On the relationship between the premonsoonal rainfall climatology and urban land cover dynamics in Kolkata city, India



Empirical and modelling studies show that urbanization can have an impact on the environment. Relatively few studies have investigated urban effects on precipitation in India or other developing countries experiencing rapid urbanization. Furthermore, most precipitation-related studies for India focus on monsoonal rainfall. However, premonsoonal periods (March–May) account for 12–14% of the annual cumulative rainfall in eastern India. The majority of premonsoonal rainfall (PMR) is convective and caused by mesoscale forcing, which may include urban effects. In this study, the area under scrutiny is a large urban area in eastern India, Kolkata city. Herein, our goal was to (1) produce a comprehensive characterization of historical land cover dynamics associated with the Kolkata megalopolis, (2) provide a spatio-temporal climatology of PMR in the Kolkata region, and (3) identify possible associations between Kolkata's land cover and PMR. The analysis shows that the rate of change of urban land cover has increased by 50% compared to the period prior to India's independence in 1947. A multi-scalar time series analysis with Mann–Kendall statistics indicated statistically significant increasing trends in rainfall over the last 50 years for two Kolkata stations and a nearby downwind station. Furthermore, there was no significant trend for cumulative PMR in less urbanized stations, the country of India, or the East Gangetic region. This finding suggests that the anomaly of the three stations, showing increasing trends in PMR, could be the effect of urban land cover change. Copyright © 2011 Royal Meteorological Society

1. Introduction

By the middle of the 21st century, almost two thirds of the world's population will be living in towns and cities. Although currently only 1.2% of the Earth's land is considered urban, the spatial coverage and density of cities are expected to rapidly increase in the near future. It is estimated that by the year 2025, 60% of the world population will live in cities (UNFP, 1999). An estimated 90% of this increase will occur in developing countries (Maktav et al., 2005). By 2030, Asia and Africa will each have more urban dwellers than any other major area, with Asia (including China, Japan, and Korea) alone accounting for over half of the urban population of the world (Roth, 2007).

This study presents a robust and detailed assessment of spatio-temporal trends in urban land cover growth in Kolkata, India, and possible relationships with premonsoonal precipitation. The growth of the Kolkata Metropolitan Area over nine decades (Table I) is significant, although the maximum was in the last three decades. However, the real population explosion took place between 1940 and 1950 during the Indian independence phase and is still continuing (Chakraborty, 1990). Such rapid expansion provides a favourable period to conduct a spatio-temporal analysis of the premonsoon rainfall (PMR) climatology.

Table I. Change in urban land cover area over 300 years in Kolkata city
YearsTotal area (km2)Change in area (km2)Area change/year (km2)
Before 17564.0

Climatologically, the area under study (Figure 1) is located in the tropical Indian monsoon region in eastern India. The monsoon season rainfall initiates in early June and continues through September. There is very little rainfall during the winter season (Lohar and Pal, 1995). Monsoon season rainfall is of great concern to scientists in India because of its impact on the agricultural economy. The 3-month period (March, April, May) leading into the monsoon season is characterized by hot days with sporadic thunderstorms. The precipitation that falls during this period is called the PMR (Sadhukhan et al., 2000b). PMR contributes approximately 12% of the annual total rainfall (Sadhukhan et al., 2000a) in the study area. Figure 2 shows a climatological analysis of rainfall for the region and it clearly supports the notion that PMR is a contributor to the annual rainfall budget. During the premonsoon season, large-scale atmospheric forcing like the monsoon is minimal. As such, it is an appropriate period to investigate how urban land cover could affect precipitation in the study region. However, the predominance of monsoonal studies has substantially limited the number of climatological studies on the PMR, and there are essentially no studies from the perspective of urban effects.

Figure 1.

Map showing India, East Gangetic Plains, and rainfall stations (circles) used in the study. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

Figure 2.

