ARTICLE IN PRESS

Atmospheric Environment 39 (2005) 65506556 www.elsevier.com/locate/atmosenv

Temperature trends in twentieth century at Pune, India

Alaka Gadgil, Amit Dhorde
Department of Geography, University of Pune, Pune 411 007, India Received 7 December 2004; received in revised form 15 July 2005; accepted 15 July 2005

Abstract Climatic change is one of the most important issues of present times. Unlike the greenhouse gases, which have a predominantly warming effect, atmospheric aerosols could either warm or cool the atmosphere depending upon the size, distribution and optical properties. Of all the climatic elements, temperature plays a major role in detecting climatic change brought about by urbanization and industrialization. This paper, therefore, attempts to study temporal variation in temperature over Pune city, India, during the period 19012000. The long-term change in temperature has been evaluated by MannKendall rank statistics and linear trend. The analysis reveals signicant decrease in mean annual and mean maximum temperature. This decrease in temperature is more pronounced during the winter season, which can be ascribed to a signicant increase in the amount of suspended particulate matter (SPM) in the ambient air during the last decade. On the contrary, monsoon season shows warming. This warming can be attributed to a signicant increase in the low cloud amount. r 2005 Elsevier Ltd. All rights reserved.
Keywords: Urbanization; MannKendall rank statistics; Cooling trend; Aerosols; SPM

1. Introduction Assessing the impacts of urbanization and landuse change on mean surface temperature calculations is a challenging task. Several studies (Balling and Idso, 1989; Karl et al., 1988; Goodrich 1992) published in the last 15 years have attempted to assess the effects of urbanization on local and regional climate. A study by Jones et al. (1990) on urbanization and related temperature variation indicates that the impact of urbanization on the mean surface temperature would be no more than 0.05 1C per 100 years. This value appears to be too small when compared with the other studies (Fujibe, 1995; Hingane, 1996) using a similar technique. According to the study for Japan by Fujibe (1995) a rising trend of
Corresponding author.

E-mail address: gadgil@unipune.ernet.in (A. Gadgil).

25 1C per 100 years in minimum temperature has been observed at several large cities in Japan. While another study (Hingane, 1996) estimates rising trends of 0.84 and 1.39 1C per 100 years in the mean surface temperature calculated for Mumbai and Kolkata, respectively. Another study (Wibig and G"owicki, 2002) related to the variability of minimum and maximum temperature in Poland reveals that the strongest increase in minimum and maximum temperatures occurs in mid and late winter. However, beginning of winter and summer indicates decreasing tendencies. A similar study by Thapliyal and Kulshreshtha (1991) on temperature trends over Indian cities indicates a slight warming within the limits of 1 SD between 1901 and 1990. It is now recognized that urbanization and changing land-use inuence minimum temperature. Due to very high levels of energy consumption and friction in cities, signicant amount of waste heat is stored in the walls

1352-2310/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2005.07.032

ARTICLE IN PRESS
A. Gadgil, A. Dhorde / Atmospheric Environment 39 (2005) 65506556 6551

of buildings and streets which gets released during the night, thereby making the night warmer. Local temperature is thus one of the major climatic elements to record environmental changes brought about by industrialization and urbanization. In view of the importance of air temperature, as indicated above, it would be of interest to study the long-term variation of surface air temperature in Pune city, which, during the last two decades saw phenomenal rise in urbanization and industrialization. Therefore, the objective of the present work is to investigate the annual and seasonal temperature trends over Pune. It is also of interest to nd out whether the overall change in temperature is due to change in minimum or maximum temperature. This will help to know changes in night and day temperatures as well. Pune, belonging to Maharashtra state of India, is located in a basin surrounded by uplands and hills. It is situated on the western margin of the Deccan plateau, few miles away from the main range of the Western Ghats, at the conuence of rivers Mula and Mutha. The climate of the city is on the whole dry and invigorating. The temperature characteristics of Pune city are reported in Table 1, which indicates higher variability during winter season than during monsoon season. The mean monthly temperature is high (29.8 1C) in May while December witnesses a low (20.5 1C). However, the mean maximum temperature for April is (37.9 1C) more than that of May (37.2 1C). The high mean monthly temperature in May can be attributed to higher mean minimum temperature as compared to that of April. Similarly, the low mean temperature in

December is due to lower maximum temperature than that of January.

