Im Impa pact cts s asse ssess ssmen ents s of of water er - - PowerPoint PPT Presentation

Im Impa pact cts s asse ssess ssmen ents s of

• f water

er re resour sources ces in n the he Ka Kalu Ganga nga ba basi sin, , Sri ri La Lanka nka & Pre reliminar ary y ana nalys ysis of

• f dow

downscal nscaling ng sc sche heme e by by GSMaP MaP

Cho ho Tha handa nda Nyunt yunt Hi Hiroshima a Un Univer ersity

The 7th GEOSS Asia-Pacific Symposium Tokyo, Japan, 27 May2014

1

Regional Climate Simulation

• Asia Monsoon
• ITCZ

2

Bias of GCM

• Under

estimation of extreme rainfall

• Poor seasonality
• Frequency of wet

days

3

Low horizontal resolution

• Insufficient for

basin scale run off simulation

• A lot of uncertainty

Multi-GCM selection Statistical Bias correction Spatial disaggregation (Downscaling)

Deficiencies in GCM

4

Extreme rainfall Seasonal rainfall Wet day frequency

GCM Selection

Parameter

• Precipitation (GPCP)
• Outgoing Long wave Radiation (NOAA)
• Sea Surface Temperature (HADLEY)
• Sea level pressure (JRA-25)
• Specific Humidity (JRA-25)
• Zonal wind (JRA-25)
• Meridonial wind (JRA-25)
• Geopotential Height (JRA-25)

Time Window

• Wet/rainy season

Analyzed Window

• Local domain
• Regional domain

Scoring

• Average over all GCM
• Scorr > Scorr_allgcm_avg

 1

• RMSE < RMSE_allgcm_avg  1
• Both Scorr & RMSE

 1 final 1

• Either Scorr or RMSE

 1  final 0

• Neither Scorr nor RMSE

 1  final -1

Criteria

• Highest total scores
• Precipitation (both Scorr and RMSE satisfied)
• Inter-seasonality (bimodal or single mode)

Rain Type Threshold Correction

Extreme

• > 99% of daily precipitation during analysis period
• same frequencies of extreme as insitu station as

in GCM Generalized Pareto Distribution

• Non every year statistics
• Extreme (long or short tailed) fitting
• Peak over threshold method

No rain day

Ranking order statistics

• frequency of no rain day in GCM is same as station
• less than no rain day threshold change zero rainfall.

Normal

Gamma Distribution

• monthly CDF of GCM mapping to

monthly CDF of station

• inverse of Gamma CDF in

each month is corrected rain

Extreme

GCM

No rain day Extreme

Normal rain

GCM=0

No rain day

9

Bias correction scheme

DEM Soil Land Use Vegetation Rainfall Temperature Relative Humidity Wind Speed Solar Radiation JRA25 Reanalysis

WEB-DHM

Calibration Validation Basin Parameter Fl Floo

• od

Drou

• ught

Basin Impact & Vulnerability Assessment

Decision Making & Planning IWRM

Insi-tu Rain(1981-2000) CMIP3 20C3M(1981-2000) A1Bscenario(2046-2065) Multi-GCM Selection Climate Change Assessment Statistical Bias Correction Spatial disaggregation Downscaling

20C3M : 20th Century Numerical Reproductive Experiment CMIP3 : Coupled Modeled Intercomparison Project Phase-3

Climate Change Impact Assessment Framework WEB-DHM

Kalu Ganga River Basin

Sri Lanka

Region and Parameter

Tibetan plateau

Ocean

South Asia monsoon East Asia monsoon

Spatial distribution of 10 parameters  4 region May - August  Monthly Spatial Correlation Monthly Root mean square error

GCM Selections

GPCP

Mri_cgcm2_3_2 Miroc3_2_hires Iap_fgoals Ncar_ccsm30 Giss_model_e_r Ncar_pcm1 Ipsl_cm4 Miroc3_2_medres Ukmo_hadcm3 Mpi_echam5 Ingv_echam4 Mub_echo_g Gfdl_cm2_0 Ukmo_hadgem1 Gfdl_cm2_1 Cccma_cgcm3_1t63 Inmcm3_0 Giss_aom Csiro_mk3_0 Giss_model_e_h Cnrm_cm3 Csiro_mk3_5 Cccma_cgc_3_1 Bccr_bcm2_0

Bias correction results at point Scale (1981-2000)

Before Bias-Correction After Bias-correction

Top 1% Extreme Normal rainfall (1981-2000) Extreme Frequency Analysis

Bias correction results for Basin Scale

Insitu Rain Gauge (May)

Gfdl_cm21 Ingv_echam4 Miroc32_hires Raw GCM rain Bias corrected rain

The Hydrological Model: WEB-DHM

The Water and Energy Budget-based Distributed Hydrological Model ▪ Fully couples a simple biosphere scheme SiB2 with a hillslope hydrological model GBHM. ▪ Allows consistent descriptions of water, energy and CO2 fluxes at the basin scale.

Benefits of WEB-DHM over other models: flexible, low computational requirements; and comparable to commercially available models

Source: Wang, et al., 2009, JGR; Wang, et al., 2009, HESS

Pre reliminar ary y ana nalys ysis of

• f dow

downscal nscaling ng sc sche heme e by by GSMaP MaP

Dataset GSMap_MVK+ Observed Grid resolution

0.1 d 1 degree l lat at/lon

• n

point R Rain g gauge uge

Temporal resolution

1 d 1 day ay 1 d 1 day ay

Available Domain

Global

• bal(60N

60N-60 60S) Basi sin sc scale

Available period

Monthly( y(20 2003 03-200 2008) 8) Daily( y(20 2003 03-200 2008) 8) Hourly( y(20 2003 03-200 2008) 8) Daily( y(19 1981 81-200 2008) 8)

GSMaP: Global Satellite Mapping of Precipitation

Bias Corrected GCM 1.13 Degree GSMapraw 0.1 Degree Basin Scale 0.01 Degree GSMaP corrected rainfall based spatial distributed weights Map Basin Scale 0.01 Degree Insitu rainfall station Basin Scale Monthly correction Bias corrected Downscaled rain

IDW

Insitu stations based gridded GSMap 0.1 Degree

Vs

Downscaling (Spatial disaggregation) Scheme

25

Delineating monthly distributed weight

Sensitivity of numbers of rain gauges

Sensitivity of Numbers of Rain gauges Over 3000 km2

Observed Miroc32_hire

Bias corrected + Downscaled Rainfall

41

490.04 mm/month 465.597 mm/month 488.6577 mm/month 471.5055 mm/month

Mean(mm/day) Stdev RMSE (mm/day) Observed 10.39 14.05 Cccma_cgcm31 11.38 14.88 12.04 Cccma_cgcm_t63 11.39 14.95 12.09 Gfdl_cm20 11.94 14.65 11.9 Gfdl_cm21 11.96 14.49 11.86 Ingv_echam4 11.35 14.83 12.04 Miroc32_hire 11.47 14.93 12.2 Miroc32_medres 11.38 15.04 11.81 Mpi_echam5 10.94 15.24 12.11

42