Effects of South West Monsoon (SWM) on Cloud Cover and Shape


There's a specialist from your university waiting to help you with that essay.
Tell us what you need to have done now!

order now

An investigation of the cloud shape stability and the cloud covering is presented on the immediate pre and post south-west monsoon (SWM) commencement. The study focuses on Sri Lankan region bounded by longitude 700E-950E & latitude 00-200N during the months April, May & June in the years 2012 and 2013. Monsoon onset is a remarkably abrupt event and has established the criterion to determine the onset date which is noticeably controversial and a complicated issue. The prime aim of this paper is to understand the vagaries of the cloud dynamics on the before, inception & after the SW monsoon and the analysis are carried out purely based on the computer based image processing rather than chemical & physical processes.

Cloud shape stability which is determined by the pixel orientation based on the neighbourhood begins to drop in April-May and remains in that low level in subsequent months of both years while cloud cover which is determined by the brightness of the pixel starts to go up in the same period and remains in that high level in the following months as well.

The south-west (SW) monsoon brings somewhat organized heavy rainfall to Sri Lanka. The economy of the country is largely influenced by the SW monsoon because major industries such as agriculture and power generation are highly depended on rainfall. Therefore, studying the commencement of SW monsoon and its structure is significant.

There is no universally accepted definition or criterion to determine the monsoon onset and previous studies exhibited considerable year to year variation of SW monsoon onset [1]. Further, the climate system is a highly uncertain dynamical system which involves large number of factors such as precipitation, temperature, outgoing long wave radiation, wind speed and direction, humidity etc. Monsoon is also an onset-active-break-revival life cycle phenomenon [11]. Therefore, the establishment of a criterion to determine the onset is a noticeably controversial and complicated task. Major monsoon area of the world exists in South and East Asia and the Indian Ocean monsoon interact with the South China Sea monsoon [10]. Most of the previous studies relating to the monsoon onset were based on wind, precipitation, outgoing long-wave radiation (OLR) and brightness temperature [1, 9, 11]. According to the summarization done by Wang et al from 1992 to 2001, only in one among seventeen (17) studies was involved with cloud data [1].

However, Wonsick et al document the seasonal progression of the Asian monsoon by analysing clouds and convection in the pre-peak and post-monsoon season [6]. Moreover, maximum cloud zone and the ITCZ (Inter Tropical Convergence Zone) over the Indian region during the SW monsoon was investigated by Sikka et al and it has been found that the two maximum cloud zones are present during June-September, one of 150N and the other of equatorial region [4]. Most of the monsoon studies based on clouds as well as other variables were highly pertinent to China and Indian region. To the best of our knowledge, there have been no or very little research have been done to investigate the monsoon commencement in Sri Lankan region using the cloud patterns. In addition to the monsoon studies, there are certain studies relating to clouds. Among those, identification of tracer clouds done by Nilanjan et al [7] and automated cloud classification done by Bryan et al [3] motivated us to study the seasonal cloud dynamics.

Thirty minute interval cloud images were downloaded from the Images are taken from the Indian National Satellite System (INSAT) KALPANA 1 geostationary satellite. These images are derived from the emission by the earth and its atmosphere at thermal infrared wavelength of 10.5-12.5 µm and they cover the latitude range 100S-500N and the longitude range 450E – 1050E. The ground resolution at the sub satellite point is nominally 8 Km x 8 Km. Images are stored in the above web site as colour RGB JPEG format. Pixel resolution of the images is about 1200 x 1024 (72 dpi x 72 dpi) with 24bits depth including image header.

The region bounded by the longitude 700E-950E and the latitude 00-200N extracted from the original image and the grey converted image is used for this study. This extracted bounded region provides a good coverage for the island of Sri Lanka. Moreover, this study investigates the cloud images from January 2012 to July 2013.

Our objective of this research is to distinguish the strange behaviour of the clouds during the potential SW monsoon onset period. Sequel of this paper is as follows: next section describes the methodology followed by the results and discussion. The last section is allocated for a conclusions and recommendations.


Local orientation which is used as a feature of a satellite image is a major contributor to determine the cloud shape life time. Local orientation which is also called the linear symmetry (LS) is characteristics by the least change of grey value in one direction and maximal change in the orthogonal direction [2,5]. Therefore, a linear symmetry tensor for an image is constructed with respect to the local neighbourhood for each pixel of the image.

