Operational Flood Forecasting for Operational Flood Forecasting for Bangladesh:Bangladesh:

Tom Hopson, RAL-NCARTom Hopson, RAL-NCARPeter Webster, Georgia TechPeter Webster, Georgia Tech

A. R. Subbiah and R. Selvaraju, Asian A. R. Subbiah and R. Selvaraju, Asian Disaster Preparedness CentreDisaster Preparedness Centre

Climate Forecast Applications for Climate Forecast Applications for Bangladesh (CFAB): Bangladesh (CFAB):

USAID/CARE/ECMWF/NASA/NOAAUSAID/CARE/ECMWF/NASA/NOAA

Bangladesh StakeholdersBangladesh Stakeholders: Bangladesh Meteorological Department, Flood Forecasting and Warning Center, : Bangladesh Meteorological Department, Flood Forecasting and Warning Center, Bangladesh Water Development Board, Department of Agriculture Extension, Disaster Management Bureau, Bangladesh Water Development Board, Department of Agriculture Extension, Disaster Management Bureau, Institute of Water Modeling, Center for Environmental and Geographic Information Services, CARE-Institute of Water Modeling, Center for Environmental and Geographic Information Services, CARE-BangladeshBangladesh

Climate Forecasting Applications for Bangladesh (CFAB)The CFAB project goals:

• Develop resilient forecast schemes that capitalize on skillful modeling techniques and advanced data sources at time-scales:1-6 months, 20-25 days, 1-10 days (2000)

• Develop an infrastructure within Bangladesh to:

a) make use of the forecasts -- establish pilot projects at selected sites, showing measurable improvements (2006)

b) eventually own the prediction schemes -- facilitate a technological transfer (2008)

2004 dry season river flows …

… and during the July flooding event

NASA Aqua/Modis images

Bangladesh backgroundBangladesh backgroundAbout 1/3 of land area floods the monsoon rainy season

Size: sightly smaller than Iowa (144,000 sq km)

Border countries: Burma (193 km), India (4,053 km)

Population: 140 million

36% of population below poverty line

Within the top 5 of: poorest and most densely populated in the world

Natural disasters:

Nov 1970 Bhola cyclone -- at least 300,000 died in 20 min (12m)

April 1991 Bangladesh cyclone -- 138,000 died (6m)

(World Food Program)

Damaging Floods:large peak or extended durationAffect agriculture: early floods in May, late floods in September

Recent severe flooding: 1974, 1987, 1988, 1997, 1998, 2000, 2004, and 20071998: 60% of country inundated for 3 months, 1000 killed, 40 million homeless, 10-20% total food production2004: Brahmaputra floods killed 500 people, displaced 30 million, 40% of capitol city Dhaka under water2007: Brahmaputra floods displaced over 20 million

River Flooding

Overview:Bangladesh flood forecasting

I. CFAB History -- sea-level backwater effects

II. 1-10 day Discharge Forecasting1. precipitation forecast bias removal2. multi-model river forecasting3. accounting for all error: weather and hydrologic errors

III. 2007 Floods and Warning System Pilot Areas

How important are (“forecastable”) interannual sea level variations on country-wide extreme flooding events?

Answer: effects impact river heights roughly 200km upstream …

August 2002 floodsSeptember 1998

Upper-Catchment Flooding

disasterous flood year“normal” flood season

Severe flood years affect whole country, with water depth variations of O(1m)

=> Look at precipitation-driven effects on flooding

CFAB Project: Improve flood warning lead time

Problems:

1. Limited warning of upstream river discharges

2. Precipitation forecasting in tropics difficult

Assets:1. good data inputs: weather forecasts, satellite rainfall2. Large catchments => weather forecasting skill “integrates” over large spatial and temporal scales3. Partnership with Bangladesh’s Flood Forecasting Warning Centre (FFWC)

=> daily border river readings

Merged FFWC-CFAB Hydraulic Model Schematic

Primary forecast boundary conditions shown in gold:

Ganges at Hardinge Bridge

Brahmaputra at Bahadurabad

Transforming (Ensemble) Rainfall into Transforming (Ensemble) Rainfall into (Probabilistic) River Flow Forecasts(Probabilistic) River Flow Forecasts

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0 1 2 3 4 5 6

Rainfall Probability

Rainfall [mm]

Discharge Probability

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

10,000 30,000 50,000 70,000 90,000

Discharge [m3/s]

Above danger level probability 36%Greater than climatological seasonal risk?

