Daily to Seasonal Operational Flood ForecastingDaily to Seasonal Operational Flood Forecasting
Tom Hopson, NCAR and ADPCTom Hopson, NCAR and ADPCPeter Webster, CFAB and Georgia TechPeter Webster, CFAB and Georgia Tech
A. R. Subbiah, ADPCA. R. Subbiah, ADPC
Overview:Bangladesh flood forecasting
I. Overview of daily to seasonal weather forecast productsII. Seasonal forecasting: Bangladesh CFAB exampleIII. Short-term forecasting: Bangladesh CFAB example
1. Where does good predictability derive?1. precipitation forecast bias removal2. multi-model river forecasting3. accounting for all error: weather and hydrologic errors
IV. Future Work: Dartmouth Flood Observatory
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)
30 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.
forecast products for hydrologic applications• Seasonal -- ECMWF System 3
- based on: 1) long predictability of ocean circulation, 2) variability in tropical SSTs impacts global atmospheric circulation
- coupled atmosphere-ocean model integrations- out to 7 month lead-times, integrated 1Xmonth- 41 member ensembles, 1.125X1.125 degrees (TL159L62), 130km
• Monthly forecasts -- ECMWF- “fills in the gaps” -- atmosphere retains some memory with ocean variability impacting atmospheric circulation- coupled ocean-atmospheric modeling after 10 days- 15 to 32 day lead-times, integrated 1Xweek- 51 member ensemble, 1.125X1.125 degrees (TL159L62), 130km
• Medium-range -- ECMWF EPS- atmospheric initial value problem, SST’s persisted- 6hr - 15 day lead-time forecasts, integrated 2Xdaily- 51 member ensembles, 0.5X0.5 deg (TL255L40), 80km
• Short-range -- RIMES- 26-member Country Regional Integrated Multi-hazard Early Warning System (RIMES) WRF Precipitation Forecasts- 3hr - 5 day lead-time, integrated 2X daily- 9km resolution
1) Greater accuracy of ensemble mean forecast (half the error variance of single forecast)
2) Likelihood of extremes3) Non-Gaussian forecast PDF’s4) Ensemble spread as a representation of forecast
uncertainty
Motivation for Generating Ensemble Discharge Forecasts (from ensemble weather forecasts)
Overview:Bangladesh flood forecasting
I. Overview of daily to seasonal forecast productsII. Seasonal forecasting: Bangladesh CFAB exampleIII. Short-term forecasting: Bangladesh CFAB example
1. Where does good predictability derive?1. precipitation forecast bias removal2. multi-model river forecasting3. accounting for all error: weather and hydrologic errors
IV. Future Work: Dartmouth Flood Observatory
Seasonal rainfall prediction for 2006Seasonal rainfall prediction for 2006An example of An example of
seasonal seasonal predictions of predictions of precipitation issued precipitation issued in JFMA 2006 (left) in JFMA 2006 (left) and MJJA 2006 and MJJA 2006 (right), to be (right), to be compared with the compared with the observed rainfall observed rainfall (dotted line) and (dotted line) and climatology climatology (dashed line).(dashed line).
The seasonal The seasonal forecasts correctly forecasts correctly indicate months in indicate months in advance ‘higher advance ‘higher than normal’ than normal’ rainfall.rainfall.
