Contributions to the Theory of Environmental Sampling

Abstract

Environmental monitoring plays a crucial role in guiding climate change and conservation policy decisions. To obtain reliable insights from environmental populations, it is essential to adopt probability sampling. Furthermore, the availability of auxiliary variables can greatly enhance the quality by reducing estimator variability. Auxiliary information can be used in different ways in a sampling design. Some designs aim to satisfy the balancing equation, i.e. selecting samples where the sample means of the auxiliary variables equal the population means. Other designs are constructed in an attempt to obtain samples well-spread, or spatially balanced, in auxiliary space, creating the sample as a miniature of the population. Paper III provides an improvement of an existing design, making it possible to increase the average spread of the sample. In Paper IV, a novel metric is introduced to assess a design’s capability to yield spatially balanced samples. Papers II and V introduce sampling designs for different types of populations one might encounter in nature. The variant of adaptive cluster sampling developed in Paper II facilitates the study of rare and clustered populations, utilizing circular plot shapes popular among practitioners. Paper V addresses the sampling of linear objects like storm-felled trees, employing aerial photographs from drones in the data collection processes. When data are gathered from multiple surveys, various methods exist to consolidate results. A common approach involves constructing a linear combination weighted by variances. Paper I introduces a novel estimator that employs a linear combination, particularly valuable when a correlation is suspected between the estimator and the variance estimator – a frequently encountered scenario in studies involving environmental populations. In conclusion, this thesis contributes to the field of environmental monitoring by emphasizing the critical role of probability sampling, utilization of auxiliary variables, and introducing innovative sampling designs tailored to the intricacies of environmental populations.