Skip to end of metadata
Go to start of metadata

 Status: Finalised Material from: Fernando, Linus


1. Impact

After heavy rainfall in mid August, severe flooding in the southern Indian state of Kerala have killed more than 350 people.

2. Description of the event

For the India Meteorological Department web site where was shortly after the event a report for the public. Some of the plots below were copied from that document. Before going to the point a few words about the large scale dynamics which led to a very wet days . The chart below shows the 850hPa wind speed and MSLP across the southern Asia on the 15th of August. We can see the low level monsoon circulation extending from East Africa to China sea. In the Bay of Bengal a depression centred to east of the coast of the India state of Orissa, is moving inland to northwest-west direction interacting with the monsoon winds (strengthening)  on the southern flank of the cyclone.

The next plot shows the average of the daily values over Kerala region since the begin of the monsoon season (1st June). Last week was particular wet as we can seen from the graph, namely  the 15-16th of August. The departure from the normal is quite exceptional for that region.

The totals for 9-15 August are displayed below. Values above 500 mm can be seen in Kerala mountains. The rainfall distribution highlights the Western Ghats mountain range situated along the west coast of India.

The plots below show TRMM rainfall (calibrated micro+IR) plots for 14, 15 and 16 August.

3. Predictability


3.1 Data assimilation

 An interesting feature which may or may not related with the dramatic shift of the ENS rainfall distribution in the last forecast run compared with the previous (see plots below) is the fact that the computed wind vector increments at the begin of the assimilation window for that particular forecast (15th August 0Z) tend to slow down the monsoon  circulation over the south-southwest region of India, so the significant increase of the rainfall in the ENS in that particular forecast run cannot be attribute to a strength of the winds passing the mountains responsible for orographic precipitation enhancement. On the other hand there is a significant increment of the specific humidity at lower levels (small panel) in the south-southwest region.

The tephigram below shows the first guess, analysis and true sounding for a location in Kerala. As expected the true  profile shows a saturated deep layer with both model profiles trying to mimic the reality. The freezing level is lower in the model profile when compared with the observation. Also the model profile suggests the presence of a cloud layer top below 700 hPa which is not evident in the observed profile.

3.2 HRES

The first plots below shows (the few) observations of 24-hour precipitation 15 August 03z to 16 August 03z and the following plots HRES forecasts from different initial times valid the same time. it is evident that the last forecast before the event (15 August 00UTC) is more extreme than the previous ones for Kerala but also over the sea to the west.

The left panel below shows the accumulated precipitation on a point in Kerala region (10N, 77E) for the HRES forecasts started at 14@12, 14@18 and 15@00 for the first days (for the first 2 runs the amounts valid at 15@00 are subtracted).  The forecast started at 14@18 shows larger values compared to the previous forecast run, but for the last forecast run the precipitation values are twice the ones seen by the previous forecast in the first hours. The total column of water vapour time series (right) for the same lat-lon point shows a sharp rise during the first hours (purple curve) before dropping to a level which is still larger than the other two previous forecasts.

3.3 ENS

The plots below show EFI and SOT for 3-day precipitation (15-18 August). Also the last ensemble forecast was more extreme than previous ones. Some signal appeared on 13 August Kerala region.

The plot below shows the evolution of ENS (blue), HRES (red dots) forecasts and model climate (red) for 3-day rainfall in 10N, 77E (black hourglass in HRES plots above). The plot includes ENS forecasts every 6 hours from 12 August and onwards. This plot confirms that the jump in the forecast appeared from 14 Aug 18UTC to 15 Aug 00UTC.

3.4 Monthly forecasts

The plots below show the weekly anomalies (13-19 August) of precipitation from extended-range forecasts. The forecast from 9 August showed a weak anomaly, while earlier forecasts had a weak wet anomaly further to the north.

3.5 Comparison with other centres

4. Experience from general performance/other cases

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

  • Large difference in rainfall between 14 Aug 18UTC and 15 Aug 00UTC forecasts for 15 August.

6. Additional material