Overview:Bangladesh flood forecasting
I. Overview of daily to seasonal forecast productsII. Seasonal forecasting: Bangladesh CFAB exampleIII. Short-term forecasting: Bangladesh CFAB example
1. Where does good predictability derive?1. precipitation forecast bias removal2. multi-model river forecasting3. accounting for all error: weather and hydrologic errors
IV. Future Work: Dartmouth Flood Observatory
CFAB Project: Improve flood warning lead time
Problems:
1. Limited warning of upstream river discharges
2. Precipitation forecasting in tropics difficult
Good forecasting skill derived from:1. good data inputs: ECMWF 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 used in data assimilation scheme
1) Rainfall Inputs
1) Rain gauge estimates: NOAA CPC and WMO GTS0.5 X 0.5 spatial resolution; 24h temporal resolutionapproximately 100 gauges reporting over combined catchment24hr reporting delay
2) Satellite-derived estimates: NASA TRMM0.25X0.25 spatial resolution; 3hr temporal resolution6hr reporting delaygeostationary infrared “cold cloud top” estimates calibrated from SSM/I and TMI microwave instruments
3) Satellite-derived estimates: NOAA CPC “CMORPH”0.25X0.25 spatial resolution; 3hr temporal resolution18hr reporting delay precipitation rain rates derived from microwave instruments (SSM/I, TMI, AMSU-B), but “cloud tracking” done using infrared satellites
4) Weather forecasts: ECMWF GCM 51-member ensemble weather forecasts at 1-day to 15-day forecast lead-times (nominal resolution about 0.5degree)
Comparison of Precipitation Products:
Rain gauge, GPCP, CMORPH, ECMWF
-- Increase in forecast skill(RMS error) with increasingspatial scale
-- Logarithmic increase
2) Spatial Scale
Merged FFWC-CFAB Hydraulic Model Schematic
Primary forecast boundary conditions shown in gold:
Ganges at Hardinge Bridge
Brahmaputra at Bahadurabad
3) Benefit: FFWC daily river discharge observations used in forecast data assimilation scheme (Auto-Regressive Integrated Moving Average model [ARIMA] approach)
Daily Operational Flood Forecasting Sequence
Forecast Trigger: ECMWF forecast files
Updated TRMM-CMORPH-CPC precipitation estimates
Updated distributed model parameters
Updated outlet discharge estimates
Above-critical-level forecast probabilities transferred to Bangladesh
Lumped Model Hindcast/Forecast Discharge Generation
Distributed Model Hindcast/Forecast Discharge Generation
Multi-Model Hindcast/Forecast Discharge Generation
Discharge Forecast PDF Generation
Calibrate model
Statistically corrected downscaled forecasts
Generate forecasts Generate hindcasts Generate forecasts Generate hindcasts
Update soil moisture states and in-stream flows
Generate hindcasts
Calibrate AR error model
Calibrate multi-model
Generate forecasts Generate hindcasts
Generate forecasted model error PDF
Convolve multi-model forecast PDF with model error PDF
Generate forecasts
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?
ECMWF 51-member Ensemble Precipitation Forecasts
2004 Brahmaputra Catchment-averaged Forecasts-black line satellite observations-colored lines ensemble forecastsBasic structure of catchment rainfall similar for both forecasts and observationsBut large relative over-bias in forecasts
5 Day Lead-time Forecasts=> Lots of variability
Pmax
25th 50th 75th 100th
Pfcst
Pre
cipi
tatio
n
Quantile
Pmax
25th 50th 75th 100th
Padj
Quantile
Forecast Bias Adjustment -done independently for each forecast grid
(bias-correct the whole PDF, not just the median)
Model Climatology CDF “Observed” Climatology CDF
In practical terms …
Precipitation 0 1m
ranked forecasts
Precipitation 0 1m
ranked observations
Bias-corrected Precipitation Forecasts
Brahmaputra Corrected Forecasts Original Forecast
Corrected Forecast
=> Now observed precipitation within the “ensemble bundle”
Daily Operational Flood Forecasting Sequence
Forecast Trigger: ECMWF forecast files
Updated TRMM-CMORPH-CPC precipitation estimates
Updated distributed model parameters
Updated outlet discharge estimates
Above-critical-level forecast probabilities transferred to Bangladesh
Lumped Model Hindcast/Forecast Discharge Generation
Distributed Model Hindcast/Forecast Discharge Generation
Multi-Model Hindcast/Forecast Discharge Generation
Discharge Forecast PDF Generation
Calibrate model
Statistically corrected downscaled forecasts
Generate forecasts Generate hindcasts Generate forecasts Generate hindcasts
Update soil moisture states and in-stream flows
Generate hindcasts
Calibrate AR error model
Calibrate multi-model
Generate forecasts Generate hindcasts
Generate forecasted model error PDF
Convolve multi-model forecast PDF with model error PDF
Generate forecasts
Discharge Multi-Model Forecast
Multi-Model-Ensemble Approach:
• Rank models based on historic residual error using current model calibration and “observed” precipitation
•Regress models’ historic discharges to minimize historic residuals with observed discharge
•To avoid over-calibration, evaluate resultant residuals using Akaike Information Criteria (AIC)
•If AIC minimized, use regression coefficients to generate “multi-model” forecast; otherwise use highest-ranked model => “win-win” situation!
