Global Flood and Drought Prediction

1
Global Flood and Drought Prediction

• European Geosciences UnionGeneral Assembly
  2006Vienna, Austria, 2006 April 4th
• Nathalie Voisin, Dennis P. Lettenmaier
• Department of Civil and Environmental Engineering
• University of Washington
• Seattle, USA

Credit Philip Wijmans/ACT-LWF Trevo, Mozambique,
February 2000 , http//gbgm-umc.org/umcor/00/mozph
otos.stm
2
Outline

• Background and Objective
• Data and models
• Toward developing global hydrology forecast
  capability
• Approach
• Data Processing bias correction and downscaling
  of the forecasts
• Preliminary results forecast issued on Feb 4th
  2000
• Future work

3
-1-Background
4
Need for flood prediction globally?
www.dartmouth.edu/floods, Dartmouth Flood
Observatory
5
Global Floods and Droughts

• Floods
• 50-60 billion USD /year, worldwide ( United
  Nations University)
• 520 million people impacted per year worldwide
• Estimates of up to 25,000 annual deaths
• Mostly in developing countries Mozambique in
  2000 and 2001, Vietnam and others (Mekong) in
  2000.
• Droughts
• 1988 US Drought 40 billion (1988 drought NCDC
  )
• Famine in many countries 200,000 people killed
  in Ethiopia in 1973-74

Source United Nations University,
http//update.unu.edu/archive/issue32_2.htm http/
/www.unu.edu/env/govern/ElNIno/CountryReports/insi
de/ethopia/Executive20Summary/Executive20Summary
-txt.html 1988 drought NCDC http//lwf.ncdc.noa
a.gov/oa/reports/billionz.html
6
Objective

• Predict streamflow and associated hydrologic ,
  soil moisture, runoff, evaporation and snow water
  equivalent
• 1. At a global scale
• Spatial consistency
• To cover ungauged or poorly gauged basins
• 2. Time scales
• Short term for floods
• Seasonal (or longer) for drought
• 3. Freely disseminate information for
  agriculture, energy, food security ,and
  protection of life and property

7
-2-Data and models
8
Meteorological Data

• - Surface observations
• Uneven global coverage
• Various attempts to grid globally
• We use Adam et al. (2006) 1979-1999 (0.5 degrees)
  and ERA-40
• - Precipitation derived from satellite
• Various products available, mostly either passive
  microwave and/or infra-red
• Issue with climatology and consistency (
  especially important for seasonal prediction)
• - Climate Models ECMWF and NCEP
• Re-analysis products, for at least 25 years
• Ensemble forecast products
• Quasi all or all required input data for our
  hydrologic model available

9
The Hydrologic Model VIC

• - Already calibrated and validated at 2 degree
  resolution on over 26 basins worldwide(Nijjsen
  et al. 2001)
• Calibrated and validated at 0.5 degree over the
  Arctic domain
• Ongoing with UW and Princeton globally at 0.5
  degree resolution

10
Real time forecasting using VIC

• The Seasonal Westwide Forecastoperational over
  the entire western US
• seasonal forecast of streamflow
• The Surface Water Monitor
• operational over the entire western US daily
  analysis of soil moisture

11
-3-Toward developing global hydrology forecast
capability
12
Forecast System Schematic
soil moisture snowpack
streamflow, soil moisture, snow water
equivalent, runoff
local scale (1/2 degree) weather inputs
Hydrologic forecast simulation
Hydrologic model spin up
Ensemble Reforecasts NCEP Reforecasts (Hamill
2006), bias corrected and downscaled ( NCEP
GFS, ECMWF ESP)
ECMWF ERA40 (or Analysis)
Downscaling using observations (Adam et al 2006)
Later on CMORPH, MODIS, AMSR-E, others
SNOTEL Update
NOWCASTS
SEASONAL FORECASTS (drought)
Several years back
Month 0
SHORT TERM FORECASTS (flood)
Similar experimental procedure as used by Wood
et al (2005) West-wide seasonal hydrologic
forecast system
13
Spin Up

• ECMWF ERA40 reanalysis for retrospective
  forecasting
• Assume ERA40 is the truth
• Later use ERA40 analysis field, bias corrected to
  match ERA40 characteristics

14
The Meteorological Forecasts

• Retrospective forecasting Reforecasts
• Tom Hamill (2006) NOAA
• NCEP-MRF, 1998 version
• 1979-present
• 15-day forecasts issued daily
• 15 member ensemble forecast
• 2.5 degree resolution
• Real Time forecasting ECMWF and/or NCEP (future)

