Hi everyone,
I need to use the Meteosat SIS dataset in MJ/m2/day. As the dataset is provided in W/m2, I was just wondering if I could use the relationship 1 W/m2 = 1 J/m2/s to make the conversion. This would give 1 W/m2 = 0.0864 MJ/m2/day.
Many thanks and best wishes
João
Hi Joao,
I am using the SIS dataset for my research and in order to make the W/m2 I use the conversion formula (SIS *60sec *60min*24h) /1000000 which makes it into MJ/m2/day.
or simply (SIS * 3600*24)/1000000
As the Daily SIS are the mean estimates of the day taking into account 24hours.
Cheers,
Theo <--- A passionate SIS user ;)
Hi Theo,
Great to hear from you and many thanks for the quick feedback.
I'm using the Meteosat SIS monthly product. Yes, the conversion formula you indicated just result in a factor of 0.0864 to convert from W/m2 to MJ/m2/day. However, you also need to apply a scale factor of 0.1 to the Meteosat SIS data downloaded from CM SAF.
Cheers
João
Hi Joao,
I am not sure about the scale factor that you suggested. The reason is that with the formula that I mentioned before, my results come with a correlation of 0.97-0.98 when compared to station based observations. Therefore I am very confident that the 0.1 scale factor is not nessasary at all.
cheers,
Theo
Hi Theo,
to get the SIS values in W/m2 you need to apply a scale factor of 0.1. I used IDL to read and process the NetCDF files, so maybe I'm getting different values from you. As for the correlation with station based data, the range that you get would be the same if you applied the scale factor or not.
Cheers
João
Hi Joao and Theo,
the values in the netcdf-files of the SIS Meteosat data set (1983 to 2005) are multiplied by a factor of 10 to reduce the size of the data. So a scale_factor of 0.1 has to be applied before processing of the data. No scale-factor has to be applied to the operational data from CM SAF.
One advantage of using netcdf as data format is the availability of a convention, in this case the CM SAF data follows the Climate and Forecast convention. Within this convention a number of attributes and standard names for physical quantities are defined for easy access netcdf data that follow these conventions. The local attribute 'scale_factor' is defined in Section 8.1. Together with 'add_offset' it is intended to reduce the size of the data files.
Some software (e.g., the ncdf-package from R) does apply these factors and the offsets directly to the data when reading the data from the file, so the user does not need to apply this factor. In case the reading-software does not account for the scale_factor and the add_offset attributes the user has to do this manually by reading out these attribute from the netcdf file and applying them to the data.
So, it does depend on the software you are using whether you have to apply the scale factor manually or not. When using R with the ncdf-package the scale-factor does not need to be applied, while IDL seems not to take the scale_factor-attribute into account automatically.
Please note that the ncdf-package does not apply the _FillValue-attribute for missing data, but uses the 'missing_value'-attribute, which was used in previous version of the CF-convention to mark missing data. Hence, in the R-scripts provided by CM SAF the _FillValue attribute is explicitely read from the netcdf file and applied to the data.In this case, particular attention has to be given when a factor and/or offfset has been applied to the data....
Hope this clarifies some issues, kind regards,
Jörg
I guess then that when I export timeseries using the R scripts provided by the CMSAF the results are correct, hense the good correletion with the station data,
but when I use only CDO to merge the data into one and the use McIDAS or other software, my results come with a factor of 10 error :(((.
Average annual sum of SIS across the British Isles based on 1983-2005 period (MJ/m2/year)
Is this figure a realistic one or the values should be of a magnitude of 280MJ/m2/year to 450MJ/m2/year?
Further to my analysis,
Surface based observations are reported in KJ/m2/day ( from SYNOP stations)
I convert the daily satellite estimates from Wm2 to KJ/m2/day by using this formula : (SiS*60*60*24)/1000 in R
The extracted data from the R are then correlated with the station obs with correlation around 0.97-0.98 therefore the data in this stage are in same units without any doubts.
The Image above shows the average annual sum of SIS for the period 1983-2005.
to create this I had to:
merge all the timesteps in one file using CDO.
then convert the values from Wm2 to MJ/m2 i used : (24*60*60)/1000000 and divided this by the number of years--> 23
I seriusly dont see any mistake in my calculation and the figures look realistic.
Your comment made me a bit confused :/ Could somebody create a figure which will show the average sum of SIS in MJ/m2/year based on SIS from CMSAF for 1983-2005?
Thanks.
Theo,
cdo does account for the scale_factor attribute in the netcdf-file, so this is no problem! Usually the output file of cdo after some arithmetics does not have a scale_factor, so the output can be used without apply any scale_factor.
You can check the scale_factor using ncdump -h file.nc or panoply. In both cases you can check the header of the netcdf-file and look for the scale_factor attribute.
Kind regards, Jörg
Thank you Jorg for the quick and clear answer.
This is very comforting indeed. I wasnt aware of the scale factor that need to be applied in R. My results are NOT affected because I used CDO for the initial manipulation of the data.
Useful knowledge added.
Greetings,
Theo
Just to clarify:
No scale factor has to be applied when using the R ncdf-package to open and access netcdf files, e.g., from the CM SAF. When reading the data into R a scale factor (if specified in the data) is automatically applied to the data. The user does not have to bother about applying any scaling factor. It seems this is differnt from the reading of netcdf data in IDL.
All R-scripts provided by CM SAF to work with the CM SAF data use the ncdf package to access the data, so no scale factor has to be applied!
Best regards,
Jörg
Dear Jörg,
many thanks for the clarification about the scaling factor. However, it was not clear to me if a missing value (NoData?) actually exists in the Meteosat SIS data. When I read the data with the IDL code I wrote I got -9999 outside the Meteosat disk.
Cheers
João
Hi,
just an example of the averaged SIS value between 1983 and 2006. Please note that the values are in W/m2 x 10.
I hope this makes sense.
Cheers
João