| Monsoon Pundits closer to accurate prediction |
| |
The Monsoon Onset over Kerala and its subsequent march across the Indian subcontinent is one of the much-awaited events by the Indian farmers. Farmers from this area are dependent on the monsoons for their crop productivity. The country's economy flourishes if there is good monsoon. A high GDP in India has always been led by growth in agriculture a result of good monsoon. Monsoons are therefore of paramount importance to the policy makers to take bold decisions in furthering the economy of the country. The monsoon especially the southwest (SW) monsoon, also called the summer monsoon, bring about 80% of annual mean rainfall - is most uncertain. How much would be the level of uncertainty? The data in past shows that in 1918, the first spell of showers in Kerala, India (technically termed as onset of monsoon') was on 8 May and in 1972 it was as late as 22 June ! Normally the date of onset of monsoon in Kerala is around 1 June. In other parts of India, it progresses slowly.
|
The onset dates of monsoon
over different parts of the Indian subcontinent
|
|
|
Thus there is no wonder that the monsoon predictions are important. Accurate predictions help farmers in selecting right seed and time for sowing for a better yield. To predict monsoon, three important aspects, are always considered: (a) its onset, (b) quantum of rainfall and its space and time variation during the season, and (c) monsoon intensity (spells of showers - i.e. active, weak or break in monsoon conditions).
In spite of several attempts, the predictions fail and the mystery of air-sea interactions make scientists to work and rework, find alternative tools, methods and models. In attempting the predictions of the onset of monsoon, the closer to reality method was suggested earlier by using weekly March-May (pre-monsoon) rainfall data at certain locations at south of 13 degrees North. This was based on the observations that the Pre-Monsoon Rainfall Peak (PMRP) is about six weeks prior to the onset of the monsoon over Kerala. In scientific terminology this data yielded a statistically significant correlation between predicted date and actual date.
With the recent research at NIO, we now have a better chance to predict the monsoon onset. The correlation coefficient between predicted and actual onset date is 0.64 the closest of existing prediction methodologies.
|
| |
 |
Compsite of GPCP rainfall for 23 years (1979-2001) with the onset date coinciding with the 0 pentad. The PMRP can be clearly seen six pentad prior to the Monsoon Onset over Kerala (MOK) |
|
|
The scientist used satellite data (Global Precipitation Climatology Project GPCP) well distributed both in space and time for PMRP. The real onset of monsoon values during last 23 years were correlated with the PMRP data obtained from 8-13°N and 70-95°E. The choice of this area is based on the shift of the tropical convergence zone to this region during pre-monsoon months. Of course, earlier experiment also used weekly rainfall data from this region but only 4 data points pertaining to the four meteorological sub divisions. The satellite data in this experiment, allowed using more data points (2.5° X 2.5° grids) which was representative of the complete area (70-95 degrees E). Based on this data, the PMRP was found to takes place about six pentads (1 pentad =5 days, i.e. about 30 days) prior to monsoon onset.
|
Considering this as the basis, the tests were carried out for subsequent three years. Analysis using 23 years (1979-2001) of GPCP data indicated that 70% of the predictions can be as close as 4 days (earlier or later) to the onset date.
|
Difference (days) between prediction and onset date |
Number of predictions |
Cumulative predictions |
Predictions (%) |
0 |
1 |
1 |
4 |
1 |
2 |
3 |
13 |
2 |
6 |
9 |
39 |
3 |
3 |
12 |
52 |
4 |
4 |
16 |
69 |
5 |
2 |
18 |
78 |
6 |
2 |
20 |
86 |
7 |
2 |
22 |
95 |
8 |
1 |
23 |
100 |
|
Difference in days between the predicted and actual onset dates
(as given by the India Meteorological Department) |
| |
| So, one can now safely say that about 70% of our predictions would be close to the actual onset of monsoon within the span of 4 days. Interestingly, no Indian station data as well as Indian National Satellite (INSAT) data has gone into the GPCP data. We feel that with the addition of the station (both island and land stations) and INSAT data, the above method has a chance to give much more accurate results as the frequency of data collected by INSAT is much higher than other satellites contributing data to GPCP. |
| |
|
|
To consult original article, refer :
RameshKumar, M.R.
Forecasting of onset of southwest monsoon over Kerala coast using satellite data.
IEEE Geosci. Remote Sens. Lett.: 1(4); 2004; 265-267. |