Mean monthly rainfall for the area (82–89°E, 19–26°N) using Legates and Willmott (1990) dataset. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

In this study, the research objective is to analyse climatological precipitation data to assess current trends in premonsoon precipitation over the past 50–100 years, for urban and rural weather stations. Also, we seek to quantify the extent of change in urban land use/land cover change in the past 300 years and determine any link between urban land cover growth and cumulative PMR. It is important to note that we did not include aerosol effects (Shepherd, 2005; Rosenfeld et al., 2008; Lacke et al., 2009) on precipitation in the analysis. Historical aerosol data are fairly limited in the study region. In addition, the intent of the study was to isolate urban land cover effects, independently. Future analysis should certainly consider the synergistic land cover and aerosol effects; however, the approach herein is instructive.

2. Background

2.1. Urbanization of Kolkata

The city of Kolkata, a 300-year-old city, is located in the eastern part of India at 22°82′N, 88°20′E. It has spread linearly along the banks of the river Hooghly. Kolkata was the first major city developed by the British East India Company in the early 1700s. The city ultimately evolved into a provincial city and eventually became the headquarters of the British India government (Kundu and Nag, 1996). This established a level of importance for the city and stimulated significant growth in size over the years. The magnitude of its population, its volume of trade and commerce, the avenues of employment that it offers, the variety of its inhabitants speaking diverse languages that give it a cosmopolitan character, and its continued importance as a centre of art and culture attract people from everywhere (Roy, 1996).

2.2. Urban precipitation effects

Urban precipitation studies date back to Horton (1921), who noted that cities are associated with increased thunderstorm activity. Horton observed thunderstorms over some cities (e.g. Albany, NY and Providence, RI), which originated immediately over the city and affected surrounding areas. Early investigations (Landsberg, 1956; Atkinson, 1968; Changnon, 1968; Landsberg, 1970; Huff and Changnon, 1972) highlighted the possibility of urban effects on rainfall patterns. The Metropolitan Meteorological Experiment (METROMEX) was a landmark field experiment started in the 1970s to investigate the effect of St Louis on convective processes and precipitation. Examination of historical data at St Louis and other cities in the United States showed that summer rainfall was larger in quantity in the immediate downwind area. Results of METROMEX also suggested that areal extent and magnitude of urban and downwind precipitation anomalies were related to size of the urban area (Changnon et al., 1971).

In the past 30 years, numerous observational studies (Balling and Brazel, 1987; Bornstein and LeRoy, 1990; Jauregui and Romales, 1996; Dai et al., 1997; Selover, 1997; Bornstein and Lin, 2000; Changnon and Westcott, 2002; Shepherd et al., 2002; Changnon, 2003; Dixon and Mote, 2003; Inoue and Kimura, 2004; Burian and Shepherd, 2005; Diem and Mote, 2005; Shepherd, 2006; Mote et al., 2007; Hand and Shepherd, 2009; Lacke et al., 2009; Bentley et al., 2010) and modelling studies (Hjemfelt, 1982; Yoshikado, 1994; Thielen et al., 2000; Craig and Bornstein, 2002; Rozoff et al., 2003; Shepherd et al., 2007; Shem and Shepherd, 2009; Trusilova et al., 2008; Shepherd et al., 2010) have been conducted to understand how urban land cover affects precipitation, a key component of Earth's climate and water cycle system. Most of these studies continued to suggest that there was an upward trend in the rainfall amount in relation to the growth of the cities.

