High Resolution Modelling of the Conditions over the Brue Catchment

1
High Resolution Modelling of the Conditions over
the Brue Catchment

• Philip Younger (Lancaster)
• Alan Gadian (Leeds)
• Chang Wang (UWERN)
• Keith Beven (Lancaster)

2
Aim

• To study the processes which affect the feedback
  between the precipitation (UM) and the surface
  hydrological model.

1/14
3
Introduction

• Conventional approach is to use radar and gauge
  data as inputs. These can be inadequate for long
  lead times (Beven, 2001). Our approach is to use
  numerical forecast models to provide
  precipitation input for hydrological models,
  especially useful for ungauged catchments. Such
  an approach can be made in a statistical manner
  to develop sensitivity studies.

2/14
4
Background objective

• To improve our understanding of flood forecasting
• To understand the important feedback processes
• To develop a knowledge base that can lead to new
  tools
• To make sensitivity studies to examine possible
  feedback effects.

3/14
5
Hydrology Requirements

• Catchment Characteristics
• Topography, subsurface, land use etc.
• Precipitation Input
• Good forecast ahead of time (for warning)
• Local scale (for valid input)
• Concerned with quality/uncertainty
• (for a distribution of flow levels)

4/14
6
Brue the HYREX catchment
HYREX Study period late 1990s. Good data sets.
49 raingauges. Brue 135km².
5/14
7
Modelling Technique

• UM v5.5
• Running on Newton, up to 16 processors
• ECMWF reanalysis data
• Series of nested grids
• Resolutions Global, 12km, 4km, 1km, 250m
• Run Length, 3 days for Global, 1 day for others
• Time periods
• 18th December 1999 frontal rain
• 22nd May 1994 convective rain

6/14
8
Domains

• Global
• 12km resolution
• 4km resolution
• 100x100 (gp)
• 1km resolution
• 150x150 (gp)
• 250m resolution
• 240x220 (gp)

7/14
9
Model Results
Cumulative rainfall (mm) maps for the catchment
over preceding hour.
Observed
Modelled
8/14
10
Model Results
Cumulative frequencies for rainfall amount
distribution, based on grid blocks, over the
preceding hour.
Red Observed Blue - Modelled
9/14
11
Analysis Techniques

• Distribution analysis
• Kolmogorov-Smirnov, Chi-square
• Block-by-block
• Total volumes and summary statistics
• Fuzzy approach

10/14
12
Model results
Surface Heat Flux (left) and Surface Latent Heat
Flux (right) (W/m²) output for 250m domain, 1300
18th December 1999.
11/14
13
Future work

• Uncertainty on observations to be built in
• Sensitivity studies
• Bowen Ratio
• Catchment wide
• Tiles
• Summer scenario

12/14
14
Summary

• I have
• Highlighted why this research is taking place
• Shown some initial results
• Highlighted the future direction, with emphasis
  on the uncertainties
• NB These are just preliminary results.

13/14
15
Acknowledgements

• Thanks to
• Pete Clark
• Humphrey Lean
• Richard Forbes
• Reference
• Beven (2001), Rainfall-Runoff Modelling The
  Primer, John Wiley Sons, Chichester.

14/14