Consistent with more rainfall in northern Europe is the prediction of a northward shift and intensification -probably with greater precipitation- of the North Atlantic storm track by Hall et al. (1994) (using the UKMO GCM).
Von Storch (1991), giving an example of a statistical approach in the downscaling of climate estimates (MPI) to regional scales that is consistent with observations, projected a decreasing trend in winter rainfall over the Iberian peninsula.
Schönwiese (1993) indeed observed upward trends in rainfall during winter in Western and Central Europe (period 1891-1990). The statistical significance are highest for the North Sea basin and Scandinavia (p<0.05). However, similar trends for the period 1961-1990 are not significant in statistical terms. Over the period 1961-1990, summers became drier in the North Sea basin (not significant). A decrease in rainfall is observed in the Mediterranean area (Spain, Southern France, Italy and Greece) during the period 1891-1990 (not significant). He also notes a decrease in pressure over Northern Europe and an increase in pressure over Souther Europe in the winter period (over the period 1961-1990). This induces an intensification of zonal circulations in winter.
At a weather station in the Netherlands (De Bilt) the annual means of total precipitation amount and duration increased (+5% and +11%, respectively) for the period 1961-1990 compared to 1931-1960 (Zwart, 1993). (However, one station is not representative for a larger area.) Also, precipitation amounts increased significantly in Bavaria (Germany) during the last 3 decades (Dister, 1995).
The precipitation time-series produced by Jones et al. (1994) for England and Wales confirm these long-term trends in seasonal rainfall.
In this respect, it is remarkable that the Waterloopkundig Laboratorium, 1994) also reveals that in the catchment area of the Meuse in Wallonie the precipitation in wintertime this century (1911-1993) has increased with more than 20% relative to 1882-1910. The KNMI (1995) points to the persistence of zonal circulations during the 1993 and 1995 winters (causing the excessive frontal rainfall). In 1993/94 a westerly flow occurred on 55% of the winter days (average is 40%). In the winter of 1994/95 this is above 60%.
Analysing runoff conditions over the last sixty years in Germany, Schumann (1993) finds similar trends. In total, 14 runoff series from rivers in different regions in Germany were analysed for the period 1931-1988. It appeared that the winter runoff of alpine-influenced river basins has increased by warmer winters since 1965. For non-alpine influenced river basins, average runoff values for the months March to June were larger after 1964. The average runoff values were increased by 26% as compared to the time period 1931-1963. It was also found that the average discharge of annual low water periods was increased. The annual lowest 60-day average of daily runoff was, on average, 29% higher from 1964 to 1988 than in the period 1931-1963. These results were found to hold for river basins differing in size and situated in different hydrological regions of Germany. Because mostly large river basins were analysed, Schumann assumes that changes reflecting human intervention can be neglected. In this context, the remark of Dister (1995) is noteworthy: "Apart from infrastructural changes, the main cause for the 1995 flooding of the river Rhine is a climatological one, i.e. the increase of 40% in precipitation in Bavaria between 1960 and 1990".
A study by Boardman et al. (1994) reveals that, in the last twenty years, there has been an increase in the incidence of flooding of property by runoff from agricultural land in certain areas of northwestern Europe. Some of the high risk areas considered were the Ardennes and the south part of the Dutch province of Limburg. The Royal Meteorological Office at Ukkel (Demaree, 1990) found evidence of an increase in the maximum precipitation intensity for the first catchment over the period 1934-1976/1983, which explains in part the observed increase in flooding incidents in this area. However, in the southern Limburg area, the changes in land use were identified as important factors contributing to the increase in runoff. The final conclusion for these catchments was that "increases in flooding in recent years are primarily the result of changes in land use and the intensification of farming, and that changes in climate (if any) are unimportant" (for the period under consideration). An explanation why the observed increase of rainfall intensity in certain regions is not important was not given.
In the period 1950-75 there were 66 floods in Italy (2.5 per year). During the 17-year period thereafter (1976-1993), the frequency was 4 floods per year (with a total of 126). Since 1950, there have been seven severe droughts, of which four were concentrated in the 1980s. These data seem to confirm that Italy is undergoing a "climate partitioning" into a wet northern region and a southern semi-arid region. This "splitting" is in agreement with general circulation model predictions: rainfall trends should go towards extremes (Ferrara, 1993).
Kwadijk (1993), and Kwadijk and Middelkoop (1994) estimated the impact of climate change on the peak discharge probability of the Rhine river by means of a water balance model. Scenarios for temperature changes between 0oC and 4oC and precipitation changes between plus 20% and minus 20% have been applied. Within this range, flood frequencies appear to be more sensitive to a precipitation change than to a temperature change. From the study it can be deduced that a biennial flood doubles in frequency to become an annual occurrence under the influence of a 20% increase in precipitation. The associated peak flow volume raises by approximately 30%. (The method was only applied for estimation of probability changes of events having relatively low recurrence times.)