In this way, local symmetry tensors of the concerned bounded area of the 30 minute interval satellite images were constructed. By preserving the first image as the reference image, comparison takes place with its LS tensor and that of the subsequent images till the correlation drops below a specified threshold. In this study the threshold is 0.9. As long as this correlation of the subsequent image is higher than the threshold, is considered as the same cloud shape with the reference image. The time period until the drop of the correlation of the LS tensors below the threshold is considered as the life time of that particular cloud shape. This process was continued by taking the next immediate image as the new reference image and mean value of the life times is taken as the cloud shape life time for the particular day.

In addition, the cloud cover was determined in the said area of the image using the feature of brightness of the pixel. If the pixel brightness is higher than the global threshold which is determined by the otsu’s method, it is considered as the cloudy pixel. As such the cloud cover area was calculated for the chosen area of image by counting the number of cloudy pixels. Similarly, the average was taken as the cloud cover of the particular day.

Finally, the cloud shape life time and the cloud cover were analysed in the months of April, May and June which include the SW monsoon potential onset period [8, 10], of the years 2012 & 2013 which is presented in the following section.

All calculations and analysis were done using the software MATLAB and Microsoft Excel.


To look at the vagaries of the cloud dynamics in the immediate pre and post south-west monsoon commencement, monthly variation of cloud shape life time and cover from January 2012 to July 2013 are calculated and presented in figure 1 with the standard deviation. In both years, cloud shape stability life time start to drop in April-May and after the drop, this low level sustains during the entire peak monsoon period (June-September) in year 2012. Similar trend can be seen up to July in the year 2013. On the other hand, cloud cover starts to increase in same period and remains in that higher value during the whole peak period of monsoon in year 2012. Once again that trend can also be observed up to July 2013.

Figure 1 Cloud shape stability (right) and cloud cover percentage (left) including standard deviation, 2012 (top) and 2013 (below)

According to the previous studies, Wang et al (2004), Wonsick et al (2009) and Tsing-Change Chen et al monsoon build up on average from 2nd week of May to 1st week of June[1, 6, 8, 10]. When we closely look at how a cloud behaves in this previously identified monsoon onset period, interestingly, we can see a remarkable change of behaviour in both cloud shape life time and the cloud cover occurred in the data considered for this study. First of all, highest cloud cover is increased by 48.56% than the previous day on May 19, 2012. Similarly in 2013, this happens on May 25 which is a 48.47% increase than the previous day. After goes up that high level of cloud cover is averagely 40 per cent or higher in 2013 while it is 40 per cent or higher in most of the days in 2012 remains till mid-June. This is shown in the figure 2.

While cloud cover is increased on May 19 2012, the cloud shape stability time is strangely dropped by 51% on the same day when compared with that in the previous day. As shown in figure 3, after this drop the average low life time sustains till mid-June. For example, cloud shape life time of May 18, 19 and 20 are 12.42, 6.08 and 7.36 hours respectively and then it does not increase to a level as high as 12 hours.

Figure 2 Cloud cover (%) variation form April 1 to June 30, no bar means data not available, 2012 (top) 2013 (bottom)

Similar behaviour can be observed in 2013 too. As higher cloud cover increased occurred on May 25, 2013, cloud shape life time brings to 6.42 hours on that day from 9.83 hours on previous day. This is a 34.75% drop which remains at that low life time as low as 5 hours up to mid-June.

Figure 3 Cloud shape stability variation form April 1 to June 30, no bar means data not available, 2012 (top) 2013 (bottom)

We can also see that the cloud shape life time drop occurred on May 20, 2013 by 37.66% than the previous day. However, the following day again it goes up by 50%. In addition, on May 28, 2013, once again we can observe the life time diminish by 38.98% but that drop brings 4.92 hours of life time on May 27 to 3.0 hours on May 28 which is not higher than 5.0 hours either of days and it is not a reasonably long run downward trend. Therefore, the life time drop occurred in May 25, 2013 is unusual.

According to the literature, summer monsoon first appeared in South China Sea (SCS) and moves progressively northward [1]. The consistency between the above said dates and onset dates of the previous studies on south china sea and indian monsoon region are significant [1, 9].

