Results - Response of Flood Events to Land Use and Climate Change - page 193

The results of statistical methods for frequency analysis of the Ubaye River in
Barcelonnette show that the 1957 flood event may, on average, be expected
annually every 11 years with a 90 % lower confidence level and every 66 and
100 years with a 90 % upper confidence level.

6.11.2 Gumbel Extreme Value Distribution

The objective of these distributions was to build the relation between the proba-
bility of the occurrence of a certain event, return period, and its magnitude. To this
purpose the Gumbel Method was applied in this research. Table 6.18 shows the
calculated discharge, return period analysis and probability analysis with 90 %
confidence in lower and upper levels.
Figure 6.65 represents the hydrograph for different return periods based on the
Gumbel method.
Figure 6.66 shows a simulated flood hazard for a different return period of
66 years (blue) with 0.015 probability of occurrence, and of 100 years (red) with
0.01 probability of occurrence as well. Orange represents a 100-year return period
when implementing a dyke break. In both return periods of 66 and 100 years there
is strong overtopping on the right and left banks, causing higher risk to the banks.
The results of this overtopping are the banks getting wet and a discontinuous flow
in the floodplain as inundations. By assuming a dyke break, the inundation is
clearly obvious, and it posed a higher risk to nearby elements (e.g., houses and
farm lands). Flood hazard maps should cover the geographical areas which could
be flooded according to different scenarios:

• Floods with a low probability or extreme event scenarios
• Flood with a medium probability (likely return period of 66 years)
• Floods with a high probability

One of the aims of this section was to investigate how flood frequency distribu-
tions along the river reaches are influenced by the adoption of dyke breaks. The
simplest method for the flood frequency analysis in terms of flood hazard assessment
was transferring of hydrograph of different return periods at Barcelonnette gauge.
In some cases, it is applied to the situation with and without dykes, thus giving a
rough idea of the dykes' flood defense effects. In flat lowland areas located at the
intersection of DEM and water levels at the gage, unrealistic inundation extends

Table 6.18 Return period and its probability based on flood 1957 (maximum discharge of

480 m 3 /s)

Maximum discharge Return period Probability

90 % confidence (lower) 210 11 0.09

90 % confidence (upper) 507 66 0.015

Over topped 600 100 0.01