Validation of the data products
Currently no real validation of SO2 slant column or vertical column
densities has been undertaken as yet. A problem is that there are few
ground-based measurements of SO2 that are appropriate for validation. It is
planned to compare the results presented here with data from
other satellite instruments. Furthermore, there is a close link with the
NOVAC project, which
intends to set up a network of ground-based stations for monitoring volcanic
emissions of, among others, SO2.
In setting up and fine-tuning the
SO2 slant column (SCD) retrieval with the
DOAS method, the fitting results have been studied so as to optimise the
settings for this retrieval. The slant columns have been compared to other
results in an ad-hoc way for this, and in a few case studies. The SO2 SCD
values can thus be considered to be fairly accurate, but a more rigorous
quantitative validation is necessary.
The SO2 vertical column data (VCD) presented on the website are computed
air-mass factor using look-up tables
and an assumed height of the SO2 plume. This approach provides only a
rough estimate of the value of the VCD. Hence, the values given for the VCD
should not be taken as hard numbers.
The criteria for exceptional SO2 concentrations
for the SO2 data have been set up by examining the results of the retrieval
and analysis processes. The results of notification events found by the processing
are the source for the continuous (qualitative) validation of the notification
In the near-real time service, the notification events are listed on dedicated web
pages. At the bottom of these dedicated web pages there is room for adding
notes and remarks, e.g. to link different notification to one and the same SO2
event, to link to external information, etc. This has to be done by hand and
is therefore done on an an-hoc basis.
In the archive service, the processing provides an overview of notification cases
for each month in the form of a table with some information on each SO2 peak
event found. Most of this table is generated automatically, but there is
room to add remarks and other info by hand to each table.
We have observed that daily and monthly SCIAMACHY data (from January to June 2009; from January to June 2010) have abnormally strong measurements;
see monthly plot of March 2010
We reprocessing these data. Do not hesitate to contact us if you want to know when reprocess will be finished.