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 Operational Flood Forecasting Sequence
Forecast Trigger: ECMWF forecast files
Updated TRMM-CMORPH-CPC precipitation estimates
Updated distributed model parameters
Updated outlet discharge estimates
Above-critical-level forecast probabilities transferred to Bangladesh
Lumped Model Hindcast/Forecast Discharge Generation
Distributed Model Hindcast/Forecast Discharge Generation
Multi-Model Hindcast/Forecast Discharge Generation
Discharge Forecast PDF Generation
Calibrate model
Statistically corrected downscaled forecasts
Generate forecasts Generate hindcasts Generate forecasts Generate hindcasts
Update soil moisture states and in-stream flows
Generate hindcasts
Calibrate AR error model
Calibrate multi-model
Generate forecasts Generate hindcasts
Generate forecasted model error PDF
Convolve multi-model forecast PDF with model error PDF
Generate forecasts
Significance of Weather Forecast Uncertainty Discharge Forecasts
3 day 4 day
Precipitation Forecasts
1 day 4 day
7 day 10 day
1 day 4 day
7 day 10 day
Discharge Forecasts
What do we mean by “calibration” or “post-processing”?
Pro
babi
lity
calibration
Basin Rainfall [mm]
Pro
babi
lity
Basin Rainfall [mm]
Post-processing has corrected:• the “on average” bias• as well as under-representation of the 2nd moment of the empirical forecast PDF (i.e. corrected its “dispersion” or “spread”)
“spread” or “dispersion”
“bias”obs
obs
ForecastPDF
ForecastPDF
Producing a Reliable Probabilistic Discharge Forecast
Step 1: generate discharge ensembles from precipitation forecast ensembles (Qp):
1/51
1
Qp [m3/s]
Prob
abil
ity
Step 3: combine both uncertainty PDF’s to generate a “new-and-improved” more complete PDF for forecasting (Qf):
Qf [m3/s]
1Pr
obab
ilit
y
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)
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
2 day
3 day 4 day
5 day
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
Confidence Intervals
2007 Brahmaputra Ensemble Forecasts and Danger Level Probabilities
7-10 day Ensemble Forecasts 7-10 day Danger Levels
7 day 8 day
9 day 10 day
7 day 8 day
9 day 10 day
Overview:Bangladesh flood forecasting
I. Overview of daily to seasonal forecast productsII. Seasonal forecasting: Bangladesh exampleIII. Short-term forecasting: Bangladesh example
1. Where does good predictability derive?1. precipitation forecast bias removal2. multi-model river forecasting3. accounting for all error: weather and hydrologic errors
IV. Future Work: Dartmouth Flood Observatory
Satellite-based River Discharge Estimation
Bob Brakenridge, Dartmouth Flood Observatory, Dartmouth College
2000
2200
2400
2600
2800
1-Jan-056-Jan-0511-Jan-0516-Jan-0521-Jan-0526-Jan-0531-Jan-055-Feb-05
10-Feb-0515-Feb-0520-Feb-0525-Feb-052-Mar-057-Mar-05
12-Mar-0517-Mar-0522-Mar-0527-Mar-051-Apr-056-Apr-05
11-Apr-0516-Apr-0521-Apr-0526-Apr-051-May-056-May-0511-May-0516-May-0521-May-0526-May-0531-May-05
5-Jun-0510-Jun-0515-Jun-0520-Jun-0525-Jun-0530-Jun-05
5-Jul-0510-Jul-0515-Jul-0520-Jul-0525-Jul-0530-Jul-054-Aug-059-Aug-05
14-Aug-0519-Aug-0524-Aug-0529-Aug-053-Sep-058-Sep-05
T, degrees K x 10010000200003000040000500006000070000
Discharge, c.f.s.
