Scottish Road Network Climate Change Study: UKCP09 update Autumn 2011
6 Precipitation related variables
Rain is one of the most important factors affecting the design and operation of the road network. It affects the design of drainage systems to collect and discharge surface water, these systems being designed to accommodate the 1 in 1-year storm without system storage being required and to store the 1 in 5-year storm without flooding of the road surface. It also affects the sizing of river bridges/culverts, which are designed to accommodate a much larger rainfall event, typically the 1 in 100 year storm for the catchment concerned. Rain also creates a hazard to road users when it is not shed sufficiently quickly from the carriageway, resulting in loss of visibility and skid resistance, both of which are frequent contributing factors in road accidents.
Rain also has the potential to cause significant landslide events, for example those witnessed in August 2004. These occur through large volumes of surface water eroding the land surface and/or through changes in groundwater levels reducing the stability of cuttings. In addition, rain, together with temperature, can significantly alter the soil moisture condition within a catchment, creating a situation where the volume of water that the catchment sheds may be much higher than the 15% to 50% currently used in the design of drainage systems. It can also lead to river flooding impacting on culvert and bridge performance. The rainfall events currently used in road design are based on historical records of rainfall events and therefore a particular concern is that if rain is considered likely to increase, these records may no longer correctly describe the design storm event.
Snow is also a significant factor affecting the operation of the road network. Measures such as gritting are usually implemented to try to prevent ice forming or snow deposits remaining on the road surface. Snow clearing is required where heavier falls occur. Snow and ice also create a hazard to road users and are contributing factors in some road accidents. In addition, snowmelt has the potential to increase catchment runoff by releasing volumes of surface water previously held in a frozen state.
6.1 Annual precipitation
Table 6.1 provides the projected changes by the 2080s to annual mean precipitation for the three target locations. Figures 6.1 to 6.3 provide Scotland-wide changes for the 2080s medium emission scenario for each of the 10, 50 and 90 percentile probabilities
| Emission scenario | Glasgow | Aviemore | Dundee |
|---|---|---|---|
| High 2080s |
+0.6 [- 11.4 to +14.6] |
-1.2 [ -6.4 to +3.8] |
+2.7 [-4.1 to +10.6] |
| Medium 2080s |
-0.4 [-9.0 to +9.3] |
-0.9 [-5.6 to +3.8] |
+1.4 [-4.9 to +8.5] |
| Low 2080s |
+0.5 [-7.6 to +9.5] |
+0.6 [-3.5 to +5.0] |
+2.6 [-2.2 to +8.0] |
Figure 6.1 - Change in annual mean precipitation 2080s medium emission scenario (10th percentile).
Figure 6.2 - Change in annual mean precipitation 2080s medium emission scenario (50th percentile).
Figure 6.3 - Change in annual mean precipitation 2080s medium emission scenario (90th percentile).
6.2 Seasonal precipitation
Figures 6.4 to 6.6 provide projected changes in seasonal precipitation across Scotland for the 2080s medium emission scenario for each of the 10, 50 and 90 percentile probabilities.
Figure 6.4 - Percentage change in winter and summer mean precipitation 2080s medium emission scenario (10th percentile).
Figure 6.5 - Percentage change in winter and summer mean precipitation 2080s medium emission scenario (50th percentile).
Figure 6.6 - Percentage change in winter and summer mean precipitation 2080s medium emission scenario (90th percentile).
Given the relatively uniform projected percentage increases in winter rainfall across Scotland it is useful to recognise that due to the steep east-west precipitation gradient this will result in greater actual depths of rainfall in the west compared to the east. For example, a 10% increase in winter rainfall at Fort William (800mm Dec - Feb) is equivalent to an additional 80mm, whilst at Dundee (200mm Dec - Feb) is only equivalent to an additional 20mm.
6.3 Extreme storm rainfall depths
Table 6.2 provides the 10-year 1-day rainfall depths for the 2080s medium emissions scenario and the baseline period, together with the current observed estimate. The "observed" values are taken from the Flood Estimation Handbook depth-duration-frequency rainfall statistics (IH, 1999). The simulated baseline values from the Weather Generator agree well with the observed estimates giving confidence in the Weather Generators ability to predict this variable. Table 6.3 shows the same as for Table 6.2 but for the 2020s time horizon.