Coupled numerical modelling studies have investigated some physical mechanisms that may lead to urban hydroclimate effects. Studies are conducted using two-dimensional (2D) models Yoshikado (1994) in Japan, Thielen et al. (2000) in Paris, Baik et al. (2001)], as well as three-dimensional models [Ohashi and Kida (2002) in Japan, Craig and Bornstein (2002) in Atlanta]. These studies provide valuable insight but with the exception of Rozoff et al. (2003), the land surface representation was simple or non-existent (Shepherd, 2005). Even with such simple urban representations, the studies suggested that enhanced convergence, sensible heat flux, and atmospheric destabilization were factors in convective forcing under relatively weak large-scale forcing. For Atlanta (Shem and Shepherd, 2009), the Weather Research and Forecasting (WRF) model was employed to simulate both urban and non-urban situations for two case days and found 10–13% more rainfall to the east of the city for the urban scenario as compared to the non-urban scenario. This finding emphasized that Atlanta may have enhanced pre-existing storms through convergence and flux enhancement. Another recent study by Shepherd et al. (2010) used a coupled atmosphere–land surface model to investigate a typically sea breeze convection day near Houston. Simulations were run using two scenarios: ‘urban land cover’ and ‘no urban land cover’. The ‘urban’ simulation showed evidence of heavier rainfall over the city and just northwest of the city, which was confirmed by radar observations on that day. This study established how enhanced convergence and destabilization occurred over Houston even within a region experiencing sea breeze-forced convergence. In many ways, the urbanization and coastal geography associated with Kolkata are similar to Houston and thus, it is plausible that similar processes may be occurring.

2.3. Urbanization and premonsoonal precipitation in the East Gangetic region

India is an agriculture-based economy that is dependent on seasonal precipitation. The monsoons (June–September) play a very important role in this cycle (Figure 2). Previous and current investigations have studied the dynamics of the monsoons (Krishnamurti and Bhalme, 1976; Krishnamurti et al., 1981; Krishnamurti, 1985; Vines, 1986; Kulkarni et al., 1992; Kripalani et al., 1995; Goswami et al., 2006). Indian monsoon studies are not limited to observational approaches. Numerical prediction experiments and modelling approaches have also been conducted (Krishnamurti and Ramanathan, 1982; Bhaskar Rao et al., 2004; Dash et al., 2006). Although the monsoons are the most important source of rainfall in the Kolkata region, the focus of this study is the PMR (March–May).

The PMR is characterized by mesoscale phenomenon, i.e. thunderstorm activity. There are mainly two types of thunderstorms: (1) deep convection resulting from the dryline, which forms where the moist low-level southerlies converge with arid westerlies causing severe thunderstorms (locally known as ‘Kalbaishaki’; IMD Report 1941; Weston, 1972) and (2) thunderstorm convection initiated at the sea-breeze front (Lohar, 1996).

An analysis of the near-surface (1000 hPa) and mid-level (500 hPa) flow regimes during the premonsoon season was performed (Figure 3(a) and (b)). In this analysis, we composited the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCAR–NCEP) reanalysis of 1000-hPa (mb) and 500-hPa (mb) vector winds for the period 1950–2008. It is evident that the prevailing 500-hPa flow is zonal with a westerly component. The surface flow in the Gangetic plain is characterized by moist southerly flow from the Indian Ocean. This analysis corroborates the notion that the formation of Nor'westers or ‘Kalbaishaki’ is attributed to a warm, moist, southerly low-level flow from the Bay of Bengal and an upper-level dry, cool, westerly or north-westerly flow giving rise to an atmosphere with high latent instability (Mukhopadhyay et al., 2005).

Figure 3.

(a) Mean 1000-mb vector winds and (b) mean 500-mb vector winds in the premonsoonal season (1950–2008). This figure is available in colour online at wileyonlinelibrary.com/journal/joc

A number of studies on PMR regime were conducted to emphasize its importance. Sanderson and Ahmed (1979) conducted a study in neighbouring Bangladesh to understand the variability and significance of PMR. They concluded that PMR was less significant than monsoon rainfall, but provided 20–90 cm of rainfall annually. Lohar and Pal (1995) carried out a one-station study on a 20-year dataset of PMR. They concluded that irrigation reduced the low-level moisture and increased soil moisture. They hypothesized that these processes decreased the intensity of sea breeze circulation and related convection. Sadhukhan et al. (2000a) also conducted a study on the PMR (1901–1992) variability in the Gangetic West Bengal, which includes this study area. They found that there were no long-term trends present in the data series, unlike the monsoons, although there was an increase in PMR in 1970s and afterwards. Short-term fluctuations were present, which they linked to PMR being mainly dependent on local features. Sadhukhan et al. (2003) investigated the changes in land use pattern by using satellite data over Gangetic West Bengal extending from 20°N to 25°N and 85°E to 89°E and its possible impact on the local climate through numerical modelling. They found, similar to Lohar and Pal (1995), that changes in vegetation patterns have an influence on the moisture content of the inland-moving sea breeze and development of convective thunderstorms in coastal and neighbouring areas.

