Scottish Government

3 Flood Characterisation

3.1 Introduction

The flood severity can be determined by assessing rainfall and river flow data. The majority of hydrometric data in Scotland is collected by Scottish Environment Protection Agency and the MET office. Only SEPA data was analysed as part of this scoping study.

3.2 Hydrological summary

The hydrological summary ¹¹ analyses the rainfall and flow data and provides a concise summary of the period. The information is presented in a format that is easily interpreted and provides a nationwide picture. The analysis does not present detail of individual catchments which may have cause the most damage in terms of flooding.

The summary shows that August was a very wet month with rainfall substantially above average and provisionally the wettest August since 1956. It also indicates that there were a number of high intensity rainfall events that set new records for gauges.

August rainfall recorded very high precipitation particularly over the east of Britain. The most exceptional rainfall with a 200 year return period (0.5% Annual Exceedance Probability ( AEP)) over a 16 day period was recorded in Boreland and flooding was recorded nearby at Sibbaldbie, Dumfries and Galloway. The summary also shows that the north east of Scotland received much lower rainfall than usual.

Notes: MET Office data as received through Scottish Executive

Figure 3-1 below shows the location of all target catchments identified as a result of assessing the areas that were worst affected by flooding in August.

River flow data and rainfall data requested as part of this scoping study are detailed in Table 3-2 and Table 3-3. The aim of such a data request was to complement the data which can be readily downloaded from the National River Flow Archive ¹² maintained by CEH Wallingford.

Notes:
1. NRFA - National River Flow Archive - Maintained by CEH Wallingford ( http://www.nwl.ac.uk/ih/nrfa/index.htm )
2. Shaded sites are those where data was received prior to study time constraints and therefore allowing analysis to be carried out

Constraints on data supply led to the analysis of information made available at the River Don at Bridge of Alford and Parkhill, the River Tweed at Norham and the Lyne Water at Lyne Station (an upstream tributary of the Tweed).

Notes:
1. Shaded sites are those where data was received prior to study time constraints and therefore allowing analysis to be carried out
2. TBR - Tipping Bucket (15 minute data), D- Daily, M- Monthly

3.3 Flow Analysis

3.3.1 Daily Mean Flow & Event Analysis

Minimum and Maximum daily mean flows have been compared with the daily mean record for August (see Figure 3-2, Figure 3-4, Figure 3-6 and Figure 3-8 overleaf). This allows August 2004 to be examined against the full record of daily mean flows.

The daily mean flows recorded at Bridge of Alford during August 2004 (65.8 m3/s) are the greatest when compared against the August daily mean flows only, suggesting that such flows with the Don at Bridge of Alford are unusual for this time of year. A similar analysis at Parkhill lower down the catchment for August 2004 (119.5 m3/s) does not show such unusual flows as experienced at Bridge of Alford. A similar pattern is exhibited in the Lyne Water and Tweed catchment, where the daily mean flows recorded at Lyne Station during August 2004 (32.2 m3/s) are the greatest considering August Daily mean flows only, suggesting that these flows were unusual for this time of year, whereas the daily mean flows for August 2004 at Norham (603.3 m3/s) while high, were not the unusual within the record.

Event analysis has been carried out as a count of the number of high flow days per year and also high flow days for the month of August for each year of the record, including August 2004. This is considered to identify periods of high flows or flooding. High flow is considered to be that above the 97% percentile (Q97). Q97 is determined by calculating flow duration curves for the full record of each of the sites, flow non exceeded 97% of the time can then be defined, as detailed in Table 3-4 below for each of the target catchments.

Q97 event analysis at Bridge of Alford for the months of August resulted in a count of 2 days in 2001 and 2 days in 2004 (a total of 4 days over the 30 year record). For Parkhill a count of 7 days in 1970, 1 day in 1979, 1 day in 2000 and 1 day in 2004 (a total of 10 days over the 34 year record). For the Don catchment at both Bridge of Alford and Parkhill gauging stations this shows the rarity of high flows (equalling or exceeding Q97) for August 2004.