| Site | Observed | Baseline 1961-1990 | 2080s Medium | ||||
|---|---|---|---|---|---|---|---|
| 10% | 50% | 90% | 10% | 50% | 90% | ||
| Glasgow | 47.4 | 40.9 | 45.9 | 53.7 | 50.4 | 59.8 | 72.0 |
| Aviemore | 48.9 | 42.9 | 48.6 | 56.7 | 48.3 | 58.2 | 72.7 |
| Dundee | 48.9 | 41.8 | 47.0 | 53.8 | 50.9 | 61.3 | 77.9 |
| Site | Observed | Baseline 1961-1990 | 2020s Medium | ||||
|---|---|---|---|---|---|---|---|
| 10% | 50% | 90% | 10% | 50% | 90% | ||
| Glasgow | 47.4 | 42.0 | 47.6 | 52.7 | 46.7 | 54.1 | 61.8 |
| Aviemore | 48.9 | 44.1 | 49.9 | 56.7 | 47.5 | 53.1 | 67.7 |
| Dundee | 48.9 | 41.1 | 47.4 | 54.8 | 44.2 | 54.1 | 64.0 |
Tables 6.4 to 6.7 present the percentage changes likely to the 10-year and 2-year daily rainfall depths for different scenarios. Some regional difference is predicted with similar changes suggested for both Glasgow and Dundee whilst Aviemore is projected to be less affected. The range of the uncertainty is particularly large for this variable and the implications of this need to be recognised when adaptation measures are being considered.
| Emission scenario | Glasgow | Aviemore | Dundee |
|---|---|---|---|
| High 2080s |
+36% [+4% to +72%] |
+22% [-1% to +75%] |
+39% [+7% to +72%] |
| Medium 2080s |
+29% [+2% to +61%] |
+20% [-8% to +55%] |
+31% [+1% to +70%] |
| Low 2080s |
+16% [-3% to +47%] |
+13% [-8% to +50%] |
+17% [-5% to +55%] |
| Emission scenario | Glasgow | Aviemore | Dundee |
|---|---|---|---|
| High 2080s |
+28% [+12% to +49%] |
+16% [+1% to +30%] |
+26% [+11% to +44%] |
| Medium 2080s |
+21% [+9% to +38%] |
+12% [-2% to +23%] |
+19% [+4% to +11%] |
| Low 2080s |
+11% [-3% to +22%] |
+10% [-7% to +24%] |
+15% [+0% to +32%] |
| Emission scenario | Glasgow | Aviemore | Dundee |
|---|---|---|---|
| High 2020s |
+15% [-10% to +47%] |
+12% [-6% to +44%] |
+19% [-9% to +45%] |
| Medium 2020s |
+14% [-5% to +38%] |
+9% [-7% to +39%] |
+16% [-18% to +44%] |
| Low 2020s |
+10% [-10% to +47%] |
+11% [-11% to +50%] |
+13% [-12% to +50%] |
| Emission scenario | Glasgow | Aviemore | Dundee |
|---|---|---|---|
| High 2020s |
+10% [-2% to +13%] |
+8% [-6% to +26%] |
+7% [-5% to +26%] |
| Medium 2020s |
+8% [-4% to +21%] |
+4% [-5% to +20%] |
+5% [-7% to +21%] |
| Low 2020s |
+9% [-2% to +23%] |
+3% [-10% to +21%] |
+5% [-5% to +20%] |
Frequency of design rainfall
Further analysis was undertaken to estimate the change in the frequency of the 2-year and the 10-year one day rainfall. The scenario used for this exercise was the 2080s medium emission scenario. The results are presented in Figures 6.7 to 6.9. The return period on the x axes are drawn in logarithmic scale. For all three locations the baseline (1961-1990) 10-year one day design rainfall magnitude is likely to be as frequent as the 5-year event in the 2080s. The baseline 2-year one day rainfall event is also likely to be more frequent, becoming approximately a 1.3 year event.
Figure 6.7 - Glasgow: reappraisal of the frequency of the baseline 10-year and 2-year 1-day rainfall depths by the 2080s (based on the medium emission scenario).
Figure 6.8 - Dundee: reappraisal of the frequency of the baseline 10-year and 2-year 1-day rainfall depths by the 2080s (based on the medium emission scenario).
Figure 6.9 - Aviemore: reappraisal of the frequency of the baseline 10-year and 2-year 1-day rainfall depths by the 2080s (based on the medium emission scenario).
Comparison to estimates derived from UKCIP02 projections:
UKCIP02:
The earlier "Scottish Road Network Climate Change Study" based on the UKCIP02 scenarios suggested that extreme storm short and long return periods are likely to increase by between 10% to 30% by the 2080s (by 4% to 13% for the 2020s) compared to the 1961-1990 baseline period. These changes were assumed to be suitable surrogates for adjusting short duration low return period events (typically used in surface water drainage situations); as well as being suitable surrogates for adjusting longer and rarer events suitable for design circumstances of natural watercourses. Based on this assumption the following was predicted:
i) By the 2080s a 1-year storm of 15 to 60 minutes would become on average a 2 +/- 0.6-year storm when assessed on the present day rainfall depth-duration-frequency relationships. (Similarly by the 2020s the storm would become a 1.4 +/- 0.2-year storm).
ii) By the 2080s peak river flood flows of 2-year to 100-year were predicted to increase by 10% to 30% compared to the current situation. This suggests that the current 100-year river flood would be twice as likely by the end of the century (ie the 50-year event).