These studies theorize that land cover dynamics related to deforestation, agricultural use (Sadhukhan et al., 2000a, 2003), or irrigation (Lohar and Pal, 1995) could account for precipitation changes, yet the limited studies have been contradictory or lacked robustness. Furthermore, none of these studies considered the influence of urban-related land cover. This study, probably for the first time, will try to link urban land cover change to precipitation anomalies occurring in the area.

Urban rainfall studies for India have been limited. The first work of this kind was done by Khemani and Murty (1973), who analysed rainfall data from 1901 to 1969 for the city of Bombay and two other stations in the nearby rural region. Their study indicated that there was an increase of about 15% rainfall in the industrialized city of Bombay compared to nearby non-urban cities. The increase was observed during 1941–1969, which was a period of intensive industrialization. No data related to a possible heat island effect in the Bombay region were available to consider urban effects.

Simpson (2006) employed a mesoscale model and observed rainfall data to investigate the impact of Chennai (a large city in the southeast of India) urban land use on the sea breeze circulation and rainfall amounts during the southwest monsoon. Rainfall was more pronounced during late evening and nocturnal hours, possibly because of the interaction between the receding sea breeze circulation and the urban heat island.

Devi (2006) discussed the nature and intensity of heat islands at Visakhapatnam, the tropical coastal city of South India, for over 10 years. They also noted that the land and sea breeze circulations interact with the heat island as recently described by Shepherd et al. (2010). De and Rao (2004) conducted a study involving 14 major cities (over 1 million population) of India for the monsoonal period. They found that between 1901 and 2000, seven cities including Kolkata had increasing trends. The increasing trend in the monsoonal rainfall (an approximate rate of 80 mm per decade) is most evident over Kolkata for the period 1951–2000, which is characterized by an intense change in land cover. Finally, Kishtawal et al. (2010) have shown a positive correlation between heavy rain events and urban land cover growth for large urban areas in India. On the other hand, Sen Roy's (2008) observational study on extreme hourly precipitation trends (1982–2002) for large urban areas in India with populations greater than 1 million showed that Kolkata had a negative trend in its rainfall pattern in the dry PMR months (March–May).

3. Research objectives

The PMR period with sporadic thunderstorms offers an appropriate framework for this type of study because urban effects on rainfall are most evident when strong large- scale or synoptic forcing is weak and mesoscale processes are more dominant (Shepherd, 2005). Furthermore, very few studies have examined the effect of urban land use in developing nations, yet the world's fastest growing urban areas are precise in these regions.

Previous studies highlighted an anomaly in rainfall but did not identify the possibility of any causal effect of urbanization on precipitation. In some cases, the results were also conflicting. Such studies motivated this study of urbanization and its effects on PMR. Specifically, the research objectives are (1) to delineate urban land cover change using historic cartography and remote sensing data; (2) to analyse historical rainfall data and quantify the trend in the spatio-temporal PMR climatology of the area; and (3) to ascertain any link between trends in PMR over the last 50 years and urban land cover change.

