Best mean air temperature data equations

     For the standard deviations parameters of the coefficients (first order coefficient and 
     intercept) the equations showing the smallest values were the regressions done on the data 
     including all stations. 

     The standard error of estimate was smallest for the regressions done on the data excluding 
     the Sabana stations (SA and SF). 

     Although all types of regressions did best at least once for the correlation coefficient R 
     and the Rsqr, most of the best regressions resulted from running the linear regression on 
     the non-closed-canopy forest stations data (i.e excluding SF). 

     The power parameter was almost equally better for equations including all 
     non-closed-canopy forest data and for equations including all stations data.

     The best PRESS values resulted from the linear regressions performed on the data excluding 
     the Sabana stations.