To the list of courses || GAT2018 || To the theme || Estonian

Exercise 3675. Points 4, theme: Non-parametric Correlation

Open exercise
  1. Numerically how strong, which direction and statistically how significant is the Spearman rho correlation between cover of coniferous forests (land cover ="okasmets") and values in Landsat TM 8th channel (grey tones) in the attached file?
  2. Is higher coniferous cover coinciding with darker Landsat image?
  3. What can be the reasons of such a weak correlation?
  4. Why is Pearson R not suitable for measuring this correlation?


Solution in the SDC.

Filter the observations in coniferous forests according to the land cover unit using Excel.

Copy the cover values and TM8 values to the Non-spatial correlation in the SDC.

Solution in Statsoft Statistica

Import the Excel file to Statistica. Select Statistics → Nonparametrics → Correlations. Select needed cases using Select casesmaakate/land cover = "okasmets".

Open the panel Detailed report from the calculation results.

Be aware that the pixel value in a remote sensing image indicates the amount of rediation, not darkeness of the site.
Log in to send your results and to see the expected answer and responses from other students.