2. Data and methodology Monthly maximum and minimum temperature data during the period 19012000 along with cloud data (19582004), were obtained from India Meteorological Department (IMD), Pune. The data of suspended particulate matters (SPM) are taken from National Air Monitoring Program (NAMP), Pune, under the control of Central Pollution Control Board (CPCB), New Delhi. The authors are associated with the monitoring program since December 1994. From the basic temperature data, mean maximum (Tmax), mean minimum (Tmin) and mean temperature, along with their standard deviation (SD) and coefcient of variation have been computed for each month and three seasons, viz summer, monsoon and winter, that are depicted in Table 1. December, January and February are considered for the analysis of winter temperature as these 3 months record lower temperatures (Table 1). While computing the mean for winter season December of the previous year is included. March, April and May are months with highest mean maximum temperatures and, therefore, represent the summer season. June to September months constitute monsoon season. These data were then subjected to a 11-year running mean to nd the trends. A linear trend line was added to the series to simplify the trend. Temporal changes in the annual and seasonal values were also analysed by

Table 1 Monthly and seasonal temperature means Month January February March April May June July August September October November December Annual Winter Summer Monsoon Tmax (1C) 30.2 32.3 35.8 37.9 37.2 32.0 28.1 27.6 29.2 31.7 30.4 29.3 31.8 30.6 37.0 29.2 Tmin (1C) 11.6 12.7 16.3 20.1 22.3 22.8 22.0 21.3 20.6 18.9 14.8 11.8 17.9 12.0 19.6 21.6 Mean temperature 20.9 22.5 26.1 29.0 29.8 27.4 25.0 24.4 24.9 25.3 22.6 20.5 24.9 21.3 28.3 25.4 SD (1C) 0.9 1.0 1.0 0.9 0.6 0.8 0.6 0.5 0.5 0.8 1.2 1.1 0.4 0.8 0.6 0.4 CV (%) 4.3 4.4 3.8 3.1 2.0 2.9 2.4 2.0 2.0 3.2 5.3 5.3 1.5 3.7 2.1 1.7

Note: SD and CV are computed for mean temperature.

ARTICLE IN PRESS
6552 A. Gadgil, A. Dhorde / Atmospheric Environment 39 (2005) 65506556
27.00 26.00 Temp (C) 25.00 24.00 23.00 22.00 Annual Mean Temperature

MannKendall rank statistics (t) to conrm the signicance of the observed trend. The value of t can be used as the basis of a signicant test by comparing it with s 4N 10 , tt 0 tg 9N N 1 where tg is the desired probability point of the Gaussian normal distribution. In the present study, tg at 0.01 and 0.05 points has been taken for comparison. Apart from this, the linear trend tted to the data was also tested with t-test to verify results obtained by the Mann Kendall test. Using the cloud data trends were worked out. The NAMP data for SPM were also analysed to support and validate the possible evidences of signicant temperature trend.

1901

1911

1921

1931

1941

1951

1961

1971

1981 1971

Year 20 19 Temp (C) 18 17 16 Annual Mean Minimum Temperature

3. Annual temperature trends The mean annual, Tmax and Tmin along with 11-year moving mean and trend line are presented in Fig. 1. The mean annual temperature shows a signicant long-term decreasing trend. However, it can be seen from the gure that there had been a relatively warm period during 19401960. Similar features are also seen in annual Tmin and Tmax with slight differences in the relative dominance of warm and cool periods. A conspicuous cooling after mid-1950s can be seen in both Tmin and Tmax temperatures. However, it is observed that there is no signicant trend in Tmin, while Tmax shows cooling trend, signicant at 0.05 level.
15

1901

1911

1921

1931

1941

1951

1961

1981 1981

1991
Linear Trend

Year 34 33 Temp (C) 32 31 30 29 Annual Mean Maximum Temperature

Annual

11 yr mov mean

1901

1911

1921

1931

1941

1951

1961

1971

4. Seasonal temperature trends The mean temperature and the Tmax and Tmin for summer, winter and monsoon seasons during the period 19012000 are presented in Fig. 2. The gure also gives 11-year moving average of the temperature. 4.1. Winter The winter mean temperature shows a decreasing trend in spite of intermittent increases, which is statistically signicant at 0.01 level (Table 2). Tmin and Tmax also show cooling. However, this cooling trend of Tmin is signicant at 0.01 level. The Tmax during winter though shows decrease, is not statistically signicant. The 11-year running mean indicates that winter temperature is increasing up to mid-1950s and then decreasing. While minimum temperature depicts two epochs of warming around mid-1920s and mid-1950s, signicant decrease in winter mean temperature can be

