Fig. 1

Framework of the two-part modeling procedure, using: (I) combined field data (FD) variables and remote sensing (RS) variables – to uncover the relatedness between the FD and the RS variables using Redundancy Analysis (RDA); (II) RS variables alone – to predict the probability of occurrence of both Erigeron spp. and Solidago spp. at the pixel- and site-levels; in both (I) and (II), three machine learning techniques are employed and compared: Gradient Boosting Machines (GBM), Support Vector Machines (SVM), and Random Forest (RF).