Integrating remote sensing and past inventory data under the new annual design of the Swiss National Forest Inventory using three-phase design-based regression estimation

Abstract

In 2009, the Swiss National Forest Inventory (NFI) turned from a periodic into an annual measurement design in which only one-ninth of the overall sample of permanent plots is measured every year. The reduction in sample size due to the implementation of the annual design results in an unacceptably large increase in variance when using the standard simple random sampling estimator. Thus, a flexible estimation procedure using two- and three-phase regression estimators is presented with a special focus on utilizing updating techniques to account for disturbances and growth and is applied to the second and third Swiss NFIs. The first phase consists of a dense sample of systematically distributed plots on a 500 m × 500 m grid for which auxiliary variables are obtained through the interpretation of aerial photographs. The second phase is an eightfold looser subgrid with terrestrial plot data collected from the past inventory, and the third and final phase consists of the three most recent annual subgrids with the current state of the target variable (stem volume). The proposed three-phase estimators reduce the increase in variance from 294% to 145% compared with the estimator based on the full periodic sample while remaining unbiased.

Countrywide Stereo-Image Matching for Updating Digital Surface Models in the Framework of the Swiss National Forest Inventory

“…In the framework of the Swiss National Forest Inventory, the data is used as auxiliary information to post-stratify forest areas according to canopy height for the estimation of different forest characteristics, see also [8,44]. This allows for a further reduction of estimation errors, especially when it comes to small area estimations [1,45].…”

“…Forest characteristics are traditionally estimated in forest inventories based on sample plots. The combination of design-based sample inventories and remote sensing data, e.g., satellite data or airborne laser scanning data, reduces estimation errors and makes field-based inventories more efficient [1]. Stereo optical imagery or laser scanning data have tremendous potential for obtaining area-wide three-dimensional data over several thousands of hectares.…”

“…As vegetation height is the most important proxy for a wide range of forestry variables, we focus in our analysis on the accuracy and agreement of modeled tree and canopy heights. The main objectives of the present study are: (1) to demonstrate the highly automated workflow to calculate a nationwide surface model out of regular national aerial images; (2) to evaluate the accuracy and completeness of this digital surface model, which is a prerequisite for its subsequent usage; and (3) to present the potential of this data for National Forest Inventories.…”

Updating Canada’s National Forest Inventory with multiple imputations of missing contemporary data

“…It also appears that the combination of T 1 differences and relatively weak contemporaneous and temporal correlations in the data prevented us from achieving practically important improvement in precision. In forest inventories, the MI approach works best for undisturbed plots where the temporal correlation between the state at two points in time is generally strong for shorter periods (years) between observations, declining as the number of years between observations increases (Massey et al 2014, McRoberts et al 2016. In this regard, a stand-replacing disturbance event acts as a 'correlation-breaker' .…”

A Double-Sampling Extension of the German National Forest Inventory for Design-Based Small Area Estimation on Forest District Levels

“…An example can be found in von Lüpke et al [5] who illustrated an extension of the existing two-phase FDI of Lower Saxony to a three-phase design that uses updates of past inventory data as additional auxiliary information and allows for a significant reduction of the terrestrial sample size in intermediate inventories. Another example is Massey et al [19] who developed a triple-sampling extension based on the ideas of Mandallaz [20] for the Swiss NFI that can significantly reduce the increase in estimation uncertainty caused by the new annual inventory design.…”