4. Data and methodology

4.1. Data

Data from the National Data Centre–Indian Meteorological Department have been acquired for several stations in the study area. Most of the data cover the period 1900–2003; however, there are several stations with gaps. In this study, only five stations were analysed, demarcating a time period spanning 1950 to 2000. In addition, data from the Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis (MPA) (Huffman et al., 2007) and the Delaware surface dataset (Legates and Willmott, 1990; Legates, 1995; Willmott and Matsuura, 1995; Willmott and Robeson, 1995) were employed for spatio-temporal precipitation analysis in the region. Further, the NCAR–NCEP reanalysis data (Kalnay et al., 1996) archived by the NOAA Earth System Resource Laboratory (ESRL) were composited to provide the climatological flow patterns during the premonsoon period. Compositing was accomplished using the ESRL online compositing tool (http://www.cdc.noaa.gov/).

Data used to delineate the urban growth of Kolkata city include historical paper maps, Landsat images (1990 and 2000), and topographical maps. All maps used in this research, except Landsat images, have been collected from different Indian Government offices.

The first remotely sensed image was a Landsat TM image, dated 14 November 1990. It was a WRS 2 (Worldwide Reference System) image and the path/row is 138/044, respectively. The second remotely sensed image was a Landsat ETM (Enhanced Thematic Mapper) satellite image, dated 17 November 2000. It is a WRS 2 image and the path/row is 138/044, respectively. The Landsat images were used as they provide a mechanism to monitor land cover and land use globally by remote sensing from space (Lillesand and Kiefer, 1987; Morain, 1998).

A historical paper map of Kolkata city showing the growth of the city from pre-1756 to 1990 was also used. The scale of the map is 1:100 000. The source of the map is the National Atlas & Thematic Map Organization, India. Two topographical maps, # 79 B/6 & 79 B/7, showing Kolkata city and adjoining area were also used (1:50 000; Survey of India).

4.2. Methodology

4.2.1. Urban growth delineation

Digitization of paper maps was validated by remotely sensed images from recent years. The scanned paper maps were rectified with the base paper maps, mosaiced together, and georeferenced to have a larger view of the area. The extent of growth in the city was delineated using ‘heads-up’ digitizing methods from the urban growth map and LANDSAT images using the ArcGIS software. Although the process of digitization was laborious, it was necessary to delineate the extent of growth of Kolkata city over 300 years. The next step was to overlay the digitized maps onto satellite images to further digitize and detect the changes in the growth pattern between 1990 and 2000.

4.2.2. Trend analysis

Statistical trend analysis, primarily a Mann–Kendall trend test, was conducted on the IMD (Indian Meteorological Data) and Delaware data to determine whether statistically significant trends in PMR exist for urban and non-urban stations. The Mann–Kendall is a nonparametric test that helps determine trends in the time series (Mann, 1945; Kendall, 1955). The nonparametric trend tests require only that the data be independent and can tolerate outliers in the data, unlike parametric tests, which require data to be independent and normally distributed (Hamed and Rao, 1998). Hipel and McLeod (1994) documented that Mann–Kendall test was more powerful as compared to log-one series correlation tests when dealing with normally distributed data. Yue et al. (2002) and Yue and Pilon (2004) also comparatively tested the robustness of the Mann–Kendall method for trend detection and concluded that it was equally as effective as other trend detection methods like the Bootstrap tests and t tests. The Mann–Kendall test has been used to analyse precipitation trends in China and the United States (Karl and Knight, 1998; Gemmer et al., 2004).

The Mann–Kendall test is applicable to data where observations have been made annually at numerous locations, such as at water wells and rainfall data, and one overall test is desired to determine whether the same trend is evident across those locations (Helsel and Hirsch, 1992). The Mann–Kendall test computes Kendall's tau nonparametric correlation coefficient and its test of significance for any pair of X, Y data. This was directly analogous to regression, where the test for significance of the correlation coefficient r is also the significance test for a simple linear regression (Helsel and Hirsch, 1992). The Mann–Kendall test through tau's correlation (Table II) can be interpreted as a simple function of the probability that as X increases Y will increase too. This probability is rescaled to range from − 1 to 1 as customary for a correlation (Stephan et al., 1999).

Table II. Mann–Kendall analysis of rainfall data for the six study stations, East Gangetic Plains, and India
StationsP-valueTau correlation
  1. P-value > 0.05 (not significant); P-value < 0.05 (significant).

