Bias correction of Dynamic Downscaled Typhoons Rainfall Data for Hydrological Applications National Science and Technology Center for Disaster Reduction Dr. Yuan-Fong Su 27 May 2015 2015 International Workshop on Typhoon and Flood (IWTF) APEC Experience Sharing on Hazardous Weather Events and Risks Management
Questions for climate change dataset What scenario is it? What the spatial and temporal resolutions are it? Can we use the downscaled data directly for impact assessment? Similar in mean value (climatology) Does they similar in other statistics? Two totally different distributions can have similar mean values.
Objectives Dynamic downscaled rainfall from a regional climate model provides useful insight for future climate changes at local scale.
Objectives Contributing from various sources of uncertainties, bias in the high-resolution model output is inevitable. To conduct a reliable assessment, the bias in the models should be corrected before it was applied for further applications, such as flood and storm surge simulations.
Aims Correct the bias of hourly typhoon rainfall derived from MRI-WRF by quantile mapping approach. After bias correction, possible changes in spatial pattern of the hourly typhoon rainfall are also illustrated.
Materials Dynamic Downscaled Data : MRI-WRF MRI data from Japan WRF Dynamic Downscaling Single model A1B Scenario Spatial: 20 km Temporal: 6 hour Spatial: 5 km Temporal: 1 hour Projection periods Base (1979-2003) Near Future (2015-2039) Far Future (2075-2099)
Materials: Ground Observation 84 gauges from WRA and CWB. Typhoon events OBS: 108 Base: 88 Near Future: 81 End of Century: 82
Comparison of averaged typhoon rainfall Total rainfall of typhoon events At 84 gauges and its corresponding MRI-WRF cells. Underestimation of rainfall is obvious. OBS. OBS MRI-WRF MRI-WRF Before BC (Base) (mm)
Method Bias Correction Quantile Mapping XX CC,ii = FF OO 1 qq PP,jj XX pp,ii RR(XX PP,ii ) = XX CC,ii XX PP,ii
Results Hourly typhoon rainfall After correction, scatter points are closer to the line of equality. Ratio are all larger than 1.
Results Averaged rainfall amount of typhoon events OBS. MRI-WRF Before BC (Base) MRI-WRF After BC (Base) (mm)
Results Averaged rainfall amount of typhoon events Applied the ratio to all 1566 MRI-WRF cells. Base Near Future End of Century (mm)
Change rate of typhoon rainfall Near Future End of Century CCCC % = XX FFFFFFFFFFFF XX BBBBBBBB pppppppppppp XX BBBBBBBB pppppppppppp 100%
Conclusions of bias correction. Quantile mapping applied to MRI-WRF datasets worked quite well by correcting the original model outputs to having similar population characteristics of observations and the correction factor for daily solar radiation was proposed.
The biases in the original MRI-WRF model outputs, such as the underestimation of hourly typhoon rainfall, was markedly reduced.
An marked increasing of typhoon rainfall, up to ~50% in far future and ~20% in near future, were observed in the west and central parts of Taiwan while a decreasing, up to ~-30%, also clear in the north and east parts.
Further applications of bias-corrected typhoon rainfall in TCCIP project. Estimation of sediment volume induced by debris flow and landslide. River overflow (1D) and urban inundation (2D) modelling (SOBEK). Storm surge and tidal modelling (FVCOM).
Averaged duration of typhoon rainfalls Base Near Future End of Century
Top 1 event in 3 periods Base Near Future End of Century
48 hours 72-hours Compare the Top 20 typhoon events in base and end of century. Base End of Century 3 hours 6-hours 12 hours 24-hours
Estimation of sediment volume induced by debris flow and landslide. Landslide susceptibility by TRIGRS Landslide occurrence threshold The estimate of sediment volume includes debris flow and landslide. The sediment volume of landslide is 74,326,000m 3 and of debris flow is 14,730,000m 3. Debris flow subcatchment classification Debris flow simulation by Flo- 2D Sediment volume estimation The total sediment volume from the sub-basin in the upstream area is 89,056,000m 3. But the number is still less than the sediment amount generated by Typhoon Morakot (91,080,000m 3 ). (Dr. Ting-Yeh Wu and Ms. Hung-Ru Shih)
River overflow (1D) and urban inundation (2D) modelling (SOBEK). tide(m) 0.60 0.40 0.20 0.00-0.20-0.40-0.60-0.80 Settings of scenario: End of century TOP1 typhoon and sea level rising 140cm 0 4 8 12 16 20 24 28 32 36 40 44 time(hr) rainfall(mm) 100 80 60 40 20 0 Regional rainfall estimated by Thiessen s polygon method TOP1_2D_rainfall 1 5 9 13 17 21 25 29 33 37 41 45 time(hr) Tsengwen and Yanshui rivers Maximum inundation depth(m) >3(m) (Dr. Hsiao-Ping Wei and Dr. Wei-Bo Chen)
Storm surge and tidal modelling (FVCOM) Present No seawall Prototype typhoon Near future End of century seawall (2m) No seawall Less impact We shift the track of typhoon 12km to the west of original one. The seawall is set a height of 4.54m. seawall (2m) Blocking drainage (Dr. Wei-Bo Chen)
Thank you for attention. Q & A