Q97 event analysis at Lyne Station for the months of August result in a count of 2 days in 1985, 1 day in 1998,2 days in 2001 and 7 day in 2004 (a total of 12 days over the 35 year record). For Norham for the months of August resulted in a count of 1 day in 1963, 1 day in 1966, 1 day in 1971, 2 days in 1985, 2 days in 1986, 1 day in 2002 and 5 days in 2004 (a total of 13 days over the 42 year record). For both the Lyne Water and Tweed catchments this shows the rarity of high flows (equalling or exceeding Q97) for August 2004.

3.3.2 N Day Annual Maxima

N Day Annual Maxima have been derived from using daily mean flows, covering various durations from 1 to 60 days. The longer duration annual maxima are more likely to pick up on occurrences of repeated flooding.

30 day and 60 day Annual Maxima have been plotted in Figure 3-10 and Figure 3-11 with linear trend lines for the river Don. In the case of Bridge of Alford there is a slight downward trend in 30 day annual maxima and whereas there is no change for the 60 day annual maxima. At Parkhill there appears to be an upward trend in both the 30 day and 60 day annual maxima.

30 Day Annual Maxima

60 Day Annual Maxima

30 Day Annual Maxima

60 Day Annual Maxima

30 day and 60 day Annual Maxima have been plotted in Figure 3-12 and Figure 3-13 with linear trend lines for the Lyne Water and River Tweed. In the both these cases there is a slight upward trend in 30 day annual maxima and 60 day annual maxima.

These predominantly upward trends exhibited by the Don, Lyne Station and Tweed catchments is likely to evidence that we are experiencing high flows for longer durations, this is increasingly important when considering catchments where the volumes of flood events rather than the peak alone are critical in trigger flood events.

30 Day Annual Maxima

60 Day Annual Maxima

30 Day Annual Maxima

60 Day Annual Maxima

3.3.3 Highest Monthly Flows

The highest instantaneous flow value for August 2004 for the Don at Bridge of Alford and Parkhill, Lyne water (a tributary of the Tweed) at Lyne station and the Tweed at Norham were analysed against the highest instantaneous flow values for the record and the results shown below in Table 3-5

On the river Don firstly at Bridge of Alford a maximum of 123.3 m3/s was recorded, this equates to 578 % of the August Average (1973 to 2004) and 317% of the total record average. At Parkhill a maximum of 138.7 m3/s was recorded, this equates to 355% of the August Average (1970 to 2004) and 233% of the total record average.

On the Lyne Water a maximum of 45.28 m3/s was recorded, this equates to 443 % of the August Average (1973 to 2004) and 310% of the total record average. While at Norham on the Tweed a maximum of 801.6 m3/s was recorded, this equates to 414% of the August Average (1986 to 2004) and 234% of the total record average.

Notes:

Although Instantaneous monthly data for the Lyne Water and Tweed were only received from 1986 to present, they appear to compare well with the averages resulting on the upland and lowland Don.

3.3.4 Flood Frequency Analysis

The WINFAP- FEH software package was used to derive frequency distributions for each of the gauging stations using both the pooling group methodology and single site analysis. WINFAP- FEH contains a database of flood data for catchments from across the UK. All FEH Statistical calculations were carried out using the UKHiFLOWs dataset. Sites are selected for the pooling group on their hydrological similarity in terms of AREA, BFIHOST and SAAR. A 100 year target pooling group was established where the resulting total number of station years required in the pooling group is 5 times the target return period. Changes to the default pooling group included the inclusion of all HiFLOWs gauges within 40km of the individual gauging stations in line with recommendations in The Flood Estimation Handbook Statistical Procedures: Automation, Appraisal and Future Development¹³.

WINFAP- FEH software determines the suitability of growth curve fittings through a 'goodness-of-fit' test. The results of this test on this pooling group with a target return period of 100 years (1% Annual Exceedance Probability ( AEP)) showed that the Generalised Logistic ( GL) distribution was suitable for the Parkhill, Lyne Station and Norham gauging station and Generalised Extreme Value ( GEV) distribution was suitable for the Bridge of Alford gauging station.