  2. (The P-values in bold show stations with increasing precipitation trends).

Alipore (within Kolkata)0.01670.227
Dumdum (within Kolkata)0.01290.236
Krishnanagar0.3982− 0.081
Sagar Island0.60320.051
East Gangetic Plains0.17950.128

Our analysis was performed using a 3-year running mean to eliminate yearly variations in the time series. Similar analyses were conducted on regional (India and East Gangetic Plains) PMR in order to compare individual station trends to regional trends. The six stations are Alipore and Dumdum within the Kolkata urban area and Bankura, Midnapore, Krishnanagar, and Sagar Island, which are classified as semi-urban areas (Figure 4). Kolkata city has the highest population (13 205 697) and the other semi-urban areas have populations ranging from 128 000 to 150 000 (Census of India, 2001).

Figure 4.

Map showing location of the six stations in the East Gangetic Plains. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

The India–East Gangetic urban analysis established the multi-scalar approach of the study from a regional to a local scale. We have used TRMM Multisatellite data (Huffman et al., 2007) and Delaware data (Legates and Willmott, 1990; Willmott and Matsuura, 1995) on a regional scale (India and East Gangetic Plains) and then narrowed it down to traditional station data on a local scale (six weather stations).

The stations, Alipore and Dumdum, are situated within the city of Kolkata. If there is convection at both Alipore and Dum Dum stations, then it is considered as a single convective event for Kolkata, regardless of the nature and intensity of convection at the two places (Dasgupta and De, 2007). The stations Bankura, Midnapore, and Krishnanagar are located inland, within 100–200 km of the city of Kolkata. The Sagar Island is a coastal station located at the edge of the Bay of Bengal, where the influence of the sea breeze is significant.

TRMM Multisatellite data (Huffman et al., 2007) were used to extend the Delaware rainfall data from 1999 to 2003 for the India and East Gangetic trend plots. A correlation analysis conducted for a 24-month period was performed in which the Delaware data and TRMM data overlapped. The correlation coefficient (r2) values were greater than 0.95. This suggests that the data extension technique was sound.

4.2.3. Spatial analysis

The spatial analysis utilized 10 years of TRMM Multisatellite analysis. We composited the premonsoon months for the period 1998–2007 and plotted the ‘mean’ PMR map for the country of India. The goal of this analysis was to provide a comprehensive spatial climatology of PMR over India.

5. Results

5.1. Land cover dynamics

The digitization and overlaying of the maps showed the phases of growth in different time frames beginning from before 1756 to 2000. The result was a unique sketch of the growth history of Kolkata over this time period, which is shown in the table (Table I). The table indicates that the city was growing consistently until 1947. After 1947, there has been a nearly fourfold increase in the change in area for the city of Kolkata from 225 to 1102 km2. The rate of change in urban land cover growth over the period 1947–1990 was 23.20 km2/year compared to 2.50 km2/year over the period 1856–1947. The table shows that between 1990 and 2000 the change in area for Kolkata city was 29.50 km2, which was more than what was observed before 1990 (23.20 km2). Such urban land cover dynamics establishes the basis for the second and third objectives, namely, to determine if there is any trend in the PMR over the same period and ascertain any link between land cover change and rainfall.

5.2. PMR trends

We analysed the time series of the PMR of India, the East Gangetic Plains, and individual stations. The Mann–Kendall analysis was performed on mean PMR for (1) the entire country of India, (2) the East Gangetic region, and (3) the six stations within the East Gangetic region. For the East Gangetic and India trends, the blended Delaware–TRMM dataset is employed.

The time series plot (Figure 5(a)) of the mean PMR for India (1951–2003), with Mann–Kendall analysis, shows no significant climatological trend in the distribution of PMR over the entire country. Similarly, the East Gangetic region (Figure 5(b)) does not show any statistically significant increasing or decreasing trend of PMR over the 50-year period (Table II). Thus, on a regional scale there is no statistically discernible large-scale trend over the study area that might mask or explain a more local or regional trend.

