about temperature variations in Turkey. The results are helpful for people with a growing interest in global and regional climate changes. Research on past and future climate is expected to contribute better conditions for our future. Every regional research provides its service when integrated with global researches. The climatic oscillations of Turkey are expected to provide hints about world climate change, because the location of Turkey at mid-latitudes plays a strong role due to its transition zone between Europe and Asia. The warm and cold periods are of great importance to human activities like heating buildings, agriculture, irrigation, water supply, etc. The results of this study give an indication of the present temperature state of Turkey on a regional basis. It considers warm and cold seasons as well as extreme temperatures. A general consensus on a recent global warming is presented in detail concerning Turkey. Finally, it helps to augment information on our understanding based on reliable datasets.

2. Study area, data and methodology

Figure 1 shows the location of the meteorological stations in the study area, Turkey (36°–42°N; 26°–45°E), which lies in southeastern Europe and Asia. The country has an area of approximately 780 000 km2, the world's 37th largest country, with Asia Major (97%) and a European (Thrace region) part, the territory of Turkey is more than 1600 km long and 800 km wide, similar to a rough rectangular shape. Turkey is encircled by seas on three sides: the Mediterranean Sea to the south, the Black Sea to the north and the Aegean Sea to the west. Turkey also contains the Sea of Marmara in the northwest. The coastal areas of Turkey bordering the Mediterranean Sea have a temperate Mediterranean climate, with hot dry summers, and mild wet and cold winters. Conditions can be much harsher in the more arid interior. The Taurus Mountains in the south close to the coast prevent Mediterranean influences from extending inland, giving the central Anatolian plateau of the interior of Turkey a continental climate with sharply contrasting seasons. Precipitation is more in the northeast and southwest regions, decreasing in the central area and occurring mostly during the winter months (Deniz et al., 2010).

Figure 1.

Spatial distribution of the meteorological stations

The monthly extreme temperature data from 165 meteorological stations for the period 1961–2008 have been studied in the present paper. The mean monthly extreme temperatures are derived from an average of the daily extreme temperature. Series of data are supplied by the Turkish State Meteorological Service (TSMS) of Turkey. The analysis is made for the annual average values during the warm (May–September) and cold (October–April) periods in addition to diurnal temperature range (DTR). The mean annual DTR is defined as the difference between the annual anomalies of extreme temperature. The warm periods are defined as the periods where there is generally no heating of buildings, and the monthly average temperatures are also greater than the yearly average. The cold period is generally when buildings need heating, and the monthly average temperature is lower than yearly average. In Figure 2 the distribution of extreme temperatures in Turkey is presented, where the minimum (maximum) temperature average is 1.5 °C (25.4 °C).

Figure 2.

Histogram distribution of monthly temperature in Turkey. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

Homogeneity of data used in the study is tested using a procedure involving the Wald–Wolfowitz (runs test) and graphical analyses. It is a non-parametric statistical test that checks a randomness hypothesis for a two-valued time series data sequence. The test can be used to test the hypothesis that the elements of the sequence are mutually independent (Wald and Wolfowitz, 1940). The stations are found to be homogeneous for air temperature data, the 48 longest and most reliable records are chosen, with data resolution for the time period from 1961 to 2008, and have little missing data (<5%), the missing data were replaced with the values of subsequent days by interpolation. The non-parametric Mann-Kendall (MK) statistical test is used to assess the significance of trend in both precipitation time series. MK tests are non-parametric for the detection of trends in a time series. These tests are widely used in environmental science, because they are simple, robust, and can cope with missing values and values below a detection limit. MK rank correlation statistics for each element xi the number of ni elements xj preceding it (i > j) is calculated so that rank(xi)> rank(xj). Hence, by definition the test statistic is,

equation image(1)

The distribution function of t is assumed to be asymptotically Gaussian with the mean (expectation) and variance as

equation image(2)

and

equation image(3)

respectively. In a two-sided test H0 is rejected for high values of |u(t)|, which is given as,

equation image(4)

Climate change can be detected by the Kendall coefficient t (Mann test) and when a time series shows a significant trend, the period from which the trend is demonstrated can be obtained effectively by this test (Sneyers, 1990; Toros, 1993). If MK statistic of a time series is higher (lower) than 1.96 (−1.96) then there is an increasing (decreasing) trend at 95% significance level.

