The South Asian Monsoon; same as it ever was?

What is happening to the Indian Ocean monsoon? Has it become less predictable? Is it becoming affected by global warming? and finally, are droughts in Sri Lanka getting worse as a result?

The monsoon rains are important not only for agriculture in the region but also power generation.  Sri Lanka generates around 40% of its electricity from H.E.P. for example. So getting an understanding of how the monsoon seasons work is really quite important for a whole range of reasons.

a.  In the first of three linked articles I am going to be analysing rainfall and drought data to find out what is actually going on.

b.  The second article will look at what drives the monsoon, in particular the interplay of 3 factors and how they lead to changing patterns of sea surface temperature ( SST ), pressure and wind patterns, and how this affects rainfall. The three phenomena are:

  • The migration of the Inter Tropical Convergence Zone ( ITCZ )
  • The El Nino/La Nina events which occur in the Pacific Ocean ( ENSO )
  • The Indian Ocean Dipole ( IOD )

c.  The third article will focus on patterns of drought in Sri Lanka.

Analysing Rainfall: getting the data

For this article I am using rainfall data for 4 stations; Batticaloa, Jaffna, Colombo and Galle. I had data for the period 2000 – 2015 and added to that data for the same locations for the period 1985-90. (I would have liked more ie; 1980 – 2000 but I couldn’t access the data).

So my sample size is  small and skewed towards the later period but it does give some indication of trends.

I chose some simple statistical methods to analyse the data;  I looked at each month in turn  over the 15 year period and calculated for each month and for each station:

  • the mean rainfall
  • standard deviation
  • coefficient of variation; it was this measure that I was really looking for; see below:

The coefficient of variation  ( Cv )is a measure of the spread of data that describes the amount of variability relative to the mean. It is calculated by dividing the standard deviation by the mean.

values close to zero indicate that the the data set shows a lower degree of variability and vice versa; In the results section  I will give just the Cv (not the mean or SD )

So if the data shows a Cv of say 0.50 what that suggests is that for any one year  the actual rainfall received will be in a wide range: 50% above and below the mean. Let’s say average rainfall is 500mm for a month then with a Cv of 0.5  the actual rainfall could be expected to fall within a wide band 250 mm to 750 mm.

Not very predictable.

I carried out the same calculation for the 1985-90 periods so that I could compare the two. I also looked at the pattern of rainfall through the year to see if it changes at all. I wanted to know the following:

  1. How variable is the annual rainfall total from year to year for each station
  2. For any given month how variable is the rainfall total over the 15 year period, and in comparison with the shorter 1985-90 period.
  3. Whether the amount of rainfall during the monsoon periods changing and if so how?
  4. Whether the distribution of monthly rainfall changed significantly over the period; is the monsoon coming earlier or later?

 

The location of the 4 weather stations

 

The monsoon seasons in Sri Lanka;

There are four monsoon seasons in Sri Lanka:

Period Name Comment Regions Affected
March – Mid May First Inter-monsoon (FIM) limited impact
May-July South West Monsoon (SWM) S.W. Winds Heavy rain Southern Coast, South West, West
October- November Second Inter-monsoon (SIM) S.W. winds Heavy rain and tropical cyclones possible South and South West, East coast
November – December North East Monsoon (NEM) North east Winds North and East

That has an impact on the rainfall distribution for different parts of the country;

  1. Colombo and Galle  in the south west of Sri Lanka both show two rainfall peaks during the year coinciding with the SWM and SIM periods.  (not one as I guess many studnets living in Europe ans Notice that the rainfall for SIM is higher on average than for the SWM; not what you would expect from the text books?

                                                                                                           SWM                                            SIM

note: what becomes apparent is that where the monsoon seasons are concerned you cannot generalise; Sri Lanka is different from the sub continent of India. Even within India there are significant departures from the generalised “norm”; so the lesson is not to accept broad generalisations from text books where climate is concerned.

 

  1. Batticaloa is on the east coast and has a different rainfall pattern; one that is dominated by the NEM.

                                                                                                                                                                       NEM

note the vertical scales on the two graphs are not the same; the graphs are there for illustrative purposes only

Of the two, Batticaloa looks like it is the most vulnerable to drought for two reasons;

  1. there is a long dry period from March through to November when temperatures and evapotranspiration rates are high
  2. The east coast is heavily dependent therefore on the NEM; if it fails to produce enough rain in November and December, or fails altogether then there is much less groundwater available for crops following on. Reservoirs (called tanks locally) and rivers dry up.

 

when the rains fail

What the data shows

Looking at the data there are a couple of general points to begin with:

  1. The onset of each monsoon period is pretty much fixed give or take a week or so although there is a suggestion that the SWM is arriving slightly earlier ie late April / early May rather than later in May.
  2. Actual rainfall for any given month varies quite substantially from average values for that month. The coefficient of variation is high.  That is true for both the 1985 period and also the 2000-2015 period. So the monthly averages don’t mean a great deal. Rainfall is variable for any given month and from year to year. From the data I have, I suggest it always has been.
  3. The NEM is a changeable event; some years wetter some years drier., but not predictable
  4. The SWM rainfall is on the increase since 2007
  5. In the North and East the dry season seems to be getting drier, if you add that to a significantly lower monsoon rainfall  total as in 2005 – 8, it can spell big problems for farmers: in particular, drought.