Measurement Reach Calibration Target Estimated Discharge Measured Discharge at Piketon
River Watch •Day/Night Flood detection on a near-daily basis regardless of cloud cover.•Measurement of river discharge changes; current flood magnitude assessments•Immediate map-based prediction of what is under water•Access to rapid response detailed mapping as new maps are made•Access to map data base of previous flooding and associated recurrence intervals.
http://www.dartmouth.edu/~floods/http://www.dartmouth.edu/~floods/
Application to the Ganges River Basin
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
MODIS sequence of 2006 Winter Flooding
2/24/2006 C/M: 1.004 3/15/2006 C/M: 1.029 3/22/2006 C/M: 1.095
The Advanced Microwave Scanning Radiometer - Earth Observing System (AMSR-E) is a twelve-channel, six-frequency, passive-microwave radiometer system. It measures horizontally and vertically polarized brightness temperatures at 6.9 GHz, 10.7 GHz, 18.7 GHz, 23.8 GHz, 36.5 GHz, and 89.0 GHz.
Spatial resolution of the individual measurements varies from 5.4 km at 89 GHz to 56 km at 6.9 GHz.AMSR-E was developed by the Japan Aerospace Exploration Agency (JAXA) and launched by the U.S. aboard Aqua in mid-2002.
Objective Monitoring of River Status:The Microwave Solution
One day of data collection(high latitudes revisited most frequently)
Example: Wabash River near Mount Carmel, Indiana, USA
Black square showsMeasurement pixel.White square iscalibration pixel.
1000
1200
14001600
18002000
2200
24002600
2800
19-Jun-0219-Jul-0218-Aug-0217-Sep-0217-Oct-0216-Nov-0216-Dec-0215-Jan-0314-Feb-0316-Mar-0315-Apr-0315-May-0314-Jun-0314-Jul-0313-Aug-0312-Sep-0312-Oct-0311-Nov-0311-Dec-0310-Jan-049-Feb-0410-Mar-04
9-Apr-049-May-048-Jun-048-Jul-047-Aug-046-Sep-046-Oct-045-Nov-045-Dec-044-Jan-053-Feb-055-Mar-054-Apr-054-May-053-Jun-053-Jul-05
2-Aug-051-Sep-051-Oct-0531-Oct-0530-Nov-0530-Dec-0529-Jan-0628-Feb-0630-Mar-0629-Apr-0629-May-0628-Jun-0628-Jul-0627-Aug-06
AMSR-E radiance,degrees K x 10
0
2000
4000
6000
8000
10000
12000
Estimated Discharge (m3/sec)
Site 98, Wabash River at New Harmony, Indiana, USA
2/17/2003 1.18 9/1/2002 1.82 7/24/2004 2.17
Guide to Predicting Inundation Irrawaddy River, Burma
The current hydrologic status and discharge or C/M ratio can be used to determine present inundation extent.
ConclusionsConclusions
2003: CFAB Brahmaputra/Ganges forecasts went operational2003: CFAB Brahmaputra/Ganges forecasts went operational
2004: 2004: -- Forecasts fully-automated-- Forecasts fully-automated
-- forecasted severe Brahmaputra flooding event-- forecasted severe Brahmaputra flooding event
2007: 5 pilot areas warned many days in-advance during two 2007: 5 pilot areas warned many days in-advance during two severe Brahmaputra flooding eventssevere Brahmaputra flooding events
Future WorkFuture Work
Dartmouth Flood Observatory river discharge estimates Dartmouth Flood Observatory river discharge estimates assimilated for improved skillful long-lead forecastsassimilated for improved skillful long-lead forecasts
Fully-automated forecasting scheme relying on global inputs Fully-automated forecasting scheme relying on global inputs (ECMWF forecasts, satellite rainfall) rapidly and cost-effectively (ECMWF forecasts, satellite rainfall) rapidly and cost-effectively applied to other river basins with in-country capacity buildingapplied to other river basins with in-country capacity building
Thank You!
Top Related