Skip to end of metadata
Go to start of metadata

 Status:Finalised Material from: Fernando, Linus


Discussed in the following Daily reports:


1. Impact

The tropical cyclone Roanu formed from a cyclonic vortex north of Sri Lanka and moved along the Indian coast and finally made landfall in Bangladesh. A huge landslide in Sri Lanka caused more than 90 deaths after three consecutive days of heavy rain ( The worst rainfall seems to have happened on 15 May. Later mass evacuations took place along Indian and Bangladeshi coasts. The vortex seems to have been associated with an MJO event. The cyclone took place a the same time as a heatwave in western India 201605 - Heatwave - India.

2. Description of the event

The cyclongenesis seems to have been triggered by an MJO event in the Indian Ocean. Below is the a sequence of satellite (VIS) images since 30 April up to 15 May every other day. The first couple of images shows very little clouds in the equatorial region. But as the days pass the clouds start to build up and organized in large clusters. In the end of the sequence Sri-Lanka and Sourthern India were under strong convection.

The passage of the MJO event left behind twin cyclones in either side of the equator, where the northern developed into TC Roanu.

Below are figures from showing the flow on 850 hPa decomposed into balanced (Rossby) modes, unbalanced (ID) modes and Kelvin modes for the 15 May 00z analysis.

3. Predictability


3.1 Data assimilation

 The increments from 15 May generated for these twin cyclones two days ago can be seen below. The chart depicts the magnitude of the wind analysis increments above (below) the 90% (10%) percentile of model increments CDF based on the last 90 days at the begin of the 4DVar window at a model level around 600 hPa. The analysis is spinning up the cyclone in the SH while attempts to spin down the feature located in Sri Lanka.

In the early stages of the cyclone development in Sri Lanka we have noted a large region with wind increments in the lower troposphere covering most of the Bay of Bengal at the begin of the 4DVar window 17@21. The adjustment of the wind field affects not only the circulation near the tropical depression but also the large scale monsoon circulation to south and east of the Bay of Bengal. The inset box shows the analysis of winds at 850 hPa valid at 18@00.

3.2 HRES

The plots below shows the 24-hour accumulated precipitation for 15 May 00z to 16 May 00z from HRES forecasts from difference initial times.

3.3 ENS

The plots below show the forecast of tropical cyclone activity 20-22 May. In early forecast the highest probabilities were for the cyclone to be on the western side of India but changed on the 16 May to mainly be on the eastern side.

The plots below show the EFI and SOT for rainfall valid 14-16 May; the period when flash floods took place in Sri Lanka.

The plots below shows the same as above but only for 15 May:

The plots below shows the probability of more than 150 mm precipitation on 15 May from different initial times:

The plot below shows the 24-hour precipitation from ENS and HRES valid 15 May for a point east of Colombo (7N, 80E):

3.4 Monthly forecasts

The plot below shows the verification of precipitation anomaly for 9 May to 15 May (left) and 16 May to 22 May (right).

The plots below show the MJO forecasts:

3.5 Comparison with other centres

4. Experience from general performance/other cases


5. Good and bad aspects of the forecasts for the event

  • Shift from west-coast to east-coast around 16 May.

6. Additional material