 

Which bring me on to the last point in part one; why is the rainfall so unpredictable? The reason is that there are several factors at play.

  • Rainfall in the Indian Ocean basin is determined by wind direction; which in turn is heavily influenced by the migration of the ITCZ.
  • However it is also influenced by two other phenomena; which are at least partially dependent on one another..
  1. The ENSO or El Nino event
  2. The 3 phases of the Indian Ocean Dipole;

all of which affect sea surface temperatures, and therefore pressure and wind systems.

Simple it isn’t?

As a taster then here is something to be thinking about.

ENSO events:

weak            2004/5, 2006/7

moderate      2002/3, 2009/10

very strong    2015/6

IOD dipole:

positive:         2006, 2012

negative:        2010

You could have a look at the summary of rainfall data above and see where there may be potential match – ups.

Part 2 looks at how it all works

Appendix: Summary  of Results

( for those who are interested in the detail; I have the raw data available on request )

  1. Batticaloa; main rainfall season is the North East Monsoon (NEM)

The average Cv for 1985-90 is 0.67; the average Cv for 2000-2015 is 1.01

However the Cv for the NEM is marginally lower for the period 2000-15

Concentrating on the period 2000-2015:

  1. Cv is higher during March to October (dry season); generally >1.0
  2. Cv falls slightly during the NEM season; between 0.38 and 0.56
  3. overall drier years in 2001, and 2005/6/7:  ? drought ?
  4. July is the driest month and there is a suggestion that July is becoming drier over the period: (2000-07 av. 35.75; 08-15 av 24.7)
  5. The NEM generally starts in November although in 2004,2011,and 2015 it arrived in September
  6. December is the wettest month
  7. NEM rainfall was less for the period 2005-2008 and also 2013
  8. 2011 was the wettest year during the period at 3581mm (80% above average)
  1. Jaffna: main rainfall season is the North East Monsoon  (NEM) plus possibly the Second Inter-monsoon (SIM)

The average Cv for 1985-90 is high; 0.86 and is even higher in 00-15; 0.97

Looking at the period 2000-15

  1. Cv is much higher during drier months; range 0.72 – 1.46 and lower during the NEM at around 0.47
  2. Cv is also lower at 0.44 during October (SIM)
  3. drier years were; 2002/3, 2005/6, 2009, 2012/3
  4. June/July are the driest months and are becoming drier; 2002-7 av 48.5mm, 08-15 av 18.4mm
  5. NEM arrives in November in 12 of 15 years
  6. November is wettest month
  7. NEM rainfall was much lower 2006-8 and 2013/4; 2009/11 NEM rainfall was above average
  8. 2015 was the wettest year in the period at 1839 mm but was only 10% above the average for the NEM

A possible question to investigate is the degree to which Jaffna may be affected by the SIM given its location.

Common to both

  • high variability from year to year especially in the dry season
  • decreasing rainfall in June/July
  • drier 2005-8 and 2013
  1. Colombo; affected by 3 seasons. First Inter Monsoon (FIM), South West Monsoon (SWM), Second Inter Monsoon (SIM); although FIM impact is negligible

at around 0.06 the Cv for both periods (85-90 and 00-15) is high

Looking at the period 2000-15

  1. Cv is lower during both the SWM and the SIM (0.32 and 0.43)
  2. drier years overall in 2004 and 2011
  3. No significantly drier years apart from 2011 which had a lower SIM
  4. January and February showing decreasing rainfall; example Jan av. 00-07 av 108.4, 08-15 av 75.1
  5. July is the driest month October/November (SIM) the wettest
  6.  September marks SIM arrival except 2005 and 2011
  7. April/May consistently marks start of SWM
  8. signs that onset of SWM  shows higher average: 00-07 av 195mm, 08-15 av 371mm not quite so marked for SIM
  9. 2010 was the wettest year at 3370mm (43% above average)
  1. Galle: affected by 3 seasons. First Inter Monsoon (FIM), South West Monsoon (SWM), Second Inter Monsoon (SIM); although FIM impact is negligible

In comparison with Colombo the Cv’s are slightly lower than for Colombo but still >0.5 for both periods but there is no significant difference between the Cv values for the two time periods.

Looking at the period 2000-15

  1. Cv is does not drop during SWM although it does so for the SIM
  2. 2001/2 and 2013 were drier years,
  3. SWM shows increasing average 00-07 av 162.3 mm; 08-15 av 268.1 with similar increase for May; The SIM data does not show a trend betond a slightly drier October and a slightly wetter November
  4. January shows a decreasing average from 00-07 av 116mm to  08-15 av 82mm
  5. January is the driest month, October is the wettest month
  6. October marks the arrival of the SIM
  7. April/ May marks the arrival of the SWM
  8.  As with Colombo signs are that the onset of the SWM is bring heavier rainfall; 00-07 av for April was 162mm for May was 241mm and for 08-15 av for April increased to 268mm and 298mm respectively.
  9. 2007 and 2010 were the wettest years (32% above average)

Common to Colombo and Galle

  • rainfall is decreasing in January
  • rainfall is increasing in April; onset of SWM is bringing heavier rainfall
  • suggestion that monsoon is getting earlier