From the application of the Business-as-Usual (BaU) scenarios developed by the IPCC (1990; 1992), it became evident that in the Alpine part of the basin winter discharge will increase and over time, as snow and ice have melted, summer discharge will decrease. In winter, the ratio rain/snow changes due to an increase of the winter temperature, and from a shift of maximum precipitation from summer to winter. The summer discharge will decrease, due to an increased evapotranspiration and decreased melt water availability. Downstream effects from precipitation changes become increasingly dominant over those related to temperature changes. The river regime at the Lobith gauging station (where the Rhine enters the Netherlands) will change from a combined snow-melt/rain-fed river to an almost entirely rain-fed river. Application of the BaU scenario ("best guess") reveals that at Lobith winter runoff will increase by 17% and summer runoff will decrease by 15% in the course of the next century.
That changes in magnitude and frequency of heavy floods accompanied changes
in the (European) climate has already been observed earlier (Lamb, 1977): Many
parts of Europe experienced anomalous wetness and great floods from about AD
1150 onward till about AD 1500 (Medieval warm period). However, southern
Europe experienced very rare flooding in that period (which agrees with current
GCM projections, showing that regional responses to climate change may differ).
4.1.4. Analyses of recent floods
Germany, France, Belgium and the Netherlands, 21-31 December, 1993
In northwestern Europe, the worst flooding in 60 to 100 years took place just
before Christmas 1993. In Germany, the flooding was most severe through the
Rhineland and neighbouring regions. River water poured into many riverside
towns. Cologne experienced its the worst flood in 100 years. In France, the
flooding burst the banks of the Oise, Meuse and Moselle. Hundreds of people had
to be evacuated. In Belgium, the river Meuse cut off the city of Dinant. It was the
highest flood since 1926. The Dutch government declared a state of emergency
along the basin of the Meuse. In total, 170 sq km of Dutch territory was
inundated. The water level at Maastricht was 10 cm higher than measured. The
1993 peak runoff volume in Maastricht (3120 m3/s) was higher than it was in
1926 (3000 m3/s). It was thought that a flooding of this magnitude occurred
once every 155 years. The total estimated damage was 265 million Dutch
guilders (Rijkswaterstaat, 1994).
Excessive rainfall over a large area was the primary cause of these floods. December 1993 was a very wet month in large parts of west and central Europe. Active western circulation forced repetition of active rainstorms in central Europe. Along the French-Belgium border, 100 mm fell in two days. Fortunately, there was no snow in the Ardennes. Otherwise, water levels in the Meuse would have risen considerably higher (several decimeters).
There is a clear relation between daily rainfall and runoff. On 19 December, over 30 mm had already fallen in the French part of the Meuse catchment. The precipitation on 20 December originated from a disturbance passing somewhat more north compared to the rainfall of 19 December. Hence, the rain fell in an area that had already suffered from higher levels due to the upstream rainfall. This coincidence might be the reason that water levels were higher than expected.
A statistical survey of rainfall data for three stations in the Ardennes on the prevalence of two- to four-day rainfall events revealed that such events take place once every 10 to 20 years (van Meijgaard, 1994). However, the recurrence period will be larger over the whole catchment area of the Meuse. Indeed, the total precipitation in december 1993 in Belgium was characterised by the Belgium Meteorological Organisation (KMI) as very exceptional, which means that the event has an occurrence of the order of one in the 100 years (van Meijgaard, 1994). The precipitation amount was 2.6-3.6 times the average precipitation.
Germany, France, Belgium and the Netherlands, 24 January-4 February, 1995
January was wet: the precipitation amount in this month was 1.5 to 3 times the
average monthly total. The flooding of the rivers Rhine and Meuse was of
comparable magnitude with the 1993 flooding. Again, it was called "the flood of
the century". Over a period of 9 days the water level rose above the critical level
(when inundation begins). In 1993, this period was 5 days. Although information
is not yet fully consistent, it is generally assumed that compared to the 1993
flood, the rainfall was less intense but of a longer duration, and coincided with
frozen soils (which enhances surface runoff) and later snow melting in the
Ardennes and the Alpes. The peak runoff of the Meuse at Maastricht was around
3100 m3/s. The peak runoff of the Rhine was six times as high as the average
runoff (2200 m3/s). The precipitation in the period 22-30 January in the
catchment areas of the Rhine and the Meuse is characterised by the Royal
Netherlands Meteorological Organisation (KNMI, 1995) as exceptional: in this
century it only occurred two times earlier (1926 and 1993).
Alps: France, Italy, Switzerland, 23 September-8 October 1993
After three days of heavy rain, rivers overflowed and mudslides occurred.
Flooding in the Savoie region was the worst in 35 years. In Italy, towns were
flooded all along the Ligurian coast and up into Piedmont towards the Alps. In
the port of Genoa it rained non-stop for 22 hours: 525 mm in 4 days compared
to an annual average of 250 mm. Obviously, the flooding was caused by the
abundant rainfall.