3. Results and discussion

As mentioned earlier in this paper, in addition to 165 stations, 9 of them are selected to represent different geographical regions, and temperature records are divided into two periods. The first part represents an early period from 1961 to 1984, and the following one covers 1985–2008. Such a selection is based on the idea that intensive industrialisation has started in 1984 in Turkey.

There has been a clear tendency towards increase in extreme temperatures in global land areas. However, so far as DTR's are concerned more than 70% decreasing tendency is observed compared to the global land mass since the middle of the twentieth century. On the other hand, for the period 1979–2005, the DTR shows no significant trend since extreme temperature trends for the same period are virtually identical; both with strong warming signal (NOAA, 2009). A variety of factors are likely to contribute to such a change in DTR, particularly on a regional and local basis, including changes in cloud cover, atmospheric water vapour, land use, and urban effects. There has been a generally suitable anomaly in the extreme and DTR temperatures with some annual variations in the increasing rates of extreme temperatures, whereas, decreasing rates are observable in DTRs (Figure 3).

Figure 3.

Time series of the temperature anomalies of the annual mean extremes and DTR. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

3.1. Maximum temperature

Figure 4 represents statistically significant increasing (positive) and decreasing (negative) trends in addition to no trend existence. A total of 69% out of the stations show positive trends, with 66% of them significant, and scattered throughout Turkey. Another 24% of the stations indicate negative trends, whereas only 1% of them have significance.

Figure 4.

Statistically significant trends for annual average maximum temperature time series over Turkey. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

Trends for the warm season average maximum temperatures are shown in Figure 5, where 70% of stations show positive trends with 63% of them as significance at 10% level. Although there is no significant decreasing trend, 13% of the stations have negative trends.

Figure 5.

Statistically significant trends for warm period average maximum temperature time series over Turkey. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

As can be seen from Figure 6, there is a significant warming trend in the southern–southeastern parts, and significant cooling in northeastern parts of Turkey. A total of 28% of the stations yield positive trends, with 22% of them as significance. Additionally, 54% of the stations show negative trends, with only 2% of them with significance.

Figure 6.

Statistically significant trends for cold period average maximum temperature time series over Turkey. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

In Figure 7, there are annual, warm, and cold period maximum average temperature time series with linear trends, and their statistical significance, according to MK trend test u(t) and u′(t) time series. The trend values over a long period of annual warm and cold period average maximum temperature of Turkey are expressed graphically according to the MK trend tests. As in Figure 7, there is an increasing trend at significant confidence interval during the whole annual warm period temperature time series, but at the beginning of the time series there is a decreasing trend until 1976 followed by an increasing trend. Up to 1984, there are neither increasing nor decreasing trends in the warm period maximum temperatures, but later there appears a significant trend. Average maximum temperature of the cold season seems to show an increasing trend with some small oscillations, but not at a significant level. These results confirm that Turkey has no trend prior to 1980s, but increasing trend starts at the beginning of 1980s in maximum temperature records.

Figure 7.

Maximum temperature time series and MK statistics over Turkey

Figure 8 illustrates both mean annual maximum temperature spatial distribution and average annual time series of maximum temperature from 9 selected stations from different climatic geography. The spatial distribution of mean annual maximum temperature varies between 16 °C and 32 °C. As in the figure, the coastal areas are warmer than inland areas in Turkey. Mean monthly maximum temperature time series records at 9 stations are divided into two periods, namely, 1961–1984 and 1985–2008. Generally, there is an increasing trend in the average value of the second period.

Figure 8.

