Analyzing BIOMEBGC Data in IDL

1
Analyzing BIOME-BGC Data in IDL

• Matt Jolly
• NTSG, Univ. of MT.

2
Introduction

• Discuss the format of common BIOME-BGC outputs
• Determine the best way to read outputs into IDL
  variables
• Apply data processing function to analyze data

3
BIOME-BGC Outputs

• Three principle output types from BGC
• Daily
• Monthly
• Annual
• Monthly and Annual outputs are the average of the
  daily values
• The outputs are defined in the Biome-BGC ini file

4
Data Format

• Data are always organized sequentially
• I.E. a series of bytes that are read in by the
  program
• BGC writes out data for each time step iteration,
  so the files are ordered as follows
• If there were two variables and two sim days
• (Var1,Day1)(Var2,Day1)(Var1,Day2)(Var2,Day2)
• If BGC writes data out as single precision
  floating point values, how many bytes should the
  above file be?

5
Reading BGC data in IDL

• The hardest part is dimensioning the array
  correctly. The rest we have done before.
• Dimension the input variable as follows
• Number of variables
• Number of intervals in a year
• 365 for daily, 12 for monthly, can be omitted for
  annual
• Number of years

6
The IDL Variable

• Lets say we are reading in daily data and in the
  INI file we gave it 23 variables to write out and
  told the model to run for 44 years.
• Dimension the IDL variable as follows
• Bgcdata fltarr(23,365,44)
• Better than bgcdata ftlarr(369380L)
• Note organization into logical groups
  (days,years) is done by IDL

7

• DAILY_OUTPUT (keyword)
• 23 (int) number of daily variables to output
• 20 0 ws.soilw
• 21 1 ws.snoww
• 38 2 wf.canopyw_evap
• 40 3 wf.snoww_subl
• 42 4 wf.soilw_evap
• 43 5 wf.soilw_trans
• 44 6 wf.soilw_outflow
• 70 7 cs.cwdc
• 509 8 epv.proj_lai

8
Hierarchical Data Organization
Year1
Year2


Day1
Day2
Day365
Day1
Day365



Vars1 - 10
Vars1 - 10
Vars1 - 10
Vars1 - 10
Vars1 - 10
Only this level is actually stored, the other
levels are just for organization
9
Recommendations

• Subset the data
• Trans data5,,
• Define variables to index the array
• Trans 5
• Plot,datatrans,,

10
Summarizing data

• The TOTAL function sums an array based on an
  index
• Syntax
• Result total(array, arraydim)
• Array is the array to sum and arrayindex is the
  dimension to sum by
• For example
• Trans data5,,
• Help, trans -gt fltarr(365,44)
• Transtot total(trans,1)
• This sums the trans for each year

11
Total used to get the mean

• Simply divide the result by the original size of
  the dimension
• Mean total(trans,1) / 365

12
Regressions

• IDL is a powerful statistical package, especially
  once you have read the data into local variables
• A common task is to determine whether or not
  there is a linear relationshiop between variables
• We can do this with regressions

13
REGRESS

• Ordinary least squares regression
• Result REGRESS( X, Y,, CONSTvariable
  , CORRELATIONvariable , FTESTvariable)
• T REGRESS(LAIMAX,TRANSTOT,CORRELATION CORR)

14
Exercise

• Read a BIOME-BGC .dayout file into IDL
• Subset transpiration and NPP
• Sum both variables
• Plot Transpiration as a function of NPP