Genoa low signal in the GFS output for May 11 to 13

Today’s 06:00 UTC operational run of NCEP’s Global Forecasting System (GFS) shows a signal predicting the formation of a Genoa low in the second decade of May, leading to noticeable precipitation.

Precipitation from GFS output (operational run 30 April 2014, 06:00 UTC).

Precipitation from GFS output (operational run 30 April 2014, 06:00 UTC).

In the past, cyclones with this genesis have led to severe flooding in Central Europe. For instance,

However, GFS signal’s lead time is larger than 10 days and thus quite uncertain, of course. Therefore, it is a good idea to have an eye on the further meteorological development but we better (saying it with Douglas Adams) don’t panic…

PS: Thanks to Karol Krizka for sharing his experience with ImageMagick as an animated GIF creator.

Locally intense convective rainfall event

The current atmospheric conditions over Saxony feature a zone of labile and relatively moist air, influenced by a low pressure area over the Atlantic and a high pressure ridge, extending from the Norwegian Sea over South Scandinavia to South-eastern Europe. These conditions cause convection which leads to thunderstorms with locally intense rainfall.

Yesterday, the Weißer Schöps catchment was hit by such a convective event (see Fig. 1) with intensities of up to 30 or 40 mm between 17:00 and 19:00 Central European Summer Time (UTC+2h).

Figure 1: Rainfall radar.

Figure 1: Rainfall radar data, 04/22/2014, cropped to the extends of Saxony. Rainfall heights given in mm. Data courtesy of Meteomedia Group.

The rainfall led to an immediate water level rise at gauge Holtendorf (54.23 sq kilometre catchment), less than one hour after the onset of the rainfall. The water level went up from a mean low flow level (27 cm) and exceeded the second alert level (fixed at 200 cm) by 13 cm in less than two hours (Fig. 2). During this span of time, the discharge multiplied by factor 100.

Figure 2: Gauging data for Holtendorf, Särichen (both Weißer Schöps) and Schöps (Schwarzer Schöps).

Figure 2: Gauging data for Holtendorf, Särichen (both Weißer Schöps) and Schöps (Schwarzer Schöps).

Due to the very local character of the event, alert levels at the next gauge in downstream direction (Särichen) were not exceeded. Moreover, neighboring gauges like Schöps at Schwarzer Schöps (around 10 km air-line distance from Holtendorf) showed absolutely no signal.

A glance at the rain gauge data underlines the strong spatial heterogeneity of the rainfield. Whereas the DWD gauge (German Weather Service) at Görlitz (approximately 1 km southwest of gauge Holtendorf) recorded around 23 mm in 2 hours, the Görlitz gauge, ran by Meteomedia (located approx. 2 km west of the DWD gauge) recorded only 6 mm in the same time interval (see Fig. 3).


Figure 3: Rain gauge data of the event.

Flood forecasting by empirical means ;-)



The shots were taken on Dresden’s Prießnitzstraße, located along the final mile of the Prießnitz Creek, a small tributary to the Elbe River. The area was flooded in 2002 by backwater effects. Although the flooding was not caused directly by the Elbe flood, the high water mark is related to the highest observed water level at Elbe gauge Dresden-Altstadt. And of course, be prepared for what is to come!

End-user’s sight on lead times of flood warnings

The last couple of weeks, I was conducting an online survey on different aspects related to the perception and applicability of specific flood warning products, distributed by the Saxon Flood Centre.

In this flood-warning context, I was wondering, what would be the smallest lead time which is considered sufficient by the end-users in order to undertake appropriate measures. Although the survey was carried out anonymously, I was able to correlate lead times and the area of the users’ focal catchments.

This is how it looks like:

Lead Times

Desired lead times vs. users’ (log) catchment area

Essentially, the results tell that there is only a slight tendency towards smaller (empirically) required lead times in smaller catchments. The question is, what moves people to desire a three-days lead time for a 40 square-kilometer catchment?

Another question would be: do people actually believe that hydrological forecasts with a three-days lead time in small catchments could be more accurate than quantitative precipitation forecasts? No doubt, anyone of the end-users would be OK with inaccurate three-days rainfall forecasts but does this apply for hydrological forecasts, especially if damage-prone events are not detected?

The survey covers a number of other interesting questions which are related to my current work in a way that I am trying to line out the interplay of end-users requirements and the technical feasibility of a limited area flood-early warning system for the Free State of Saxony.

Soon, I will inform on more results of the survey (a paper is in preparation).