Daily Automated Operational Flood Forecasting Sequence

ECMWF 51-member Ensemble Precipitation Forecasts

2004 Brahmaputra Catchment-averaged Forecasts

5 Day Lead-time Forecasts

Pmax

25th 50th 75th 100th

Pfcst

Pre

cipi

tatio

n

Quantile

Pmax

25th 50th 75th 100th

Padj

Quantile

Bias Adjustment

Model Climatology “Observed” Climatology

Bias-corrected Precipitation Forecasts

Brahmaputra Corrected Forecasts Original Forecast

Corrected Forecast

Daily Automated Operational Flood Forecasting Sequence

2003 Model Comparisons for the Ganges (4-day lead-time)

hydrologic distributed modelhydrologic lumped model

Resultant Hydrologic multi-model

Multi-Model Forecast Multi-Model Forecast Regression CoefficientsRegression Coefficients

- Lumped model (red)- Lumped model (red)- Distributed model (blue)- Distributed model (blue)

Significant catchment variationCoefficients vary with the forecast lead-timeRepresentative of the each basin’s hydrology

-- Ganges slower time-scale response

-- Brahmaputra “flashier”

Daily Automated Operational Flood Forecasting Sequence

Significance of Weather Forecast Uncertainty Discharge Forecasts

3 day 4 day

5 day

7 day 8 day

9 day 10 day

2004 Brahmaputra DischargeForecast Ensembles

Corrected Forecast Ensembles

7 day 8 day

9 day 10 day

Producing a Reliable Probabilistic Discharge Forecast

Step 1: generate discharge ensembles from precipitation forecast ensembles (Qp):

1/51

1

Qp [m3/s]

Probability

PDF

Step 3: combine both uncertainty PDF’s to generate a “new-and-improved” more complete PDF for forecasting (Qf): Qf [m3/s]

1Probability

Step 2: a) generate multi-model hindcast error time-series using precip estimates;b) conditionally sample and weight to produce empirical forecasted error PDF:1000

-1000

forecasthorizon

time

PDF 1

-1000 1000Residual [m3/s]

[m3/s]

Residuals

=>

a) b)

2 day

3 day 4 day

5 day

Confidence Intervals

7 day 8 day

9 day 10 day

50% 95%Critical Q black dash

2004 Brahmaputra Forecast Results

Above-Critical-Level Cumulative Probability

7 day 8 day

9 day 10 day

Overview:Bangladesh flood forecasting

Five Pilot Sites chosen in 2006 consultation workshops based on biophysical, social criteria:

Rajpur Union -- 16 sq km-- 16,000 pop.

Uria Union-- 23 sq km-- 14,000 pop.

Kaijuri Union-- 45 sq km-- 53,000 pop.

Gazirtek Union-- 32 sq km-- 23,000 pop.

Bhekra Union-- 11 sq km-- 9,000 pop.

A v e r a g e D a m a g e ( T k . ) p e r H o u s e h o l d i n P i l o t U n i o n

7 , 2 5 5

2 8 , 7 4 5

6 0 , 9 9 3

6 4 , 0 0 0

4 0 5 8

0

1 0 , 0 0 0

2 0 , 0 0 0

3 0 , 0 0 0

4 0 , 0 0 0

5 0 , 0 0 0

6 0 , 0 0 0

7 0 , 0 0 0

U r i a G a z i r t e k K a i j u r i R a j p u r B e k r a

U n i o n

Average Damage (Tk) per

Household

(annual income: 30,000 Tk; US$400)