Selected station maximum temperature data for the periods 1961–1984 and 1985–2008, and spatial distribution of maximum temperature over Turkey. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

3.2. Minimum temperature

Figure 9 illustrates the spatial distribution of MK statistics of mean annual minimum temperatures over Turkey. A significant trend towards warming is evident in annual and warm period average minimum temperatures. It is higher in the warm period. Almost 52% of the stations indicate positive trends as against 43% of them with significance level. About 22% of the stations have negative trends out of which 6% of them have significance.

Figure 9.

Statistically significant trends for annual average minimum temperature time series over Turkey. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

General warm period temperature increasing trend is a common feature with characteristic fluctuations in Turkey. Figure 10 presents the trends in warm period average minimum temperatures. Again, the increasing trend dominates all over Turkey. It was found that 66% of the stations have positive trends while 62% of the stations have significance. On the other hand, 26% of the stations show negative temperature trends with only 4% of the stations having significant negative trends in northeast Turkey.

Figure 10.

Statistically significant trends for warm period average minimum temperature time series over Turkey. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

Figure 11 represents the spatial distribution of MK statistics of cold period average minimum temperatures over Turkey. Across the country, in the cold period minimum temperature series significant changes could not be observed. However, about 28% of the stations show positive trends, with 12% of them with significance. Almost 24% of the stations show negative trends, with 7% of them with significance. In 47% of the stations, there were no increasing or decreasing trends. Figure 11 shows that during the cold period there is no effective warming or cooling in minimum temperatures.

Figure 11.

Statistically significant trends for cold period average minimum temperature time series over Turkey. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

The least squares method and MK trend test results for time series of average annual minimum temperature time series are presented in Figure 12. There is neither increasing nor decreasing trend prior to 1990s, but there is a significant increasing trend afterwards in annual average minimum temperature, and there are fluctuations in warm period temperatures. There is warming until 1973, then cooling until 1984, and there is an increasing trend after this time. If one investigates all warm period time series in Figure 12, there is an increasing trend at significant confidence intervals within the minimum temperature time series. There is only some oscillation in cold period minimum temperature time series, which does not appear in warming and cooling temperature records.

Figure 12.

Minimum temperature time series and MK statistics over Turkey

The spatial distribution of mean annual minimum temperature is represented in Figure 13, according to which mean minimum temperatures in Turkey vary between − 12 °C and 12 °C. The coastal areas are warmer and farther inland areas are colder. Nine stations from different topographic regions illustrate minimum temperature monthly average values in 1961–1984 in comparison with 1985–2008. In general, there is an increase in the recent period average values, but there are decreasing trends at the cities Erzurum, Antalya and Zonguldak stations.

Figure 13.

Selected station minimum temperature data for the periods 1961–1984 and 1985–2008, and spatial distribution of minimum temperature over Turkey. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

4. Conclusions

This paper represents an updated revision of temperature trends in Turkey. It is possible to draw the following conclusions.

  • The result of this work is in agreement with previous studies, but differentiates from other investigations by using an updated dataset, which includes records up to 2008. Also, more stations are considered.

  • Turkey has warmed up during the last 25 years, and during the recent period, warming is more concentrated in maximum than in minimum temperatures. There has been a strong warming in the warm period compared to the temperatures of the annual and cold period.

  • More than 60% of the stations show significant positive trends in warm period extreme (maximum and minimum) temperatures.

  • There has been a clear tendency towards increasing in extreme temperatures for global land areas. There is a general suitability in the extreme and DTR temperature anomalies in Turkey.

  • Spatially, stations with significant increasing or decreasing trends in extreme temperature are almost uniformly distributed all over the country.

  • The differential rate of warming between the maximum and minimum temperatures is apparent in cold periods.

  • There is an increasing trend on the average extreme temperatures after 1985 in 9 selected stations.

  • This study shows that the increase in temperatures is not the only result of urban heat island effects.

Acknowledgements

I am grateful to the Turkish State Meteorological Office (TSMS) for providing the data; and thanks to Ozan Mert Göktürk (ITU) for preparing NCL code.

Ancillary

Advertisement