CM SAF 2012: difference between SIS data

Hi,

I would like to ask your help to explain our result derived from METEOSAT MFG and NOAA global surface incoming short wave radiation data.

We have used the SIS from METEOSAT/MFG monthly mean and SIS from polar orbit monthly mean data between 2001 and 2005 for Hungary.

We derived the difference of these two kind global radiation maps for that period. At first we made the difference between two set of SIS data (MFG minus NOAA), after it mean was calculated for the whole country.

As you can see in the enclosed image the difference between the SIS data derived from MFG and NOAA seems to be an increasing trend during the 5 year period. We have compared these values with observed surface measurements to check and investigate this result. A very good correspondence be seen between the Meteosat and the ground measurement, but if we compare it with NOAA data we can see increasing difference (like a trend) during the examined period.

We have thought it might be caused by the NOAA path delay. But we checked it in our dataset, and we did not see remarkable delay in the pathes.
Thanks a lot!

Regards, Ildikó

Re: difference between SIS data

by Jörg Trentmann - Monday, 17 March 2014, 11:06 AM

Dear Ildikó,

thanks a lot for sharing your results on your comparison between the CM SAF solar surface radiation data sets and their evaluation using surface observations.

I understand you are using the CM SAF data sets (not the operational products) based on Meteosat (i..e., http://dx.doi.org/10.5676/EUM_SAF_CM/RAD_MVIRI/V001) and AVHRR (aka the CLARA data set: http://dx.doi.org/10.5676/EUM_SAF_CM/CLARA_AVHRR/V001). So far these data sets have not yet been compared quantitatively and your contribution and your results are very valuable.

In general we find that the MVIRI-based surface radiation data set tends to overestimate the surface radiation compared to surface measurements, while the AVHRR-based data set tends to underestimate the surface radiation. It seems this is consistent with your result, i.e., the positive difference between MVIRI - AVHRR.

Regarding a possible trend in the differences between these surface radiation data sets (satellite- and ground-based) the five-year period might be too short to come to a firm conclusion. I agree that the differences in 2004 and 2005 are larger than in 2001 and 2002 between the two CM SAF data sets, and I do not have a clear explanation for this.

In a previous work we did look at the long-term trends in both data sets and found them to be very comparable, also in Hungary (see http://presentations.copernicus.org/EGU2013-10394_presentation.pptx, slide 18). Also we did compare both data sets with the surface observations from Budapest, that are availble in the GEBA data base and found no major trend in the bias after 2000. As in your evaluation, the MVIRI-based data set shows a lower difference to the surface measurements than the CLARA data set at Budapest. Of course, I am happy to share these results with you, just let me know!

I would be very interested in more details on your validation using surface observations. Maybe you could provide some figures that show the performance of the CM SAF data sets compared to the surface observations.

Kind regards,
Jörg

Re: difference between SIS data

by Ildikó Dobi Wantuch - Monday, 17 March 2014, 2:51 PM

Dear Jörg and Reto,

Thank you for your reply! We used monthly mean for 8 surface stations (pixels) from different part of Hungary between 2001 and 2010, so we have only five years paralell data set for all resources: Observed global radiation and SIS (climate data, MVIRI and AVHRR sensors). I attached a times series for a single station (Budapest), the others are similar. It is our preliminary results for the EUMETSAT conference.

Thanks a lot.

Regards, Ildikó

Re: difference between SIS data

by Jörg Trentmann - Monday, 17 March 2014, 4:02 PM

Dear Ildikó,

thanks for providing plot of the time series of the validation of the CM SAF data sets for time series from Budapest. This results agrees very well to our analysis using the monthly mean data available in the GEBA data set from Budapest. Please find enclosed the corresponding plots for the CLARA and the MVIRI-based data sets.

Unfortunately these plots are a bit messy (designed for internal use..), but on the second figure you can see the multi-year annual cylce from the surface and the satellite measurement. While the MVIRI-based data fits nicely to the surface observations, there is an underestimation of the CLARA data set, in particular during summer, comparable to your results.

In our analysis a temporal trend in the bias after 2001 is not obvious (first figure, middle row). The CLARA data set includes many missing data, so an interpretation can be difficult; the time series of the bias of the MVIRI-based data set suggests something suspicious happening around 1998, when the bias seems to drop suddenly. You can find more information on the GEBA data and their application for validation in the article by Sanchez-Lorenzo et al., 2013.

We are currently preparing for the next release of the Meteosat-based surface radiation data set, which will be released in fall 2014 covering the time period from 1983 to 2013. Hopefully this data set will provide a consistent and homogeneous data record of the surface solar radiation.

Kind regards,
Jörg

Attachment Budapest_CLARAValidation_1982-2009.jpg

Attachment Budapest_MVIRIValidation_1983-2005.jpg