We consider two-phase sampling schemes where one component of the auxiliary information is known in every point ("wall-to wall") and a second component is available only in the large sample of the first phase, whereas the second phase yields a sub-sample with the terrestrial inventory data based on general tree inclusion probabilities. We propose a generalized version of the classical two-phase regression estimator, for global and local estimation and derive its asymptotic design-based variance. Cluster and two-stage sampling procedures are also considered.