Year

Fig. 1. Annual temperature trends at Pune.

attributed to predominant decline in minimum temperature (Fig. 2). 4.2. Summer The summer mean temperature also shows a decreasing trend, signicant at 0.05 level. However, unlike winter season this decrease is caused by signicant fall in maximum temperature. The MannKendall test indicates that Tmax decrease is signicant at 0.05 level, while Tmin shows no particular signicant trend. Therefore, it can be inferred that daytime temperatures in summer are signicantly decreasing. The 11-year running mean indicates that the temperature during summer was higher during 1940s to early 1960s after which it has come down.

1991

1991

ARTICLE IN PRESS
A. Gadgil, A. Dhorde / Atmospheric Environment 39 (2005) 65506556
30 29

6553
Monsoon mean temperature

Summer mean temperature

24 23

Winter mean temperature

28 27

Temp (C)

28 27 26 25 1901 1911 1921 1931 1941 1951 1961 1971 1981 1991

Temp (C)

21 20 19 18

Temp (C)
1902 1912 1922 1932 1942 1952 1962 1972 1982 1992

22

26 25 24 23 22

1901

1911

1921

1931

1941

1951

1961

1971

1981 1981 1981

Year
23 22

Year
15 14

Year
24 23

Summer minimum temperature

Winter minimum temperature

monsoon minimum temperature

Temp (C)

Temp (C)

Temp (C)

21 20 19 18 17 1901 1911 1921 1931 1941 1951 1961 1971 1981 1991

13 12 11 10 9 1902 1912 1922 1932 1942 1952 1962 1972 1982

22 21 20 19 18 1901 1911 1921 1931 1941 1951 1961 1971 1991 1991

Year
40 39

Year
33 32

1992

Year
32 31

Summer maximum temperature

Winter maximum temperature

monsoon maximum temperature

Temp (C)

Temp (C)

38 37 36 35 34 1901 1911 1921 1931 1941 1951 1961 1971 1981 1991

31 30 29 28 27 1902 1912 1922 1932 1942 1952 1962 1972 1982 1992

Temp (C)

30 29 28
Annual

27 26

11 yr mov mean

1901

1911

1921

1931

1941

1951

1961

Year

Year

Year

Fig. 2. Temperature trends for summer, winter and monsoon at Pune. Table 2 Result of MannKendall rank statistics Month January February March April May June July August September October November December Annual Winter Summer Monsoon Tmin 0.235** 0.208** 0.206** 0.068 0.129 0.266** 0.341** 0.365** 0.242** 0.162* 0.057 0.073 0.111 0.203** 0.092 0.354** Tmax 0.235** 0.044 0.064 0.159* 0.175* 0.043 0.116 0.009 0.021 0.112 0.115 0.129 0.149* 0.113 0.175* 0.013 Mean temperature 0.219** 0.172* 0.144* 0.127 0.070 0.026 0.195** 0.148* 0.136* 0.188** 0.061 0.106 0.166* 0.189** 0.157* 0.155*

to the increase in Tmin being signicant at 0.01 level. This indicates that the night temperatures during recent years have gone up during monsoon. t-Test when applied (Table 3) indicates that annual mean temperatures decrease at 0.05 level and of all seasonal temperatures, winter mean Tmin and monsoon Tmin trends are signicant at 0.01 level. Thus, all the trends are also well-supported statistically.

5. Monthly temperature trends Behaviour of minimum and maximum temperature has been studied for individual months by subjecting them to the MannKendall test. The results are presented in Table 2. It is interesting to note that the Tmin shows a signicant trend in a majority of the months. The beginning of winter, though it shows decreasing trend in minimum temperature, is not statistically signicant. The later part of winter (i.e. January and February) shows decreasing trend signicant at 0.01 level. The reverse trend of Tmin is seen during the monsoon months of June to September, signicant at 0.01 level. Tmax also shows cooling trend practically for all months except July and September.

*Signicant at 0.05 level, **signicant at 0.01 level.

