Cross-classes domain inference with network sampling for natural resource inventory

There are two distinct types of domains, design- and cross-classes domains , with the former extensively studied under topic small-area estimation. In natural resource inventory, however, most classes listed in condition tables national inventory programs characterized as such vegetation type, productivity class, age class. To date, challenges remain active for inventorying because these usually unknown sampling frame spatial distribution result that inference relies on population-level opposed to domain-level sampling. Multiple noteworthy: (1) efficient strategies difficult develop little priori information about target domain; (2) domain a sample designed population, so within-domain sizes could be too small support precise estimation; (3) increasing size population does not ensure an increase domain, actual remains highly uncertain, particularly domains. this paper, we introduce design-based generalized systematic adaptive cluster (GSACS) Design-unbiased Hansen-Hurwitz Horvitz-Thompson estimators derived totals compared within GSACS (SYS). Comprehensive Monte Carlo simulations show unbiased equally efficient, whereas latter outperforms supporting one; SYS is special case while terms increased efficiency reduced intensity; variance estimator design-unbiased single sample; (4) rules-of-thumb summarized respect design effect improve precision. Because mini analogous rare variable, alternative network procedures also readily available ? Generalized devised Unbiased estimation Analytical empirical comparisons made provides estimates identical SS using observations fewer order 20%–40% Network including