Discussion of Changes in Flood Probability

Discussions of the flood probability results are done for the comparison of the simulation results using observed data series of the past (1971 to 2000), applying REMO control scenario data series of the past (1971 to 2000) and using future climate scenario data series (2040 – 2070).

6.1.2.1 Reference Scenario 0 Results [1971 – 2000]

The results of the statistical evaluations of the flood peak probability curves computed with observed data series (Scenario 0) in this thesis are discussed with the results of a previously worked out project by the cooperation of the EPK₂ Engineers GbR and the Engineers of Klütz & Collegen Itzehoe GmbH. The team of engineers has been named: Planungsgemeinschaft Überschwemmungsgebiete an der Krückau;

in short: PÜK. The PÜK developed the hydrological model used in this thesis and worked out statistics for specific nodes (river stations) upstream of Elmshorn till the node A23 with yearly flood peak data series from 1969 to 2004. For the climate change scenario studies in this thesis, it was necessary to apply a different statistical evaluation approach with partial series for the climate period from 1971 to 2000 and with a differentiation of summer as well as winter periods of a hydrological year.

For the comparability, the flood peaks in the years 2000 till 2004 computed by the PÜK had to be transferred to the scenario 0 flood peak series worked out in this thesis. The comparison of the statistical evaluations has been done for the node A23¹. The results of the Gumbel and Gamma distribution curves computed by the PÜK are illustrated in Fig. 6. 1 together with the Log-Normal Type III and Log-Pearson Type III Distribution curves calculated in this thesis.

0 5 10 15 20 25

0 50 100 150 200 250

Results of the PÜK

Results of thesis: Scenario 0

Flood Peak [m³/s]

Return Period T [a]

PÜK: Gamma Probability Distibution Curve PÜK: Gumbel Probability Distibution Curve Scenario 0: Log-Pearson III

Probability Distibution Curve

Scenario 0: Log-Normal III Probability Distibution Curve

Empirical Distribution

Fig. 6. 1 Comparison of the statistical evaluations of the PÜK and the computed Scenario 0 in this thesis

The differences between the statistical evaluations are about 7.5% for flood peaks with small return periods of once every 1 to 5years and a higher difference of up to 17% has been calculated for events with return periods of once in 100years. Because

1 The location of this node is displayed in Figure 5.7

different probability distribution functions and partial instead of yearly series have been applied, the outcomes compared with the results of the PÜK are regarded as comparable and reasonable.

6.1.2.2 Climate Control Scenario Results [1971 – 2000]

The differences in the flood probability simulations with observed data series (Scenario 0) and the REMO control scenario data series (Scenario C20) are discussed with the results of the study done for the catchment area of the Wandse in 2009. In this project the software Kalypso Hydrology has been applied for simulating a yearly series of flood peaks for three nodes in the catchment (Golder Associates, 2009). The computed average differences of the results range between 8% for events with small return periods of once in a year and increases up to 100% for events with return periods of once in 100years. This is comparable with the results in the scenario studies in this thesis for summer periods, where rather low differences for events with small return periods (T<5a) have been calculated of about 11% and the difference increases up to an average difference of 80% for a return period of the flood peaks of once in 100years (Attachment 14.3). For the yearly period differences are computed between 16% (T=1a) and 30% (T=100a) in this thesis. In both project studies, the differences between the flood peaks are significant for extreme flood events with lower probabilities of occurrence. The reason for these differences needs to be further discussed as stated in the outlook of this thesis.

6.1.2.3 Future Climate Scenario Results [2040 - 2070]

For the discussion of the computed changing rate of flood probability curves derived in future climate scenarios, the results of the project INKLIM 2012 II plus (Brahmer, 2008) and the results of the local study of the Wandse catchment are referred.

In the INKLIM 2012 II plus project the increase of flood peaks in the Lahn catchment area derived with the IPCC scenarios A1B, B1 and A2 has been analysed.

The applied climate data series have been computed with the statistical RCM WETTREG¹. For the simulation of stream flow for the past (1961 to 1990) and the future (2051 to 2080) climate periods, the hydrological model LARSIM² has been applied (Brahmer, 2008). Like in the scenario studies of this thesis, the largest changes have been computed for the Scenario A1B with an increase of about 20%

for extreme flood peaks with return periods of once in 100years (Brahmer, 2008). In this thesis, an increase of 27% has been computed for summer flood peaks in the scenario A1B in the Krückau catchment. However, it has been pointed out that there is a low confidence in computing changes in flood probability, which can vary significantly between scenarios and study locations (Brahmer, 2008).

1 Features of the RCM WETTREG are listed in Attachment 1.2.

2 LARSIM = http://larsim.sourceforge.net/ .

In the Wandse project, REMO data series for the IPCC scenario A1B have been applied for different climate periods from 2001 to 2100, from 2051 to 2100, from 2076 to 2100 and yearly series of flood peaks have been computed with the software Kalypso Hydrology. The results in this thesis approximate the statistical evaluation results of the Wandse project for the climate period from 2076 to 2100. For three nodes in the Wandse catchment, average increases of about 28% are illustrated for events with higher probabilities of occurrence of once in a year (T=1a) and for events with lower probabilities of occurrence (T=100a) an average change of about 11.9%

has been displayed (Golder Associates, 2009). In this thesis, for events with higher probabilities of occurrence (T=1a) an increase of maximal 27.4% and for events with lower probabilities of occurrence (T=100a) an increase of about 11.8% has been calculated. The results in the Wandse report for the other climate periods from 2001 to 2100 and from 2051 to 2100, display lower changes in flood peak probabilities.

In comparison with the increase of the rainfall intensities, it can be stated that in the A1B scenario the summer flood peaks increase with a higher percentage rate than the summer rainfall intensities. The yearly flood peaks with a larger probability of occurrence (T = 5a) increases less than the corresponding rainfall events, whereas for events with lower probabilities of occurrence it is the other way around. The rainfall intensities in the winter period increase only by about 5.7% to 7.8%, but the flood peaks in the winter period increases between 11.8% and 27.4%. It can be stated that the overall tendency of increase corresponds between the extreme rainfall and flood events, but the rate and magnitude of increase differs in the A1B scenario, which displayed the largest changes and is regarded as most important for the following discussion of the post-processing results.