UKCP09:
Consideration of the central projected values (50 percentile values) for Glasgow, Dundee and Aviemore suggests that the change in extreme 10-year storm depths for the 2080s needs to be further increased by 15% to 40%, and by 10% to 20% for the 2020s. However the associated ranges of uncertainty need to be recognised from a risk perspective when adaptation measures are being considered. For example increases by the 2080s could conceivably be almost zero or as high as 75%. This large spread in the uncertainty reflects the particularly difficult nature of predicting extreme daily values from the climate change projections.
Assuming that the changes to the 10-year daily rainfall event can be used as surrogates for adjusting rarer events suitable for considering design circumstances of natural watercourses, the following prediction is made:
- By the 2080s peak river flood flows of 10-year to 100-year are predicted to increase by 15% to 40% compared to the current situation. This suggests that the current 100-year river flood would become the 40-year event by the end of the century.
Assuming that the changes to the 2-year daily rainfall event can be used as surrogates for adjusting short duration low return period events (typically used in surface water drainage situations), the following observation is made:
- Analysis of the UKCP09 data does not change the earlier finding based on the UKCIP02 data.
General comment:
The apparent greater increase for rarer events should be treated with some caution as, although it may be a true reflection of the likely change, it may also be a manifestation of the difficulty of projecting extreme events.
Appreciable uncertainty may surround the assumption that the changes to the 2-year daily rainfall event can be used as surrogates for adjusting short duration low return period events (ie those suitable for road drainage designs). As indicated in Chapter 2 the climate change models cannot resolve small convective storms that typically cause summer downpours.
A slight east-west design rainfall gradient does exist across Scotland with the 1-hour 2-year rainfall depth being approximately 40% higher in the western upland regions compared to the central Scotland and eastern regions. Consequently this suggests that road drainage in the western uplands will need to accommodate slightly higher design flow increases in absolute terms than elsewhere. However existing pavement drainage is designed to accommodate regional variations in design flows reflecting the existing differences which exist in design rainfall.
6.4 Change in precipitation of the wettest day in a season
Figures 6.10 and 6.11 present the 50 percentile estimates of the changes in the precipitation depth of the wettest day in the autumn (September to November) season for the high and medium emissions scenarios respectively for the 2080s. The autumn season is presented as this, on average, tends to be the period that experiences the most extreme events across Scotland (ref FEH volume 2). The wettest seasonal day cannot be attributed an exact annual return period but can be considered to be indicative of the likely changes to daily rainfall return periods approaching the 1-year event.
Figure 6.10 - Percentage change in precipitation on the wettest day in autumn by the 2080s for the high emission scenario (50 percentile central estimates given).
Figure 6.11 - Percentage change in precipitation on the wettest day in autumn by the 2080s for the medium emission scenario (50 percentile central estimates given).
These maps suggest slightly lower percentage changes in storm rainfall than those obtained from the 10-year daily rainfall analysis. The maps (together with the low emission map not presented here) suggest increases ranging from 0% to 35% may be expected with the central mountainous area having slightly lower values compared to the more coastal areas and southern Scotland. This appears to agree with the magnitude of change predicted from the analysis of the Weather Generator output for the 2-year event for the sites given in Table 6.5.
6.5 Soil moisture deficit and groundwater recharge
Only the 2080s medium emission scenario has been run for the Glasgow area and compared to the baseline period. Figure 6.12 presents the results graphically.
Figure 6.12 - Glasgow baseline and projected 2080s soil moisture deficit (medium emissions).
The maximum magnitude of the soil moisture deficit does increase in the future, but not as much as might be expected based on the projected summer decreases in precipitation and summer increases in potential evaporation. This is because plant transpiration is constrained during particularly dry periods when roots cannot access soil water and as a result, deficits cannot increase at the same rate.
The average duration that a deficit is predicted to occur is projected to barely change in the future. This refutes, in this case, the speculation that deficits will in the future last significantly longer into the autumn and early winter period; and hence mitigate the likely increase to flood flows as a result of increased autumn and winter rainfall.
The similar average durations of soil moisture deficit also suggest that on average significant groundwater recharge will occur during the same months as it does now. However with the winters projected to be wetter in the future, it is anticipated that a higher amount of recharge on average will occur in the future leading to higher winter and spring groundwater levels.
Note:
i) The averaging used to produce these projections masks significant inter-annual variability. In the future as for now, there will continue to be large inter-annual soil moisture temporal variations.
ii) Since the climate change models do not resolve localised convective storms, the future changes to this type of event (that has been linked to landslide in Scotland) cannot yet be investigated.