Presentation by CEH scientists Steve Cole, Bob Moore and Seonaid Anderson at the RMetS National Meeting on December 19th 2018.

1. Dr Steven Cole, Robert Moore,
Dr Seonaid Anderson
Progress in Weather Forecasting, RMetS National Meeting
ECMWF, Reading, 19 December 2018
Progress in flood forecasting
across Britain from advances in
hydrological modelling and
numerical weather prediction

2. 1 km (2011)
1 km
(2011)
4 km
(2008)
12 km (2005)4 km (2008)
12 km
(2005)
Caldew (246km2)
Time (hours)
Flowm3s-1
High-resolution NWP during
orographically enhanced rainfall
events provides benefits for
flood forecasting and warning
Raingauge
Flood
Warning Level
Roberts, Cole et al.,
2009, Met. Apps
Carlisle 2005 floods – high-res NWP
Observation

3. Distributed Model (G2G)
Recent development
Lumped Model
Traditional approach
• One model for each gauging station
• Many calibrated parameters
• Flow estimates for one location only
• Uses catchment average rainfall
Gauging
station
• One model for large regions (UK)
• Few parameters, use spatial datasets
• Flow estimates in each grid (1km2)
• Uses gridded rainfall estimates
Distributed hydrological modelling
Cole and Moore (2009), JoH
Potential for ungauged sites!

4. • Uses spatial datasets on terrain, soil/geology, land-cover
• Responds to spatial variation of rainfall input
• Used operationally across Britain at a 1km, 15 min resolution
Grid-to-Grid (G2G) Distributed Model
Soil
moisture
River
flow
Moore et al., IAHS Publ. 305 (2006) Price et al.; Cranston & Tavendale, Water Management (2012)

5. Catchment-wide storm:
similar response
Lower catchment storm:
earlier, larger peak
Upper catchment storm:
later, larger peak
355000 360000 365000 370000 375000
415000420000425000430000435000
Storm rainfall total
355000 360000 365000 370000 375000
415000420000425000430000435000
355000 360000 365000 370000 375000
415000420000425000430000435000
mmhr-1mmFlowm3s-1
Time (days)
Flowm3s-1
Catchment Hyetograph
Lumped Model
Distributed Model
Catchment-wide storm
Lower catchment storm
Upper catchment storm
Impact of spatial extent and location of
storm on flood response?
Extreme Flood Response
Moore et al. (2006),
IAHS Pub. 305
Same catchment
rainfall total
Same catchment
rainfall total
Varied hydrological response
important for ensembles
355000 360000 365000 370000 375000
415000420000425000430000435000
355000 360000 365000 370000 375000
415000420000425000430000435000

6. 4 km NWP1 km NWP
Boscastle 2004
12 km NWP UK radar
20 km radius
from BoscastleCourtesy Nigel Roberts, JCMM (Met Office)
• Up to 200m fell in 4 hours from sequence of convective cells
• 1 or 4km NWP major improvement over 12km product
• Still uncertainty in NWP rainfall intensities and location
Forecasts from 03 UTC

7. Boscastle 2014 – Ensemble forecasting
1km NWP
pseudo-ensemble
G2G Model 1km
river flow ensemble
Comparison with
river flow observations
Acknowledgements:
Collaboration with JCMM (Met Office)
• Simple pseudo-ensemble method to capture NWP uncertainties
• Generate gridded ensemble flood forecasts using G2G
• Operational system now uses MOGREPS-UK 2.2km ensemble

8. Risk Maps of Flood Exceedance
Acknowledgements:
Collaboration with JCMM (Met Office)
920 km2
This example employs:
• 1km rainfall pseudo-ensemble
• 10 year return period flow
thresholds
• 24 hour forecast horizon
Potential to forecast
flood risk hotspots
Probability of exceeding a given flow threshold,
for a given forecast horizon

9. • Rapid Response Catchments are typically small & ungauged
• Challenge to develop forecast/warning capability
• Use rainfall forecast ensembles (~2km, 24h, 12 members)
• Case study experience (6-7 July 2012)
Forecast time origin
Q(T) return period
threshold used
Circles denote gauging stations
Solid outline: area <50km2
Observed flow exceeds threshold
during forecast
Percentage of ensembles that exceeded the
Q(T) threshold at some point during forecast
0 0-33% 34-66% >67%
Rapid Response Catchments

10. Threshold
Forecast
Origin
Qmed/2 Qmed Q(5) Q(50)
Case study: 6-7 July 2012
06-07-2012
07:15

11. Threshold
Forecast
Origin
Qmed/2 Qmed Q(5) Q(50)
Case study: 6-7 July 2012
06-07-2012
07:15
06-07-2012
19:15

12. Threshold
Forecast
Origin
Qmed/2 Qmed Q(5) Q(50)
Case study: 6-7 July 2012
06-07-2012
07:15
06-07-2012
19:15
07-07-2012
00:15

13. Threshold
Forecast
Origin
Qmed/2 Qmed Q(5) Q(50)
Case study: 6-7 July 2012
06-07-2012
07:15
06-07-2012
19:15
07-07-2012
00:15
07-07-2012
07:15
Early signal of
flood risk
“hotspots”

14. G2G for Scotland – Snowmelt modelling
Model
Performance
(R2)
No Snowmelt

15. G2G for Scotland – Snowmelt modelling
No Snowmelt
With Snowmelt
Model
Performance
(R2)
Major improvement in
model performance

16. Real-world example from SEPA

17. National-scale Surface Water Flooding tool
● Challenge to provide real-time national SWF guidance
 Dominance of convective rainfall events that are hard to predict
 SWF Hazard Impact Model developed by Natural Hazards Partnership
 Real-time trial at FFC ongoing, planned to be operational in 2019
SWF
HIM
Grid-to-Grid
Hydrology
SWF
Hazard
Footprint
Impact
Library
Updated Flood Map
for Surface Water
National Population
Database (NPD)
National Receptor
Database (NRD)
Real-time
SWF risk
outputs
Visual Weather
Dissemination
Flood Guidance Statement
MOGREPS-UK
Ensembles

18. Dr Steven Cole: scole@ceh.ac.uk
Robert Moore: rm@ceh.ac.uk
Dr Seonaid Anderson: seodey@ceh.ac.uk