Single site analysis for each of the gauging stations was carried out using General Logistic L-Moment ( GL- LMOM) distribution. These appear to fit reasonably well at each of the gauging stations for the lower return periods. It is important to note greatest confidence lies in the lower return period estimates up to 25 years (5% AEP) for Bridge of Alford, Parkhill and Norham and 10 Years (10% AEP) for Lyne Station due to the gauged record lengths.

Notes:

In addition to flood frequency analysis of the instantaneous Annual Maxima series, flood frequency analysis was also carried out using the N- Day Annual Maxima derived from the daily mean series for each of the sites for the whole series and for the august series only for 1 day, 5 day, 10 day and 30 day Annual Maxima, the results of which are presented in Figure 3-7. This enables an indication of rarity of the August 2004 event in terms of the whole data record and in terms of the month of August, in addition to examining the rarity of different duration flood events.

At each of the sites, in terms of the annual flood series the flows experienced during August 2004 were of low return period, apart from Lyne Station which experienced approximate 10 year return period (10% AEP) flow for 5 day duration.

When examining the August series, it can be seen than the data exhibits greater rarity, ranging from 28 year return period (3.57% AEP) for the instantaneous august maximum to a return period of 95 years (1.05% AEP) at Lyne station for a 5 day duration event. The 'upland' sites of Bridge of Alford and Lyne Station both experienced greater return period flows over the longer duration periods, than those downstream on the associated catchments at Parkhill and Norham, and is likely to be linked to the shorter response time of the smaller upland catchments.

Notes:
Return periods calculated using WINFAP- FEH Software, using single site analysis with GL- LOM distribution.

3.4 Rainfall Analysis

3.4.1 Daily

Data from three daily rain gauges in the Don catchment (Cushnie, Insch, and Tullinessle) and in the Tweed catchment (Hallmanor, Hawick, and Cappercleuch) were compared against rainfall shown in the FEHCDROM Depth Duration Frequency ( DDF) model in terms of varying durations from 1 day to 15 days the results of which are shown in Table 3-8. The FEHDDF model is not to be used for durations greater than 16 days. In addition as the rainfall data recorded daily at 0900 the exact duration of the rainfall is unknown if less than 24 hours, therefore hypothetical return periods have been calculated for durations of 12 hours and 6 hours.

Notes:
1. August - this is based on single site analysis using a Gumbel distribution with LMoments. It must be notes that this is based on short record lengths and is only statistically reliable for larger return periods and they are therefore used in this case to give indication of severity.
2. DDF - Calculated using the FEHCDROM and is based on pooled equivalent and based on the Annual series.

As with the daily mean flow data frequency analysis was carried out using the N-Day Annual Maxima derived from the daily rainfall series for the august series only for 1 day, 5 day, 10 day and 30 day Annual Maxima and compared to the DDF analysis, the results of which are presented in Table 3-8. This enables an indication of rarity of the August 2004 event in terms of the whole data record and in terms of the month of August, in addition to examining the rarity of different duration flood events.

At each of the sites, in terms of the annual flood series the flows experienced during August 2004 are common place, whereas those in terms of August alone are more unusual (note only calculated using single site analysis on short record periods, so high return periods less reliable).

3.5 Tay

Data was received but after analysis was completed.

3.6 Teith

Only 15 minute instantaneous data for Inveruglas rain gauge has been considered. Analysis carried out using the FEHCDROMDDF model on the event rarity demonstrates that rainfall experience at Inveruglas was not unusual. Analysis of this single rain gauge in this area is inconclusive, but may demonstrate the localised nature of rainfall experienced during August 2004, in that this gauge is located on the north western shore of Loch Lomond approximately 11km from the landslides on the A83 and 30km from the landslides in Glen Ogle.

Notes:
1. N Day Rainfall Totals calculated from Daily rainfall data on a sliding duration. Return periods calculated using 1km point FEHDDF model.