In the literature, we can find a considerable amount of studies for defining onset of the monsoon. Some of them are based on multivariable. Certain studies have tried to define the onset using a single variable. However, resulting onset dates for some years were greatly divers because the choice of local variables are sensitivity to the region [1]. One of salient feature of most definitions is that a remarkable change the value of chosen factor or factors and sustains that value to some reasonable period. For instance, Wang et al define the South China Sea (SCS) summer monsoon onset as an index of 850 hPazonal winds average over the central SCS (USCS) is positive and remains positive on subsequent days (3 pentads) with accumulating mean greater than one [1]. Moreover, according to Tsing Chen et al wind field and rainfall are other important widely used factors for determining the onset. Again, these factors reach a certain critical amount and sustain that value for some consecutive days which is adopted as the criterion of monsoon onset [10].Similarly, we are able to identify the remarkable change of cloud shape life time and the cloud amount and sustain that change a considerable number of days in the same fashion.


In 2012, within the potential SW monsoon onset period of 2nd week of May to 1st week of June (according to the previous studies), highest cloud cover increase occurred on May 19 and it is 48.56% higher than the previous day. After it goes up it remains in that value of around 40 per cent most of the days till mid-June and averagely higher in the entire peak SW monsoon period (June-September). On the other hand, cloud shape stability on the same day is dropped by 51 per cent than the previous day. Again this is the highest drop within the monsoon onset potential period. Similarly after the cloud shape life time goes down, it remains till mid-June and mean value is as low as the same within the whole monsoon peak period. Similar observation can be seen in that period of 2013 too. Highest cloud cover increase of 48.47% than the previous day is occurred on 25th May, 2013 and it remains till mid-June at as high as 40%. Moreover, the cloud shape life time too behaves as same as the previous year. When compared with the previous day it is dropped by 34.75% on 25th May and that this low value remains in subsequent days till mid-June.

In the light of the above findings, there is a sound reason to explain that the level of cloud cover and their shape life time are good indicators for determining the arrival date of the SW monsoon. On the other hand determining the monsoon structure such as arrival and withdrawal date, break time and its strength is a challenging demand. In the future, we expect to develop the image processing tools for capturing the structural information of the satellite cloud images in order to describe and figure out weather patterns as well as monsoon.


B. Wang, Linho, Yongsheng Zhang, and M. M. Lu, 2004: Definition of South China Sea Monsoon Onset and Commencement of the East Asia Monsoon, Journal of Climate, 17, 699-710
Bigun, J., Gosta, H. Grandlund, Optimal Orientation Detection of Linear Symmetry, 1987, Proceedings of the IEEE First International Conference of Computer Vision. London, IEEE Computer Society Press, pp 433-438
Bryan A Baum, Vasanth Tovinkere, Jay Titlow, and Ronald M Welch, 1997: Automated Cloud Classification of Global AVHRR Data Using a Fuzzy Logic Approach, Journal of Applied Meteorology, 36, 1519-1540
D. R. Sikka, and Sulochana Gadgil, 1980: On the Maximum Cloud Zone and the ITCZ over Indian Longitudes during the Southwest monsoon, Monthly Weather Review, 1840-1853
H. L. Premaratne and J Bigun, 2002: Recognition of Printed Sinhala Characters Using Linear Symmetry, The 5th Asian Conference on Computer Vision, Melbourne, Australia, 23-25 January 2002
Margaret M Wonsick, Rachel T Pinker and Yves Govaerts, 2009: Cloud Variability over the Indian Monsoon Region as Observed from Satellites, Journal of Applied Meteorology and Climatology, 48, 1803-1821
Nilanjan Ray, Dipti Prasad Mukherjee and Jyotirmoy Das, 1999: Identification of Tracer Clouds: A Shape-based Approach, Current Science, 76, 916-923
S. Ramanayake, Lareef Zubair, and H. B. Nayakekorala, 1998: Week of Onset and Withdrawal of the Southwest Monsoon in Sri Lanka, SLAAS annual conference 54:251-252, December 1998
Tsing Change Chen and Ming Cheng Yen, 1994: Interannual Variation of the Indian Monsoon Simulated by the NCAR Community Climate Model: Effect of the Tropical Pasific SST, Journal of Climate, 7, 1403-1415
Tsing Change Chen and Jau-Ming Chen, 1995: An Observational Study of the South China Sea Monsoon during the 1979 Summer: Onset and Life Cycle, Monthly Weather Review, 123, 2295-2318
Tsing Change Chen, Ren Yow Tzeng and Ming Cheng Yen, 1988: Development and Life Cycle of the Indian Monsoon: Effect of the 30-50 Day Oscillation, Monthly Weather Review, 116, 2183-2199