Status: Ongoing analysis Material from: Linus, Ivan, Mohamed
1. Impact
On 28 March 2017 the tropical cyclone Debbie made landfall on the coast of Queensland in Australia. The cyclone brought extreme wind and rainfall, leading to widespread floodings.
2. Description of the event
The plots below show analysis of MSLP and short precipitation forecasts from 26 March 12UTC to 28 March 12UTC, every 12th hour.
3. Predictability
3.1 Data assimilation
Some deterioration of the analysis of tropical cyclone Debbie was noticed for the cycle from 12UTC on 26 March. The reported core pressure of the cyclone was 976 hPa and the first guess was not bad with a central pressure of 983 hPa. The analysis though appears much shallower with 998 hPa and the symmetrical structure of the tropical cyclone was also destroyed. The deterioration of the analysis can also be seen by comparing the forecast winds with ASCAT: forecast winds around the centre are stronger and in line with the ASCAT winds.
TC Debbie and ASCAT winds.
The following investigation found suspicious observations from a ship and an operational like experiment (gp2b) were these observations were removed. By doing this it is clear that the analysis considerably improved although it is still not as good as the FG. Later it was found that before it started reporting pressure which looked wrong its position was at about 10 degrees south and suddenly it jumped to 19 degrees south and reported from that sector for most of the day. Later its position jumped again to 8 degrees south. This shows that the ship actually reported wrong position not wrong weather data. It was a bit unfortunate that the wrong position appeared to be in the close vicinity to the centre of Debbie. In the forecasting system there is a check of ship position and if it differs more than a certain threshold the ship is blacklisted. Unfortunately, this automatic check is not applied to the new BUFR data.
FG-OBS and AN-OBS for operational run (top) and the experiment (bottom).
3.2 HRES
3.3 ENS
The plots below show tropical cyclone tracks (ensemble - grey, CF - blue, HRES - red, obs - black) and positions on 27 March 12UTC (symbols coloured by intensity) for forecasts with different initial times (latest in first panel). The forecasts from 25 March and before had a strong northward bias.
The plots below show time-series of central pressure (top), maximum wind speed (middle) and propagation speed (bottom) for the forecasts from 27 March 00UTC (left), 26 March 00UTC (middle) and 25 March 00UTC (right). Note how bad the intensity was handled in the latest forecast (27 March 00UTC).
