Which is the more suitable performance score? Correlation coefficient , Root Mean Square Error or others?
Which is the best benchmark dataset? In situ observations, modelled data or others?
How to deal with the differences in the spatial scale between satellite and ground measurements?
In the scientific community there is not a consensus on that. Each study employed a different methods, sometimes with some limitations in terms of sample size, temporal co-location, ... This poses clear problems for the intercomparison between their results and for an objective evaluation of the soil moisture products.
Therefore, any comments, suggestions, guidelines are really welcome.
BTW, a workshop on that will be held in Frascati next July: http://www.soil-moisture-workshop-2013.com/.
Therefore, comments before that date are very welcome!
Wow！ you raised very important questions which we face very often while trying to judge the credibility of our results. I am not fit to answer any of the questions you raised because I am just a PhD student. But in order to invite the attention of those whose opinion really matters, I am expressing my comments:
1) I think some statistician may give more professional opinion about suitable performance score, Correlation coefficient , Root Mean Square Error or others etc.
2) I think benchmark dataset criteria depends upon the its usage. The usage decides which one is better. In some cases in situ might be better and in some cases modeled might be better. But if insitu data is very dense and represents more than 70% of area of interest, then i think in situ should be preferred over modeled data.
3) In my view, if study area doesnt have so many features and mostly uniform then by using interpolation or extrapolation techniques on insitu data, we can cover the whole area to compare it with satellite data. In case of so multiple features, ROIs should be representative of insitu data to be compared with satellite data.
My answers are naive but just in order to invite the attention of right persons for the right answers, I just expressed my limited knowledge here.
At the following link you can find "best practices" for satellite soil moisture validation:
http://congrexprojects.com/docs/default-source/13c13_docs/best-practice.pdf?sfvrsn=0
... and also have a look to this paper recently published that addresses this issue:
https://www.researchgate.net/publication/256492261_Clarifications_on_the_Comparison_between_SMOS_VUA_ASCAT_and_ECMWF_soil_moisture_products_over_four_watersheds_in_U.S?ev=prf_pub
Article Clarifications on the "Comparison between SMOS, VUA, ASCAT, ...
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing