A Three-Tier System for Forecasting Precipitation and River Discharge in the Monsoon Regions:

A Three-Tier System for

Forecasting Precipitation and River Dischargein the Monsoon Regions:

Potential Application to the Mekong River System

Tom HopsonCFAB: CU/GT/ADPC

To date, forecasting of river discharge and regional precipitation has been confined to Bangladesh and India under the auspices of

the Climate Forecast Applications in Bangladesh (CFAB)Project funded by USAID and NSF.

Currently regional precipitation and river discharge (Brahmaputra and Ganges) forecasts are made on three time scales: ==> 1-10 days ==> 15-30 days ==> 1-6 monthsThese forecasts are issued routinely by CFAB on an experimental basis and used by the Bangladesh Meteorological Department and the Flood Forecasting and Warning group.

The question is whether the system can be applied to the MekongRiver system.

• Most certainly the 1-6 month and the 1-10 day forecasts are directly applicable. Preliminary analysis indicates that the system used for the 15-30 day forecasts may also be applicable.

• To this end, we propose to develop a three-tier system for the lower, middle and upper Mekong for both precipitation and river discharge.

Overview

1. The Climate Forecast Applications in Bangladesh (CFAB) project

2. The three-tier forecast system for precipitation and river discharge

3. Potential application to the Mekong System

4. Ways forward

The Climate Forecast Applications Project CFAB

Bangladesh lies at the confluence of three major rivers, the Meghna, Brahmaputra and Ganges. No stream flow data is available in Bangladesh beyond its borders. Furthermore, the catchment areas of the latter two rivers are very large.

(b) August 2002 floods(a) September 1998

CFAB’s problem was how to produce flood forecasts for Bangladesh without detailed upstream data for mega-flood years like 1998 (a) or regular flooding during the wet season (b).

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999

(a) Observed river discharge into Bangladesh Brahmaputra and Ganges Discharge into Bangladesh

The only data available were the long records of river discharge of the two major rivers as they entered Bangladesh

A further problem was how would the forecasts be disseminated to the user communities once they were produced?

BrahmaputraGanges

Experimental Operational Forecasts in Bangladesh

• Prior to 2000, Bangladesh had no numerical meteorology/hydrological capability

• 10 Different agencies each operating independently (GoB, NGOs ….)

• Primary Job: Create cooperative infrastructure• Utilized Asian Center for Disaster Prevention

(ADPC) for in-country connections• Formed steering committee in 2000 which meets

every month

Infrastructure Development ……

• Steering group acts as interface between US scientists, GoB technical and administration and the user community

• Forecasts are now prepared and disseminated to the GoB network of agricultural extension, flood units, disaster relief and etc.

• Much work still to be done in capacity building and utilization of forecasts at base user level

• Current work in developing a user metric system for interpretation of forecasts in probabilistic and meaningful context.

Three-Tier Overlapping Forecast SystemDeveloped for Bangladesh

SEASONAL OUTLOOK: “Broad brush” probabilistic forecast of rainfall and river discharge. Updated each month. Reliable for 3 months especially early in season.

20-25 DAY FORECAST: Forecast of average 5-day rainfall and river discharge 3-4 weeks in advance. Updated every 5 days.

1-10 DAY FORECAST: Forecast of rainfall and precipitation in probabilistic form updated every day. Considerable skill out to 5-days. Moderate skill 5-10 days.

Utility of a Three-Tier Forecast System

SEASONAL OUTLOOK: Long term planning of agriculture, water resource management & disaster mitigation especially if high probability of anomalous season (e.g., flood/drought)

20-25 DAY FORECAST: Broad-scale planning schedules for planting, harvesting, pesticide & fertilizer application and water resource management (e.g., irrigation/hydro-power determination). Major disaster mitigation resource allocation.

1-10 DAY FORECAST: Detailed agriculture, water resource and disaster planning. E.g., fine tuning of reservoir level, planting and harvesting.

What if rainfall forecasts had been available for the Ganges Valley during the summer of 2002

The summer of 2002 was a major drought year in India. The total summer rainfall in the Ganges Valley was 15% below average and in other regions far worse. OverIndia, total rainfall was -19%. The extended late-June to mid-July drought was crippling.

-60%

-15%

-20%

Availability of skillful seasonal (1-6 month) forecasts

• In general, seasonal forecasts, even if skillful are of limited utility as they are generally made for a large area. They also do not give spatial or temporal variability of rainfall and omit important details such as active and break periods of the monsoon.

• However, if the forecast is for an extreme season then they are of use because most areas are of similar sign in their anomaly.

• For example, a forecast of the 1998 flood year or the 2002 drought year would have been most useful for long term, planning

Availability of skillful 15-30 day forecasts

• Arguably, the time scale of a forecast on which water resource management and agriculture can benefit is on the multi-week time scale.

• However, these forecasts have been difficult to produce with skill or accuracy. However, under the auspices of CFAB, we have started to produce multi-week forecasts of the average 5-day rainfall and river discharge.