Figure 5.

(a) Time series of cumulative pre-monsoonal rainfall (3-year running mean) for India. (b) Time series of cumulative premonsoonal rainfall (3-year running mean) for East Gangetic Plains

In contrast, the analysis of individual stations showed three station time series with statistically significant increasing trends, two within the city Kolkata and one Midnapore, which is a fairly sizeable city, slightly west of the Kolkata Metropolitan region. In all three cases, the p-value is less than 0.05, which establishes that it is significant at the 95% level.

Trend lines for each station (Figure 6) with the Mann–Kendall statistical analysis show that the stations Alipore and Dum Dum show increasing trends. But the other stations, with the exception of Midnapore, are not showing any significant trend.

Figure 6.

Time series of cumulative pre-monsoonal rainfall (3-year running mean) for the six study stations (IMD data). This figure is available in colour online at wileyonlinelibrary.com/journal/joc

6. Discussion

Kolkata city grew immensely from 225 km2 in 1947 to 1297 km2 in 2000. It reflected an approximate tenfold increase in size in the last five decades. A major factor contributing to the expansion of the city was the influx of millions of refugees across the border around the time of India's independence in 1947. An estimated 4 million people moved from Bangladesh to West Bengal between 1946 and 1971 (Chatterjee, 1990) and in subsequent decades. This rate of expansion continued for decades after the independence.

Another factor contributing to the urbanization was the influx of people from the surrounding non-urban and less developed areas to the city (Roy, 2003). Kolkata has become the target city for people of east India to migrate to and a source for financial advancement for many people. Population growth and the drive for economic development have led to unplanned and sporadic expansion of the megacity.

The Mann–Kendall analysis appears to support the hypothesis that large urban areas like Kolkata may be associated with cumulative rainfall trends in the premonsoonal season. The Mann–Kendall statistical analysis shows that the stations Alipore and Dum Dum along with semi-urban station Midnapore show increasing trends. This suggests that urban land cover could be a driving factor for these increasing trends in Alipur and Dumdum stations. As mentioned earlier, a precipitation event in either the Alipore or Dum Dum station is considered as one precipitation event for Kolkata. In this study, the two stations, Alipore and Dum Dum, are considered separately to gather independent evidence and validate results from each station of urban effect on precipitation.

The Midnapore station, although not as urban as the other two, shows an increasing trend too. We hypothesize that Midnapore is in a downwind anomaly region of the Kolkata urban area and thus may experience some of the urban enhancement. Further study is required to examine the possible mechanisms but is beyond the scope of this article. The Midnapore result is consistent with previous findings, which often suggested an anomaly, not only over the city but also on the downwind fringe of the urban land cover extent (Shepherd et al., 2002; Diem and Brown, 2003; Mote et al., 2007; Hand and Shepherd, 2009; Shem and Shepherd, 2009). All these studies showed some evidence of an increase in the precipitation, which they hypothesized to be related to heavy urban development. More recently, Shepherd et al. (2010) reviewed studies for cities around the world and established that the ‘urban rainfall effect’ is now conclusive though mechanisms are not well understood.

The results of Mann–Kendall analysis support some of the findings that have been cited earlier in the article inferring that land cover change in the Gangetic West Bengal might influence precipitation (Lohar and Pal, 1995; Sadhukhan et al., 2000a, 2002), although none of the studies were related to urban land cover change. Urban rainfall studies conducted on Indian cities like Bombay (Khemani and Murty, 1973), Chennai (Simpson, 2006), Visakhapatnam (Devi, 2006), and Kolkata (De and Rao, 2004) all show increasing trends in rainfall, except a study done by Sen Roy (2008) where Kolkata showed decreasing trends in extreme hourly precipitation.

