Table 8. First five predictors that were highly significant for RFR (based on “IncNodePurity” importance measure) and MLR analysis.

Model	Rank	Sand	Silt	Clay	CEC	SOC	Nitrogen
MLR	1	june_SWIR2	june_SWIR2	june_NIR	june_SWIR2	Elevation	Elevation
	2	june_green	June_RI	June_RI	May_RI	prep	March_NDVI
	3	June_CI	may_red	may_blue	may_RE	march_NIR	march_NIR
	4	may_green	june_red	June_SI	June_BI	March_NDVI	march_green
	5	April_HI	June_BI	June_CI	june_red	june_SWIR1	March_CI
RFR	1	june_SWIR2	June_RI	june_NIR	june_SWIR2	june_red	june_NIR
	2	may_NIR	May_SI	June_RI	june_blue	june_NIR	June_SI
	3	june_green	june_SWIR1	june_blue	May_RI	Elevation	Elevation
	4	May_SI	june_SWIR2	june_SWIR1	March_NDVI	June_SI	march_green
	5	may_green	May_CI	temp	june_red	June_BI	may_red

The names of the spectral predictors (see Table 2) here are a concatenation of the month of satellite acquisition and a spectral channel or indice. For example, “May_BI” represents the brightness index calculated from the May RapidEye image. prep: precipitation, temp: temperature.