The Southern Annual Forest inventory System (SAFIS) is in various stages of
implementation in 7 of the 13 southern states serviced by the Southern Res
earch Station. The SAFIS design is an interpenetrating design where the Is,
units (1/6 acre plots) are divided into k = 5 panels, each panel containin
g m = n/k units. Panel 1 plots are measured in year 1, panel 2 in year 2, e
tc., such that all plots have been visited by the end of year 5. The panel
cycle is repeated into perpetuity. Each panel, in effect, is a 5-year perio
dic survey with complete overlap of sample units. Numerous estimation schem
es are possible, and we explore five possible options. The five options are
(1) use existing periodic inventory programs to produce 5-year survey esti
mates by adjusting all five panels to a common year, (2) analyze each annua
l panel independently, (3) produce 5-year estimates by combining the five p
anel estimates by varying the weight given to each panel, (4) base inventor
y estimates on mixed estimation where actual and predicted values are combi
ned, and (5) use imputation techniques such that unmeasured plots are fille
d in with imputed plots. A two-phase method for forest area estimation that
uses the known map marginals from a thematic map is presented as an altern
ative to photo interpretation-based estimates.