Presentation of Findings Using Geostatistics

4.1.5 Presentation of Findings Using Geostatistics

The above variables characterizing impoundments can be used to produce maps for flood risk management purposes. This section briefly discusses some example maps for the wider central Scotland area. Two geostatistical techniques have been used for the display of the variables. Ordinary kriging provides the best linear unbiased estimations with minimum error variance and is the most commonly used type of kriging. In comparison, disjunctive kriging is a non-linear generaliza- tion of kriging. This estimation technique allows for the conditional probability that the value of a spatially variable SFRB characterization parameter is greater than a cutoff level to be calculated. A detailed discussion on spatial statistics is, however, not within the scope of this book.

Lat 56° 44' N

Capital Town County border

Engineered 20 – 40 0 – 20 60 – 100 40 – 60

North Sea Kilometers 0 10 20

Figure 4.2 Map showing the application of ordi- nary kriging for the variable engineered (%)

Long 2° 9' E

Lat 55° 22' N

Lat 4° 51' E

4.1 Rapid Assessment Methodology for the Survey of Water Bodies 175

Figure 4.3 Map showing

Lat 56° 44' N

the application of ordi- nary kriging for the vari-

able mean flooding depth Capital

County border Town

(meters)

Mean flooding depth 3– 5 0– 3 5– 9

North Sea Kilometers 0 10 20

Long 2° 9' E

Lat 55° 22' N

Lat 4° 51' E

Figures 4.2 to 4.4 show map examples applying ordinary kriging for the vari- ables engineered, mean flooding depth, and maximum flood water volume, respec- tively. High numerical values for the variable engineered generally indicate the likely necessity of high civil engineering investment to be made when planning for the construction of a new SFRB (Figure 4.2). The most engineered SFRB struc- tures are likely to be found in the southwest of the study area, which coincides with the highest density of reservoirs and lakes used for water supply purposes. In contrast, low investment for flood infrastructure is required for the study area in the North. This variable is particularly useful when a decision has to be made on where an old flood infrastructure should be upgraded or a new SFRB should be constructed.

The spatial distribution for the variable mean flooding depth is shown in Fig- ure 4.3. The mean flooding depth is relatively high in the less populated upland areas of the Northwest and South of the study area as well as within the Pentland

Lat 56° 44' N

Capital County border Town Prediction

North Sea Kilometers 0 10 20

Figure 4.4 Map showing the application of disjunc- tive kriging for the vari- able engineered (%; exceeding 30%)

Long 2° 9' E

Lat 55° 22’ N

Lat 4° 51' E

176 4 Wetlands and Sustainable Drainage

Figure 4.5 Map example

Lat 56° 44' N

showing the application of disjunctive kriging for

Capital

the variable engineered

County border Town

Probability 0.0 – 0.4 0.6 – 0.8 0.4 – 0.6 0.8 – 1.0

North Sea Kilometers 0 10 20

Long 2° 9' E

Lat 55° 22' N

Lat 4° 51' E

Hills area southwest of the capital, Edinburgh. Low values for mean flooding depth are rare and patchy.

Figure 4.4 shows the most likely values for the variable managed maximum flood water volume. This volume-based variable mirrors the depth-based variable, indicating that higher depths correlate with higher volumes, which is particularly the case for upland areas, far away from major urban settlements.

Map examples showing the application of disjunctive kriging for the variables engineered, mean flooding depth, and maximum flood water volume are summa- rized in Figures 4.5 to 4.7, respectively. Areas of low and high probabilities for the variable engineered are relatively small and patchy (Figure 4.5). This probability map can be used in conjunction with Figure 4.2 and all maps indicating flooding depth and flood water volume to determine the areas of greatest investment poten- tial, if flooding is likely to be a problem.

Lat 56° 44' N

Capital Town County border

Probability 0.0 – 0.6

North Sea Kilometers 0 10 20

Figure 4.6 Map example showing the application of disjunctive kriging for the variable mean flooding depth (meters, exceed- ing 3 m)

Long 2° 9' E

Lat 55° 22' N

Lat 4° 51' E

4.2 Classification of Sustainable Flood Retention Basin Types 177

Figure 4.7 Map example

showing the application of ordinary kriging for the variable maximum flood

Lat 56° 44' N

Capital Town

water volume (m 3 , ex-

County border

ceeding 35 × 10 –4 m 3

Probability

North Sea Kilometers 0 10 20

Long 2° 9' E

Lat 55° 22' N

Lat 4° 51' E

The map showing probabilities of exceeding 3 m flooding depth associated with the variables mean flooding depth should be used to estimate the likely return on flood infrastructure investment throughout the study area (Figure 4.6). The higher the probability, the more likely it is that an existing or planned SFRB will make

a positive impact on flood control. The greatest potential for active flood control is in areas southwest of the capital such as the Pentland Hills. Figure 4.7 shows that the areas with the highest flood storage capacity are located in upland catchments far away from populated lowland areas.