Dear all,
We had collected the ground data within forests in the year 2012, and developed the regression equation/model by correlating the ground biomass and red band of February month 2012, which was significant with good R2 value and predicted the biomass map of our study area for the year 2012.
Now, we want to apply the same regression equation/model which was developed in 2012 on the red band of February month 2017 data and want to predicate the biomass map of our study area and want to perform change detection analysis to know the changes occurred within last 5 years.
My question is, does it scientifically valid to use 2012 regression equation/model to apply on 2017 satellite image….?
I will highly appreciate for your valuable suggestions and comments.
Regards-Prashant
Dear Dr. Sudipta,
Thank you for your kind suggestion. For 2012 we developed model by using landsat data after undergoing atmospheric & radiometric corrections and fileds data. And for 2017 we want to run on Landsat data and we will go for atmospheric & radiometric correction, but here we want to use 2012 model to predicate biomass.
Dear Dr. Sudipta,
In our case we used Landsat-7 data in 2012, and we selected the points for modeling were the pixel value is present in image because Landasat-7 had serious data gap problem. And later we filled the data gap values by using back date datasets of same time and predicted the biomass values for our study area.
But as per your suggestion what I understood and the bottom-line is the 2012 model cannot be used for 2017 Landsat-8 because both are different sensors and platforms and also the geometry is differ.
We will try to collect the ground data for 2017 or may be for next year in 2018 and try to carry further research.
Thank you very much for your valuable suggestions and comments.
Regards-PTP
- Badges
- Science topic
- Similar topics
- Computer Science
- Images