CRU : Projects : SO&P : Data : Model Data : HadCM3 : HadCM3 Data

SO&P: HadCM3 Geographic pattern of sea level anomalies

control : 1243-2252 constant forcings

Important notes regarding the HadCM3 data:

Do not redistribute data to others outside the SO&P project.

Hadley Centre will require co-authorship and need to know from the start what analyses you are using the HadCM3 data for. Please contact Simon Tett (see SO&P participants table) and keep him informed.

To ensure collaboration between SO&P partners, please refer to the list of ongoing/planned model-data comparisons. Please contact others who have listed an interest in the same area, to initiate collaboration. To add your planned areas of model-data work to the list, email Tim Osborn.

A small climate drift, unrelated to the external forcings, is present in the HadCM3 simulations. It's significance depends on the region and variable being considered. Simon Tett describes how this should be dealt with, by reference to the HadCM3 control simulation, in this report (not yet available). The HadCM3 control simulation data is not yet available.

Please inform Tim Osborn of any errors that you find in the data.

For important notes regarding the control simulation: click here.

For important notes regarding the sealv variable: click here.

Obtaining fields of model data

Fields of model data, covering the entire globe, are available in netCDF format only (please email Tim Osborn if you cannot use netCDF files). For monthly and seasonal-mean data, there is usually one file for each decade (though not always and the filenames or headers will make it clear when this is not the case). For annual data there are usually separate files for each year.

The netCDF header contains all the information about coordinates and units that you need.

You have two options for downloading the files:

Use wget with the command listed below (click here to obtain the wget software, to understand its options and for an explanation of its advantages):

wget --mirror --no-parent --no-directories --accept=nc --http-user=soaphad --http-passwd=xe2005 http://www.cru.uea.ac.uk/cru/projects/soap/pw/data/model/hadcm3/soaparchive/control/nc/surface.y/

Click here to see a listing of all files, and right-click each one that you want to download/save.

Note that the data files use the HadCM3 variable name instead of the variable name used on the SO&P website. Instead of sealv, the files are named surface.y.

Note also that the units used in the data files may be different to those given in the time series plots/data, because I converted some variables to more common units. Refer to the netCDF file headers for the units used (and any notes below regarding errors in this information).
For this variable, I multiplied the data files by 0.1 to convert the data to mm.

Select alternative simulation for this variable:
control : nat : all

Select alternative variable from this simulation:
slp__ : sst__ : airtm : rlhum : swrad : precp : snowd : soilm : soilt : win10 : z850_ : z500_ : u500_ : v500_ : iceco : iceth : lwrad : sealv

Important notes regarding the 'control' simulation:

The 'control' simulation has constant external forcings, representative of (approximately) pre-industrial conditions (though, for example, the vegetation cover is set to present day).
In the SO&P web archive we have 1010 years of data from the 'control' simulation. The 'control' simulation actually began 750 years earlier, so essentially we have years 751 to 1760 of the simulation (these are not calendar years!). Even at this stage in the simulation, there are some residual climate drifts (trends) in certain variables.

What does the "year" in the control simulation mean? In the actual netCDF data files provided by the Hadley Centre, the first year of the 'control' run was given the year nominal year "1530", and so the years 751 to 1760 that we have are given nominal years "2280" to "3289". The 'nat' simulation was started from the 1000th year of the 'control' simulation (nominal year "2529" in the netCDF files). The 'nat' simulation is associated with specific calendar years (because the time series of external forcings represents calendar years 1492-1999). Therefore, to compare the 'nat' run with the 'control' run, we need to compare calendar years 1492-1999 with 'control' run nominal years "2529"-"3036" (in fact, in SO&P we are using calendar years 1500-1999, which can be compared with 'control' run nominal years "2537"-"3036"). The 'all' simulation begins in calendar year 1750, and because it starts from the calendar year 1750 conditions of the 'nat' run, its relationship with the control run is the same (i.e., calendar years 1750-1999 should be compared with 'control' run nominal years "2787"-"3036"). The following table may help:

Year conversions	nominal year in netCDF files	year of the control run	corresponding calendar year of nat/all runs
start of control run	"1530"	1	N/A
start of control run data held by SO&P	"2280"	751	1243 AD
start of nat run	"2529"	1000	1492 AD
start of nat run data used by SO&P	"2537"	1008	1500 AD
start of all run	"2787"	1258	1750 AD
end of nat and all runs	"3036"	1507	1999 AD
end of control run data held by SO&P	"3289"	1760	2252 AD

To simplify things, just subtract 1037 from the nominal year stored in the 'control' run data files to convert it to a calendar year for comparison with the 'nat' and 'all' simulations. In the time series files this has already been done.

Important notes regarding the sealv variable:

It is extremely important to realise that the sealv variable does not contain the correct global-mean changes.

The global-mean of the sealv data from any one year should be calculated (using area-weighting of each grid cell, which is proportional to the cosine of the latitude of the cell) and subtracted from the sea level at every individual grid cell for the same year. This will yield patterns of local sea level deviations from the global-mean for that year.
To obtain the actual simulated local sea levels, the correct global-mean value of sea level for the year (see HadCM3 global-mean sea level) should be added to the pattern of local sea level deviations.
By adding the sum of (1) the global ocean expansion and (2) the global glacier/small-ice-cap melt component, then the local sea levels will capture both components plus the local deviations from this.
Note also that the absolute values are arbitrary (there is no definition of "zero" sea level) and so some adjustment to a common baseline period is necessary before comparing different regions and/or different model simulations.
- A suggested adjustment is this: after combining the local and global changes as described above, subsequently add a constant to each grid cell's sea level time series so that it has zero mean over the 1500-1549 period in the 'nat' simulation. For the 'all' simulation, add a constant to each grid cell's sea level time series so that it has the same mean as the adjusted 'nat' series over the 1750-1799 period.

Last updated: August 2004, Tim Osborn