A spatial analysis of composite PMR (1998–2007) is interesting for multiple reasons. First, it is possibly the first time that a spatial presentation of PMR has been presented. Previous studies investigate individual station records or trends. TRMM MPA, a blended satellite product incorporating TRMM, infrared, and gauge-derived precipitation dataset, showed an anomalously high region of rainfall extending from the Kolkata region southwestward along the coast (Figure 7). This finding establishes that the TRMM MPA at approximately 25 km spatial resolution can identify mesoscale or topographically forced precipitation signatures, such as those associated with mountainous terrain (e.g. the Himalayas) or the sea-breeze front (along the coast). Of greater importance, the figure raises the question as to why the entire coastal plain does not display significant rainfall totals in response to sea breeze forcing, whereas regions around the Kolkata metropolis do. Shepherd and Burian (2003), Burian and Shepherd (2005), and Shepherd et al. (2010) have shown that coastal cities like Houston, Texas (USA), may amplify the precipitation signature through urban–mesoscale (Nor'wester, sea breeze) interactions. However, it is more likely that urban enhancement of the ‘Nor'wester’ storms could also explain this feature as we did not find very strong sea breeze rainfall signatures in our analysis of PMR (not shown). Our future modelling studies will hopefully shed light on this finding and what synergistic role the sea breeze, dryline convergence, and urban forcing may play in this distribution.

Figure 7.

TRMM multisatellite precipitation analysis data (1998–2008) illustrating the climatology of PMR over India (25 km spatial resolution). This figure is available in colour online at wileyonlinelibrary.com/journal/joc

7. Conclusions

This study is a unique synergy of urban land cover dynamics and premonsoonal climatology. Using a combination of historical cartographic maps and satellite data, land cover change of Kolkata was quantified over the past 300 years. Although the rainfall analysis spans a truncated period of roughly the last 50–60 years or so, the 300-year land cover analysis was critical for placing the explosive urban land cover dynamics in 1947 in proper historical context. There is no evidence from the inception of the city showing any anomalous period of growth before Indian independence. Furthermore, the study established the most current and robust assessment of PMR trends and climatology while suggesting a possible link between urban land cover change during the most recent 50 years of explosive growth. The analysis shows that the rate of change of urban land cover has increased tenfold from before India's independence in 1947. Before 1947, the growth rate was 2.50 km2/year and after 1947 up to 1990 it was 23.20 km2/year. Between 1990 and 2000, the growth rate was 29.50 km2.

Time series trend detection over the last 50 years of PMR data showed that there was an increasing trend in the PMR for the two urban stations and the Midnapore station (located in the possible downwind region) but not for the three less urbanized stations around the region. Furthermore, there was no significant trend for cumulative PMR over the entire country of India or the East Gangetic region, in which Kolkata is located. This finding suggests that the anomaly of the three stations, showing increasing trend in PMR, could be the effect of urban land cover change.

The findings of this study establishing a possible link between urban land cover change and rainfall pattern set the stage for further studies to determine how Kolkata will grow in the next 25 years using the CA Markov growth model (Pontius and Malanson, 2005; Baoying and Bai, 2008) and its influence on precipitation. Another interesting research focus of our group would be to provide novel modelling experiments to test this finding from a physical mechanism perspective, comparing an urban and no-urban case. Further future work should focus on the role of urban aerosols on climate system (Ramanathan et al., 2001; Lau et al., 2006). It is also important to understand what will be the outcome if the same methods are applied to different cities under varied climate regimes. The implications and effects of this kind of research will provide a better understanding of what to expect in our future and what could be the possible feedback of today's increased anthropogenic forcings.


The authors acknowledge support from NASA grant NNX07AF39G, Precipitation Measurement Missions Program, and the Southern High Resolution Modeling Consortium/USDA Forest Service contract AG-4568-C-08-0063. The entire research work was done at the Department of Geography, University of Georgia, Athens, GA, USA. We would also like to acknowledge Mr P. K. Haldar from the Regional Meteorological Center, Kolkata, India, for assistance with rainfall data collection.