4.3. Monsoon The monsoon season, on the contrary, depicts a signicant increase in the mean temperature. This is due

1971

1991

ARTICLE IN PRESS
6554 A. Gadgil, A. Dhorde / Atmospheric Environment 39 (2005) 65506556 Table 3 Linear equations and their signicance tested by t-test Linear equation Annual mean temperature Annual Tmin Annual Tmax Winter mean temperature Winter Tmin Winter Tmax Summer mean temperature Summer Tmin Summer Tmax Monsoon mean temperature Monsoon Tmin Monsoon Tmax *Signicant at 0.05 level,** signicant at 0.01 level. y 25.0160.0029x y 18.0660.0028x y 31.9670.0031x y 21.7490.0087x y 12.6360.0121x y 30.8620.0053x y 28.4890.0042x y 19.7490.0032x y 37.2290.0053x y 25.263+0.0036x y 21.3+0.0071x y 29.226+0.00009x Calculated t 2.2519* 1.7077 2.1313* 3.2683** 3.3229** 1.9681 2.0856* 1.4178 2.3138* 2.4676* 6.0006** 0.0442

However, this cooling is signicant for January at 0.01 level and April and May at 0.05 level.

6. Trends in cloud amount and SPM levels In order to investigate the possible causes for increasing/decreasing temperature trends during monsoon/winter seasons, data related to cloud amount and suspended particulate matter in the ambient air were analysed. The trend analysis of mean cloud amount shows decreasing tendency (Fig. 3), which is signicant at 0.01 level only during the monsoon season. The mean temperature during the monsoon season, however, shows increasing trend. This leads to a paradox, as increase in cloudiness is usually associated with increase in temperature. The monsoon season is evident with more low clouds and, therefore, the trend of low clouds was independently examined. Interestingly, it was observed that the proportion of low cloud amount has increased since 1958. Hence, the increasing trend of temperature during the monsoon season is justiable. Tmin during winter season, on the other hand, shows signicant decrease. Many authors have attributed the cooling trend to increasing atmospheric suspended particulate matter (Lamb, 1974; Barrie et al., 1976). These particles can both scatter and absorb sunlight. When a non-absorbing particle scatters solar radiation, some of the radiation is lost in space resulting in less sunlight reaching the Earth. This increases the net albedo of the Earthatmosphere system which would cause the net cooling. Pune city has witnessed phenomenal growth in urbanization and industrialization during the last three decades. This has resulted in an increase in the rate of aerosol production.

The authors of this paper are engaged in a project of ambient air monitoring of Pune city since 1994. As a part of this project the SPM levels at Pune were studied for two periods, December 1994March 1999 and June 1999December 2004. The average value of SPM of three sites is indicated in Fig. 4. Difference in the values of SPM during these periods is due to shift in the site of monitoring. In spite of this, levels of SPM over the city are increasing signicantly. From 1994 to 1999 the increase is signicant at 0.01 level and from 19992004 it is signicant at 0.05 level. It was also observed that SPM levels are higher in winter and summer and are lower during monsoon season, the latter because rains wash out SPM. Higher SPM levels during winter and summer may be responsible for scattering the incoming solar radiation, consequently decreasing its amount reaching ground surface. This may in turn lead to decreasing tendency in temperature during these seasons.

7. Conclusion An important aspect of the present study is the signicant cooling trend in mean annual temperature, which is more predominant during winter season. The summer season also shows signicant cooling trend due to decrease in Tmax. This cooling trend in Punes temperature is supported by studies conducted by other researchers (Rupa Kumar and Hingane, 1988). These authors studied the temperature for Pune during the period 18761986 and observed a cooling trend, but not signicant at any level. Against this background, in the present study, temperature data during the period 19012000 have been studied. The result indicates signicant decrease in winter temperature at 0.01 level. This suggests that the last decade has witnessed a phenomenal epoch in temperature series, leading to a

ARTICLE IN PRESS
A. Gadgil, A. Dhorde / Atmospheric Environment 39 (2005) 65506556
3 Mean Cloud Amount, Summer Cloud Amount (Oktas) y = -0.002x + 0.914 Cloud Amount (Oktas) R2 = 0.0144 2 3 Mean Cloud Amount, Monsoon y = -0.0059x + 2.3799 R2 = 0.3001

6555

0 1969 1979 1989 1999

0 1969 1979 1989 1999 y = 0.0098x + 4.1131 R2 = 0.123 1988 1998

Year 3 Mean Cloud Amount, Winter Cloud Amount (Oktas) Cloud Amount (Oktas) y = -0.0028x + 0.7455 R2 = 0.0291 6

Year Mean Low Cloud Amount, Monsoon

0 1970 1980 1990 2000

3 1958 1968 1978 Year

Year

Fig. 3. Cloud amount during summer, winter and monsoon at Pune.

