Hello Everyone,
Besides calculating Pearson's r (correlation coefficient), the coefficient of determination (R^2), and Root Mean Square Error (RMSE), what are some additional statistical/validation techniques for examining regressions/algorithms derived from plant biomass (dry wt / m2) and vegetation indices calculated from remote sensing data?
Additionally, when evaluating two plant biomass map products derived from different biomass algorithm equations (i.e. one map was created from using imagery and field samples collected in mid 2006 and the other map was created from imagery and field samples collected in late 2006; imagery used in both cases was Hyperspectral; sensor= AISA Eagle), what is the best approach for comparing and contrasting the two maps? How do I compare the two biomass maps and determine which one best captures marsh primary production (seeing that both maps were derived from algorithms yielding R^2 values around 0.75)? Besides performing the change detection technique, in order to look at the difference among the two biomass maps, are there any statistical approaches that could be use in evaluating the differences among these maps?
Any help would be greatly appreciated. Thank you!
If I have understood you correctly... one possibility that you could explore is to use a geostatistical approach. The links below may help you get some ideas
http://streaming.stat.iastate.edu/~stat506/notes/crossvalidateexamp.pdf
http://meteo.edu.vn/DATA/Books/Hydrological%20Modelling%20and%20the%20Water%20Cycle/2.1%20Geostatistical%20Tools%20for%20Validation%20of%20Satellite%20and%20NWP%20Model%20Rainfall%20Estimates.pdf
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics