Simulation-based assessment of sampling strategies for large-area biomass estimation using wall-to-wall and partial coverage airborne laser scanning surveys

Abstract

Abstract Airborne laser scanning (ALS) has been demonstrated to be an excellent source of auxiliary information for increasing the precision of estimating stand-level attributes in forest inventories. It has also been proposed to use ALS for estimating biomass and carbon stocks under the United Nations Collaborative Program on Reduced Emissions from Deforestation and Forest Degradation in Developing Countries (UN-REDD). The benefits of REDD depend among other facts on the cost-efficiency of the carbon accounting systems, which should be economically feasible and highly accurate. Acquiring full-coverage ALS data would provide highly accurate estimates but might be too expensive for limited inventory budgets. As an alternative, the ALS data might be collected as a sample by acquiring data from a portion of the area of interest. However, in surveys involving complex multi-phase and multi-stage systematic sampling designs, the efficiency of ALS-based estimates is hampered by the ability of estimating the sampling variability correctly. It has been demonstrated recently that the precision of such complex analytical estimators may be largely underestimated. In order to make an informed decision, simulated sampling from artificial populations generated from empirical data may provide a means for assessing the cost-efficiency of various sampling strategies when analytical approaches fail. This study presents a simulation-based assessment of sampling strategies employing ALS with focus on large-area (27,400 km 2 ) biomass estimation. Simulated sampling mimicking the two contrasting cases “wall-to-wall” and two-phase ALS-aided surveys is exemplified using Norwegian National Forest Inventory data for creating an artificial population. The main results indicated that (1) the gain in precision (10%) when using “wall-to-wall” ALS data may not be worth the very high inventory costs, (2) using variance estimators based on higher-order successive differences produced correct confidence intervals for two-phase systematic sampling, and (3) two-phase ALS-aided systematic surveys are cost-efficient solutions for large-area biomass estimation.