2007 Brahmaputra Ensemble Forecasts and Danger Level Probabilities

7-10 day Ensemble Forecasts7-10 day Danger Levels

7 day 8 day

9 day 10 day

7 day 8 day

9 day 10 day

Response of National Institutions for 2007 flood forecasts

• Flood Forecasting and Warning Center (FFWC) incorporated the CFAB forecasts to produce water level forecasts for many locations along Brahmaputra and Ganges well in advance

• National level Disaster Emergency Response Group prepared emergency response plans, logistics for preparedness and relief in advance

Selvaraju (ADPC)

Response of local institutions for 2007 flood forecasts

• Local project partners used community vulnerability maps to assess the risk of flooding

• Local NGOs and CBOs mobilise boats to rescue people and livestock from the “char” areas

Selvaraju (ADPC)

Community level decision Community level decision responses for 2007 flood responses for 2007 flood

forecasts (Low lands)forecasts (Low lands)

• Secured cattle, poultry birds, homestead vegetables, protected fishery by putting nets in advance

• Planed to evacuate and identified high grounds with adequate communication and sanitation facilities

Community level decision responses Community level decision responses for 2007 flood forecasts (High lands)for 2007 flood forecasts (High lands)

• Protected homestead vegetables by creating adequate drainage facilities

• Livestock was protected in high lands with additional dry fodder (paddy straw)

• Early harvesting of B.aman rice and jute anticipating floods in Gaibandha and Sirajganj, respectively.

Selvaraju (ADPC)

2007 ADPC Warnings issued …2007 ADPC Warnings issued …“We were able to inform the people in advance and on 25th July we started communicating the information to as many people as possible about the certainty of exceeding danger levels along the Brahmaputra…The local partners, non-government organization (NGO) networks and DMC members were advised to inform the poorest of the poor, especially those people living in river islands (“chars”)...”

“On the 28th and 29th, meetings were organized in villages near Rangpur (northern Bangladesh), where the Teesta River was flowing just a few inches below the rim... However, they perceived that the river water level would fall, but our forecasts showed a rising trend…We informed them the significant chance of overflow and breaches, as the embankments are weak in certain places. We engaged the local partner NGOs to prepare an evacuation plan urgently…”

“We communicated the forecast to another pilot union DMC chairman (Uria Union in Gaibandha District) directly on July 26th so he could arrange to mobilize resources for evacuation through DMC members and volunteers. All the six villages in the union were later flooded to a height of 4-6 feet on July 29th. We contacted him again on the 29th to know more and he informed us that about 35% of the people in the union were evacuated in advance.”

“The communities in Rajpur Union of Lalmunirhat (relatively medium lands), were able to use the forecast for preparedness activities like mobilizing food, safe drinking water for a week to 10 days, protecting their T. aman rice seedlings, fishing nets, and raising and protecting their fish pods.”

“For the first time we have communicated 10-day in-advance official forecasts of significant chances of exceeding danger level in all the gauge stations along the Brahmaputra River through the FFWC (and not just in the pilot regions) and directly through all local NGOs, and the DMC members obviously had 8 days extra lead-time they otherwise would not have had.”

ConclusionsConclusions

2003: CFAB forecast went operational2003: CFAB forecast went operational

2004: 2004: -- Forecasts fully-automated-- Forecasts fully-automated -- CFAB became an entity of Bangladesh government-- CFAB became an entity of Bangladesh government

-- forecasted severe Brahmaputra flooding event-- forecasted severe Brahmaputra flooding event

2006:2006:-- Forecasts incorporated into operational FFWC model-- Forecasts incorporated into operational FFWC model-- 5 pilot study dissemination areas trained-- 5 pilot study dissemination areas trained

2007: 5 pilot areas activated during two severe flooding 2007: 5 pilot areas activated during two severe flooding eventsevents

Future WorkFuture Work

Dartmouth FloodWatch Program river discharge Dartmouth FloodWatch Program river discharge estimatesestimates

Improved river routingImproved river routing

Fully-automated forecasting scheme applied to other Fully-automated forecasting scheme applied to other river basins in Africa (and elsewhere)river basins in Africa (and elsewhere)

Climate change impacts …Climate change impacts …

Thank You!