15
Data processing

• The climatology statistics to be conserved in the
  forecasts are
• - the frequency of occurrence of rain- the
  peaks - accumulated amounts (mean)
• Using quantile-quantile mapping technique

16
Data processing Bias Correction

• Non-exceedance probability plots (MRF in green,
  ERA40 in black )
• Systematic Bias Occurrence of
  Precipitation

17
Data processing Downscaling

• Inverse square distance interpolation from 2.5
  down to 0.5 degree resolution
• Integration of observation based spatial
  variability at 0.5 degree
• Use observations based Adam et al. (2006) global
  dataset (0.5 degree resolution)
• Shifting
• makes the Adam et al. average temperature field
  at 2.5 degree match ERA40,
• Derive the temperature range for each 0.5 degree
  cell within the 2.5 degree cell
• Scaling of the precipitation and the wind field
  so that the ratio Value(0.5)/Value(2.5) is
  conserved

18
Preliminary results

• February 2000 floods in the Northern Part of
  South Africa
• Tropical depression moving southward from Beira,
  then continuing west into Zimbabwe, Botswana and
  South Africa
• Sustained rain during the period 4 to about 14
  February

Tropical depression Boloetse track (pink) and
forecasted direction (red)
http//gisdata.usgs.net/sa_floods/
19
Preliminary results 2000 Feb 4th

• 5 day acc. PRECIPITATION
• ERA 40 GFS
  reforecast 15 ensembles avg

LEAD 1
LEAD 1
LEAD 2
LEAD 2
20
Preliminary results 2000 Feb 4th

• 5 day acc. RUNOFF
• ERA 40 GFS
  reforecast 15 ensembles avg

LEAD 1
LEAD 1
LEAD 2
LEAD 2
21
Preliminary results 2000 Feb 4th
Basin Avg Hydrologic Variables Prediction (ERA40
in red, GFS in black ) Zambeze Basin, Africa
NO BIAS CORRECTION BIAS CORRECTION
22
Preliminary results 2000 Feb 4th
Basin Avg Hydrologic Variables Prediction (ERA40
in red, GFS in black ) Limpopo Basin, Africa
NO BIAS CORRECTION BIAS CORRECTION
23
Preliminary results 2000 Feb 4th
Basin Avg Hydrologic Variables Prediction (ERA40
in red, GFS in black ) Colorado Basin, North
America NO BIAS CORRECTION BIAS
CORRECTION
24
Preliminary results 2000 Feb 4th
Basin Avg Hydrologic Variables Prediction (ERA40
in red, GFS in black ) Ganges Basin, Asia NO
BIAS CORRECTION BIAS CORRECTION
25
Preliminary results 2000 Feb 4th
Basin Avg Hydrologic Variables Prediction (ERA40
in red, GFS in black ) Elbe Basin, Europe NO
BIAS CORRECTION BIAS CORRECTION
26
Preliminary results 2000 Feb 4th

• The bias correction
• beneficial for ALL input variables (P, Tavg,Wind)
• does not substitute for missed precipitation/tempe
  rature peaks/lows BUT the correction for the
  occurrence of rain correction should help
• brings consistency between the control run (
  model or observations, or both) and the forecasts

27
-4-Future Work
28
Future Work

• Retrospective forecasting
• Finish up the small scale variability
  implementation in the code
• Refined precipitation occurrence correction
• Further evaluation of the retrospective
  forecasts
• using GFS reforecasts
• and eventually archived ECMWF 10 day, monthly and
  seasonal forecasts
• Predictions in forms of percentile and anomalies
  with respect to the climatology

29
Future Work

• Operational real time forecasting
• Once a week
• Use several climate model forecasts
• ECMWF 10 day forecast
• ECMWF monthly forecast
• ECMWF seasonal forecast
• GFS 6-10 day forecast
• Improvement of the initial conditions e.g.
  assimilation of satellite soil moisture snow

30
Thank You!
Credit Philip Wijmans/ACT-LWF Trevo, Mozambique,
February 2000 , http//gbgm-umc.org/umcor/00/mozph
otos.stm
31
Preliminary results 2000 Feb 4th

• Snapshots Hydrologic Variables Prediction (ERA
  40)
• 5 day acc. PRECIPITATION 5 day acc.
  RUNOFF

32
Preliminary results 2000 Feb 4th

• Snapshots Hydrologic Variables Prediction (ERA
  40)
• 5 day avg. SOIL MOISTURE 5 day avg.
  SWE

33
Preliminary results 2000 Feb 4th

• 5 day acc. PRECIPITATION
• ERA 40 GFS
  reforecast ensemble 14

34
Preliminary results 2000 Feb 4th

• 5 day acc. RUNOFF
• ERA 40 GFS
  reforecast ensemble 14