Presentation is loading. Please wait.

Presentation is loading. Please wait.

New Method for Bias Correcting GCM data

Similar presentations


Presentation on theme: "New Method for Bias Correcting GCM data"— Presentation transcript:

1 New Method for Bias Correcting GCM data
Development of climate change impact assessment technology (BA14Y9005) New Method for Bias Correcting GCM data Priyantha Hunukumbura, Daikichi Ogawada & Akiko Matsumura 応用水理G

2 Climate change Impact assessment in Brantas Basin …
GCM Scale Downscaling Statistical Meteorological Variables in GCM scale should be downscaled to point scale for basin scale hydrological modeling. Statistical downscaling is one way of doing it. Dynamical GCM Grid Point Scale

3 The method that we used…
quantile mapping method GCM Grid Compare & Mapped into Point scale Point Scale

4 Comparison of Before and after Bias Correction
Monthly Rainfall Monthly Rainfall Top 20 Top 20

5 Problem… Observed GCM-1 Will look closer year 2000 flood
When applying bias corrected downscaled data into a hydrological model Observed GCM-1 Will look closer year 2000 flood

6 Problem… Observed GCM-1 Year2000 flood

7 Identified weaknesses of the method
Not Keeping the Spatial Coherence – (this is very important in hydrological applications where the runoff is not only determine by the magnitude but also its spatial distribution) Trying to Match the large scale process directly into point scale process is not reasonable Due to scale difference Due to scale difference

8 Proposed Method Assume the same observed spatial distribution (Spatial Weight Matrix) Basin Average OBS Basin AverageGCM Spatial Weight Matrix Basin Average (mm) ST1 ST_n 5.5 0.2 1.2 0.8 . 100 1.1 0.5 2.2 GCM RF Rainfall(mm) Time

9 Basin Scale Comparison
Results Basin Scale Comparison Rainfall in January

10 Basin Scale Comparison …
Results … Basin Scale Comparison … May in January

11 Point Scale Comparison …
Results … Point Scale Comparison … GCMs cccma_cgcm3_1 cccma_cgcm3_1_t63 cnrm_cm3 csiro_mk3_0 csiro_mk3_5 gfdl_cm2_0 gfdl_cm2_1 giss_aom iap_fgoals1_0_g inmcm3_0 ingv_echam4 ipsl_cm4 miroc3_2_hires miroc3_2_medres miub_echo_g mpi_echam5 mri_cgcm2_3_2a

12 Results … Spatial Coherence Raw GCM Observed Traditional Method
New Method

13 Discussion New BC method was developed to address the issues of keeping spatial coherence in bias corrected data Strengths Simple Easy to implement and straightforward Spatial coherence is maintained Scale differences is considered Weaknesses Necessary to assume the observed spatial rainfall distribution Further study Applicability of Weather Generators keeping spatial and temporal coherence How to address the scale issues when applying weather generators

14 Thank You !!!!

15 Thank You !!!!

16 Why there is a big difference ???
GCM- GRID Rain Gauge 1 FOBS FGCM GCM RF 2000/01/05 Rainfall(mm)

17 Analyzing the Results - Brantas River Basin, Indonesia
1. Check for Spatial Coherence Observed GCM-1 Log[Rainfall]

18 2. Check for Temporal Coherence
Indices used to analyze the temporal characteristics Day Rainfall Consecutive Total Rainfall Rain Spell Size Rain Spell Amount 1999/01/01 5 1 1999/01/02 1999/01/03 3 1999/01/04 2 1999/01/05 10 15 1999/01/06 1999/01/07 1999/01/08 1999/01/09 25

19 2. Check for Temporal Coherence
OBS Will look closer

20 1994-1995 2. Check for Temporal Coherence OBS
Need to find a way to quantify the temporal similarities Then we can select better GCMs OBS

21 Histogram of Rain Spell Total Rainfall

22 2. Check for Temporal Coherence
Annual Maximum Discharge As the basin is small, we can expect that this difference is mainly due to the mismatch of temporal characteristics of the bias corrected GCM rainfall

23 Findings Though the quintile based downscaling is used for analyzing the effect of climate change on meteorological variables, it can mislead when use them for analyzing the basin scale hydrological impacts. Alternative methods (such as Dynamic downscaling, Numerical weather generators, etc) may be necessary to overcome the mismatch in the Spatial signatures. Mismatch in the Temporal signatures can be used to reject some GCMs . Uncertainty in the projected stream flow is a combination of different sources such as GCM uncertainty, Bias correction method, hydrological model, Observations etc..

24 Analyzing the Results - Brantas River Basin, Indonesia
1. Check for Spatial Coherence Observed GCM-2

25 Analyzing the Results - Brantas River Basin, Indonesia
1. Check for Spatial Coherence Observed GCM-2

26 1. Check for Spatial Coherence
Observed GCM-2

27 2. Check for Temporal Coherence

28 Combined Effect on Annual Maximum Discharge

29

30

31

32


Download ppt "New Method for Bias Correcting GCM data"

Similar presentations


Ads by Google