MSG FOR NOWCASTING - EXPERIENCES OVER SOUTHERN AFRICA Estelle de Coning and Marianne König South African Weather Service, Private Bag X097, Pretoria 0001, South Africa EUMETSAT, Am Kavalleriesand 31, D-64295 Darmstadt ABSTRACT The Global Instability Index (GII) product, which is routinely derived from Meteosat-8 measurements, comprises a list of empirical instability coefficients (Lifted Index, K-Index, KO-Index) and the total precipitable water as an additional airmass parameter. Such instability indices describe the thermodynamic state of the atmosphere, help to identify pre-convective states and potential of storm development. As a satellite product, the GII overcomes the limited spatial and temporal resolution of the traditional radiosonde derived indices. The operational GII is derived over pixel groups of 15 by 15 pixels, and it is confined to cloudfree regions. Offline computations of the GII can be done over smaller areas for selected case studies. The South African Weather Service currently evaluates the operational GII product for its regional nowcasting potential. Convection development and occurrence of severe storms over Southern Africa are analysed to assess the warning potential, in terms of lead time and storm intensity, of the Metsoat-8 GII product. Focus is here the 2005/2006 summer season. In addition, the satellite product either from the operational processing or from offline processing for selected cases - is compared to ground, radar, and radiosonde observations for further validation. 1. Severe weather over South Africa on 6 Nov 2005 During the weekend of 4-6 November 2005 heavy rainfall across South Africa brought welcome relief after very hot (heat wave) conditions over most of the country. The persistent heavy falls lead to local flooding in any areas over the south-eastern parts of the country where rainfall figures in excess of 150 mm in 24 hours were reported at in the Eastern Cape (Figure 1 and 2). Figure 1: Map of South African with provinces Figure 2: 24 hour Rainfall total for 6 November 2005, ending at 06:00 UTC on 7 November 2005, A merged product of satellite, radar and rain gauge measurements.
Over the northern parts of the country large amounts of rain fell in short periods of time leading to local flooding and considerable damage was caused by strong wind. Electricity cables were damaged which led power outages, roofs were blown off and informal settlements were destroyed. In some villages east of Polokwane (23.8S and 29.4E) a 10 year old boy drowned when his shack washed away, three women were injured when their houses collapsed and 60 newly built houses were destroyed. West of Polokwane power lines were damaged: five of the pylons were damaged and one blown over, each of these weighing about 20 tons (Figure 3 a and b). The time of this damage was reported to be 15:27 UTC. Figure 3a and 3b: Damage to power lines On the 4 th rainfall amounts over the northern most province of South Africa were less than 10 mm, but on the 5 th rainfall figures were between 20 and 50 mm. The rain of the previous two days left the ground fairly moist in this area and thus an area of cloudless, rain-cooled existed over the Limpopo province in the early hours of the 6 th. Convection started west-northwest of the province on the surface dry line and migrated towards the east-northeast (along with the upper air flow) and first entered the area of interest during the early afternoon. 2. Satellite image interpretation At 11:00 UTC the RGB composite image shows the first explosive development with strong updrafts on the western border of the moist area (Fig 4a). Thirty minutes later, a second severe storm develops further to the northeast (Fig 4b). The bright yellow colour indicates cold, thick ice clouds with small ice particles (strong updrafts). Figure 5 a and b: Met-8, 06 November 2005, 11:00 UTC (a) and 11:30 UTC (b) RGB Composite WV6.2-WV7.3, IR3.9-IR10.8, NIR1.6- VIS0.6
At 12:30 UTC (Fig 6b), both storms show strong development (strong cooling rates, strong increase of storm area, small ice particles) with signals of a U-shaped storm. Using IR10.8 colour enhancement the cloud top temperature of these very severe storms is indicated at -85 C (Figure 7). Figure 6 a and b: Met-8, 06 November 2005, 12:00 UTC (a) and 12:30 (b) RGB Composite WV6.2-WV7.3, IR3.9-IR10.8, NIR1.6- VIS0.6 Figure 7: Met-8, 06 November 2005, 14:30 UTC Channel 09 (IR10.8, colour enhanced) In the following hours, the southern storm moves into the Limpopo province, crossing exactly the area where the damage occurred to the electricity towers (west of Polokwane) as indicated in HRV (Fig. 8). Figure 8: Met-8, 06 November 2005, 14:45 UTC Channel 12 (HRV)
Using the MSG satellite imagery one would have been able to detect signs of this severe storm hours in advance which shows the capability of MSG imagery as a short term forecasting tool. 3. Indices of instability It is a widely used practice to make use of instability indices that would indicate whether or not the conditions are favourable for thunderstorm development. Traditionally we would calculate these from temperature and humidity soundings by radiosondes, but these are only done at selected site across the country and also only twice daily. Figure 9 shows the upper-air sounding profile from Polokwane at midday. It shows that the K-index was more than 32, but the Lifted index only -0.5 and indications of instability from the Showalter index were very neutral. Since numerical weather prediction models can also forecast these factors of the atmosphere, they can used to create grid point atmospheric profiles. Figure 9: Upper-air sounding at Polokwane at midday on 6 November Using a 24 hour forecast for 06:00 UTC with ECMWF, the K index would be showing values more than 30, which is also a good indicator for instability and thus thunderstorm development (Figure 10). Figure 10: ECMWF 24 hours forecast for 06:00 UTC showing K index values of more than 30 in yellow
4. Global Instability Index (GII) The Global Instability Index (GII) provides a set of instability indices which describe the layer stability of the atmosphere. A local profile is derived with: the satellite measured brightness temperatures (for MSG channels 5,6,7,9,10 and 11) and some a priori information of the atmospheric profile (from the Numerical Weather Prediction model, in this case ECMWF). Various instability indices are computed from this local profile, with the prerequisite that calculations can only be done in clear air conditions. Using the GII s K-index, and a resolution of 15X15 pixels, values of more than 40 could be seen in the area where the severe weather occurred (Fig 11) at 06:45 UTC. Figure 11: GII s K index, at 06:45 UTC, showing values of more than 40 in the area where severe weather occurred. At 12:00 UTC the storms were already developing, but in the cloud free areas the GII s K-index still shows values of above 40, indicating that thunderstorm development was extremely likely (Fig 12). Figure 12: GII s K-index still showing values of more than 40 in the relevant area at 12:00 UTC
Using the K-index as calculated by the GII, it would have been possible to see signs of the possibility and severity of these thunderstorms early in the morning already, and these signs persisted to later in the day. The GII can thus be applied as a short term and nowcasting tool, together with the other MSG images. 5. Conclusions MSG imagery offers a lot of useful information about the severity of storms in real time. Using the GII product the possibility of severe weather can, however, be anticipated more than 6 hours in advance and it can be updated every 15 min when a new image is available to see whether the features are persistent. Closer to the time RGBs and HRV provided excellent guidance on the severity of these storms. Using MSG tools the decision making process to forecast severe convection can be aided in order to reduce risk and damage to property. Future considerations for further studies would include an investigation into the possibility of calculating the rotation in the mesoscale flow off the MSG images, investigation in the effect of improved resolution and, of course, the gathering of more cases to be able to generate evaluation statistics of this methodology.