1. WFM 6311: Climate Change Risk Management
Lecture-9: Handson Exercise using SDSM to downscaling Temperature of a selected station of Bangladesh
Professor A.K.M. Saiful Islam
Institute of Water and Flood Management (IWFM)
Bangladesh University of Engineering and Technology (BUET)
10 August, 2016

2. Lecture Topic Download and Installation of SDSM Download of GCM data
Preparation of Station data
Consistency of station data
Calibration and validation of model
Future scenario generations

4. Getting Started First http://co-public.lboro.ac.uk/cocwd/SDSM/
Register yourself

5. Installing the software
After login click on the ‘software’ of the menu bar or go to
Download SDSM and the manual

6. Statistical Downscaling Input
Go to
Click on the ‘Statistical Downscaling Input’ on the left panel.
Then click on HadCM3 Predictors

7. Statistical Downscaling Input
Which box to choose?

8. Dhaka BMD station (90.38 , 23.78) Box # ?? (25x 25y) (90, 25)
(90, 22.5)
(25x 26y)
Dhaka BMD station
(90.38 , 23.78)
Box # ??

9. Statistical Downscaling Input
H3A2a_ (Hadley Centre Climate Model 3 Emissions Scenario A2 - pessimistic)
h3a2mslpas.dat
h3a2p5_fas.dat
H3B2a_ (B2 - optimistic)
h3b2mslpas.dat
h3b2p5_fas.dat etc
NCEP_
ncepmslpas.dat
ncepp5_uas.dat etc (26 files)

10. Statistical Downscaling Input
Depending on the location of the station data, choose the right box numbers and click on the ‘Get data’. For example, for Dhaka station choose 25X25Y.
Downloading of the zip file will start.

11. [source] [variable] [grid box]. Dat nceprhumee. dat. h3b2p8_zsw. dat
[source] [variable] [grid box] . Dat nceprhumee.dat ?????? h3b2p8_zsw.dat ?????? h2ggp_thas.dat ??????

14. Station Data Preparation (1961 -2000)

16. Uncheck ‘Allow -ve values’ Event threshold = 0(temp) = 1 (rainfall)
Select Settings
Check Calendar (366)
Uncheck ‘Allow -ve values’
Event threshold = 0(temp)
= 1 (rainfall)
Data 1/1/1961 – 12/31/2000
Save the settingc
For Easy selection
Select the source directory of your station data
Select the folder that has your predictand /data

17. Click on advance of settings
Option ‘None’ for temperature
Others for rainfall (i.e., fourth root)
Save the settings

18. Click Quality Control
Then Select file
Open the input station data
Click on the Check file

19. If OK appears, quality check is complete.
Manually use the judgment by checking the maximum difference of the temperature

20. Data Transformation Change of predictors (optional)

21. Screen Variables ATTENTION ATTENTION
Select the NCEP_ folder of appropriate Box
ATTENTION
Temp Unconditional
Rainfall Conditional
Select station data as the predictand file (i.e Tmax_dat)
Select your data period (61-00) and analysis period (i.e. Annual)
Check your process condition (i.e. conditional)
Amend the significance level (i.e for 5% significant value).
SELECT APPROPIATE COMBINATION OF PREDICTOR VARIABLES (<#12)
PRESS

22. Choose those combination of variables where partial correlation (r) is high but with P value ≈ 0
Go back and PRESS

23. Example: Dhaka Station
Analysis Results

24. Example: Dhaka Station
Scatter Plot
Station data Vs 1 Predictor

25. Dhaka Station (CALIBRATE MODEL)
Select those chosen combination
for Tmax_61-80.PAR

26. Dhaka Station (Calibration Results)
Standard Error = 2%
(Average)
Explained Variance = 23.65%
(Average)

27. (to Validate create output file from 1981-2000)
Weather Generator
(to Validate create output file from )
NCEP Files
1/1/1981 for Tmax_NCEP_81-00
From
Tmax_61-80.PAR

28. Summary Statistics Go to Summary statistics and press
Select suitable statistics (maximum up to 8), and go back

29. Summary Statistics
Click

30. Compare Results Check the results (i.e. mean Vs mean)
If unsatisfactory, choose another combination of screen variables and repeat the above procedures. Validation is an iterative process
If satisfactory, Go to

31. Create calibrated file
Create Tmax_61-00.PAR with same screen variables combination (validated)
Use the data period 1961 – 2000
Calculate the Explained Variance
Explained
Variance = 26.04%

32. Create Present Climatic data (NCEP_1961-1990)
Tmax_61-00.PAR
NCEP Files
Tmax_NCEP_ out
1/1/1961
Calibrate Model
61-00.PAR-> NCEP_61-00.OUT
Summary Statistics
NCEP_61-00.OUT->NCEP61-00.TXT

33. Creating Model Data Change the settings calendar to 360 (GCM)
Tmax_61-00.PAR
H3A2_
Change the settings calendar to 360 (GCM)
Create H3 A2 or H3 B2 model data (.out) through Scenario Generation
for the year

34. Creating Model data (present +Future)
Choose Modelled
INPUT FILE
Tmax_H3A2_ out
Analysis period will depend on the desired (present or future) data time
‘Summary statistics’
Choose Statistics options
Create text files (Tmax_H3A2_ txt, Tmax_H3A2_ txt etc.)

35. Sample: Delta Stats Results
Choose Modelled
INPUT FILE
H3A2_ out
Analysis period

36. Sample: Delta Stats Results……… Cont

37. Sample: Generated output data

38. Sample: Frequency Analysis
Tmax.dat as the observed data
H3A2_ out as the modelled data
Ensemble Mean (Q-Q plot)
Ensemble Member 1 (FA)