350 300 SPM (ug / cu.m) 250 200 150 100 50 Oct--98 Sep--99 Aug--00 May--02 Dec--94 Nov--95 Dec--97 Mar--03 Jul--01 Jan--04 Nov--04 0

Month & Year

Fig. 4. Average levels of SPM at Pune since December 1994.

Lohmann, 1995). In the present study, it is observed that SPM levels at Pune are signicantly increasing since 1994. The observed cooling trend for the city, therefore, may be attributed to the upward trend of SPM in the ambient air, which the city has witnessed, due to phenomenal increase in urbanization during the last two decades. However, the relationship between aerosol concentration, distribution and anthropogenic pollution needs to be examined, before quantifying the impact of indirect forcing of anthropogenic aerosols. Otherwise, it is difcult to interpret the above results in terms of cause and effect. Regular measurements and long-period data are needed for aerosol optical characteristics. There is also a need to know more about the possible dimensions of natural climatic variability. We are still a long way from understanding the complex interaction of many physical processes that determine the evolution of climate.

decreasing trend from non-signicant to signicant. Contrary to this, the monsoon season shows warming trend. This may be due to signicant increase in the low cloud amount during this season. Recently, anthropogenic aerosols are recognized as providing a signicant and yet uncertain perturbation on the global radiation balance in terms of overall cooling effect (Charlson et al., 1987, 1992; Boucher and

Acknowledgements The authors wish to acknowledge and express their thanks to Central Pollution Control Board (CPCB), New Delhi for permitting them to use the SPM data in the present study. The authors are also thankful to

ARTICLE IN PRESS
6556 A. Gadgil, A. Dhorde / Atmospheric Environment 39 (2005) 65506556 temperature ranges. Papers in Meteorology And Geophysics 46, 3555. Goodrich, J.D., 1992. Urban bias inuences on long-term California air temperature trends. Atmospheric Environment 26B, 17. Hingane, L.S., 1996. Is a signature of socio-economic impact written on the climate? Climatic Change 32, 91102. Jones, P.D., Groisman, P.Ya., Coughlan, M., Plummer, N., Wang, W.C., Karl, T.R., 1990. Assessment of urbanization effects in time series of surface air temperature over land. Nature 347, 169172. Karl, T.R., Diaz, H.F., Kukla, G., 1988. Urbanization: its detection and effect in the United States climate record. Journal of Climate 1, 10991123. Lamb, H.H., 1974. The current trend of world climatea report on the early 1970s and a perspective. Research Publication No.3, CRURP 3 Climate Research Unit, University of East Anglia. Rupa Kumar, K., Hingane, L.S., 1988. Long term variations of surface air temperature at major industrial cities of India. Climatic Change 13, 287307. Thapliyal, V., Kulshreshtha, S.M., 1991. Climate changes and trends over India. Mausam 42, 333338. Wibig, J., G"owicki, B., 2002. Trends of minimum and maximum temperature in Poland. Climate Research 20, 123133.

Shri. G. Krishnakumar, Director, India Meteorological Department, Pune, and anonymous reviewers for their useful suggestions to improve the paper.

References
Balling Jr., R.C., Idso, S., 1989. Historical temperature trends in the United States and the effect of urban population. Journal of Geophysical Research 94, 33593363. Barrie, L.A., Whelpdale, D.M., Munn, R.E., 1976. Effects of anthropogenic emissions on climate: a review of selected topics. Ambio 5, 209212. Boucher, O., Lohmann, U., 1995. The suphateCCNcloud albedo effect: a sensitivity study with general circulation models. Tellus 47B, 281300. Charlson, R.J., Lovelock, J.E., Andreae, M.O., Warren, S.G., 1987. Oceanic phytoplankton, atmospheric sulphur, cloud albedo and climate. Nature 326, 655661. Charlson, R.J., Schwartz, S.E., Hales, J.M., Cess, R.D., Coakley, J.A., Hansen, J.E., Hoffmann, D.J., 1992. Climate forcing by anthropogenic aerosols. Science 255, 423430. Fujibe, F., 1995. Temperature rising trends at Japanese cities during the last one hundred years and their relationships with population, population increasing rates and daily