The utility of a 20-25 day forecast can be seen from the following example

Year 2002

100 150 200 250 3000

5

10

15

20

day

MAY JUN JUL AUG SEP OCT

Discussing optimal forecasting times and the 2002 drought over central India, A. R. Subbiah (ADPC) noted: “The minimum length of time of a forecast that will allow a farming community to respond and take meaningful remedial actions … about 10 days although 3 weeks would be optimal… Assuming (such) were available by the third week of June 2002… farmers could have been motivated to postpone agricultural operations saving investments worth billions of dollars… water resource managers could have introduced water budgeting measures…”

Availability of skillful 1-10 day forecasts

• Accurate short- term forecasts of 1-10 days allow for modification of longer term plans and the opportunity for detailed plans water resource management and agricultural practices.

• These forecasts are the most accurate and are updated daily.

• Considerable skill and utilization has been demonstrated in CFAB for both the Ganges and Brahmaputra.

Current Experimental Forecasts Scheme:A Three-Tier System

Three concurrent overlapping precipitation and river discharge forecasts are produced continually throughout the spring and early summer for Bangladesh (and Indian rainfall districts)

SEASONAL OUTLOOK• 1-6 month forecasts in

probabilistic mode issued each month

Seasonal: May plus 6 months

Seasonal: June plus 6 months

Seasonal: July plus 6 months

Seasonal: August plus 6 months

Seasonal Outlook Forecasts

• Based on ECMWF coupled model output plus statistical hydrological modelling

• Forecast of least confidence: provides excellent 3 month forecasts spring ==> summer but loses predictability late in season.

• Provides important “broad-brush” outlook potentially allowing long-term planning

• Scheme in which we expect greatest improvement

Current Experimental Forecasts Scheme:A Three-Tier System

Three concurrent overlapping precipitation and river discharge forecasts are produced continually throughout the spring and early summer for Bangladesh (and Indian rainfall districts)

15-30 DAY FORECASTS• 15-30 day forecasts of

pentad rainfall issued every five days

20-day forecasts for Ganges Plain

20-day forecasts for Orissa

20-day forecasts for Rajasthan

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999

(a) Observed river discharge into Bangladesh Brahmaputra and Ganges Discharge into Bangladesh

20 40 600

10Ganges 1996

30 50

(b) 20-day forecast of river discharge into Bangladesh

observed forecast

5

20 40 600

10

30 50

5

20 40 600

10

30 50

5

20 40 600

10

30 50

5

pentad pentad

pentad pentad

Ganges 1998

Brahmaputra 1996 Brahmaputra 1998

H&W(2003)

Intermediate Forecasts

• Arguably the most important time scale forecasts for agricultural and water resources planning and disaster mitigation

• Provides good skill• Need to learn how to implement• Forecasts based on wide range of data• Will couple with ECMWF output next year

(model based empirical forecasts)• Provides ideas for numerical modeling

Current Experimental Forecasts Scheme:A Three-Tier System

Three concurrent overlapping precipitation and river discharge forecasts are produced continually throughout the spring and early summer for Bangladesh (and Indian rainfall districts)

1-10 DAY FORECASTS• Forecasts of precipitation and river

discharge issued every day in probabilistic mode

Brahmaputra Discharge

1 day 2 day

3 day 4 day

5 day

Ganges Discharge

1 day 2 day

3 day 4 day

5 day51 ensembleMembers in color

Observed QBlack dash

Brahmaputra Discharge Ganges Discharge

6 day 7 day

8 day 9 day

10 day

6 day 7 day

8 day 9 day

10 day

Brahmaputra Flood Probability Ganges Flood Probability

1 day 2 day

3 day 4 day

5 day

1 day 2 day

3 day 4 day

5 day95%

50%

95%

50%

Short-term Forecasts

• Based on ECMWF 51-member forecasts.• Shows excellent skill out to 5-7 days.• Extends Bangladeshi forecasts to 7-9 days• Incorporated into Bangladesh flood warning

program • Added value if used in conjunction with the

intermediate (15-30 days) forecasting system

The Mekong River system

A

B

C

Define three zones for the Mekong:

A: Upper B: Middle C: Lower

These definitions are initially arbitrary. Finalchoice would be madeon the basis of common meteorology and hydrology

• Short-term (1-10 days) and seasonal forecast schemes can be adapted immediately.

• Intermediate forecasts (15-30 days) depend on the “form” of the weather disturbances that exist in a region. There is a need for intraseasonal variability to be a major component of the climate. Most likely, 15-30 day forecasts will be possible but some basic work is required.

Can the CFAB schemes be adapted for use in the Mekong River system?

Middle Mekong precipitation rate for year 1999. Note variability on

15-30 day time scales

mid-Mekong rainfallclimatology

Plan of Action

• Determine the meteorological and hydrological homogeneity of the three regions: make adjustments of areas as necessary. Requires analysis of precipitation (in situ and satellite) and stream flow data.

• Design a three-tiered forecast system (as per the CFAB bangladesh system) for the three regions of the Mekong.

• Produce a time table of work from design to operational implementation including training, technology transfer and etc.

• Prepare a detailed proposal to the MRC if invited. Proposal and MOU would be between MRC, Georgia Institute of Technology and the Asian Disaster Preparedness Center (ADPC)

Note on the Forecast Modules

• The three forecast modules are relatively simple and can be adapted for workstation or laptop.

• The major problem is the transmission of considerable amounts of data upon which the modules depend. The Problem is simply one of band width

• To use the modules, interpret the results and modify the modules for other uses or extensions requires some basic training.

Thank you

Thank you