Dear experts,
I have snow parameters calculated using MODIS snow products and I'm trying to validation with ERA5-Land monthly averaged - ECMWF climate reanalysis images using Pearson correlation. MODIS snow products have spatial resolution of 500 m while ERA-5 has 0.1 degree (11.1 km) of spatial resolution, thus terrible for me to correlation, even resizing back to 500m doesn't make sense. Could one of you be so kind for any advices, please?
Dear Swun,
Please clarify your question, because the title of your message asks about 'calibration' while the text of the message asks about 'validation'. These are entirely different concepts!
* In general, instruments do not directly measure the property of interest but a proxy (be it a current, a voltage, a digital number or a reading on a dial), i.e., they deliver a value that is only indirectly related to the information being sought. Hence, instruments need to be 'calibrated': this is a process whereby particular values generated by the instrument are associated with specific values of the variable of interest. For instance, the height of a column of mercury in a thermometer is associated with a temperature through the graded scale next to the column. Note also that the expression 'data calibration' is meaningless: only instruments can be calibrated.
* By contrast, the process of 'validation' consists is determining the degree of veracity, the accuracy or the reliability of the measurements. In other words, once measurements have been acquired, to what extent can we trust them? In this latter case, one would typically rely on a different source of information, hopefully even more accurate or reliable (such as provided by national standards organizations) to compare measurements from multiple instruments or using different methods of analysis.
In your message, it is not clear whether you are trying to calibrate the MODIS instrument for snow parameters (i.e., find a relation between actually measured radiances in space and actually observed snow cover observations on the ground), or evaluating whether the 'ERA5-Land monthly averaged' product generated by ECMWF is reliable. To calibrate MODIS radiance data in terms of snow cover without having access to field observations, you might want to compare those measurements with data from other satellites, perhaps at a much finer spatial scale, for instance. The ECMWF products are not going to be very useful for that purpose. If, on the other hand, you wish to determine the reliability of a snow cover field contained in the ERA-Land product, then you need to aggregate the MODIS data to match the spatial resolution of the latter and investigate to what extent the two estimates concur.
Best regards, Michel.
Dear sir, your last part of the reply is the answer I was looking for. Much appreciations to you.
PS. Again, thank you for your clarification. I was having unclear between validation and calibration while I posted this. Thank you so much.
With due respect,
Swun
- Badges
- Science topic
- Similar topics
- Mathematics