An analysis of stratification in the estimation of landscape metrics at national level

Monitoring landscapes at national level is essential for understanding environmental changes. However, collecting wall-to-wall data on a large scale is very expensive and time-consuming. Sampling methods are often used to address this challenge. This study investigates how stratified sampling can improve the efficiency of national landscape monitoring. Using data from Sweden’s National Inventory of Landscapes (NILS), three stratification schemes were assessed in conjunction with two sampling densities (2.5% and 5%) and two allocation methods (Neyman and optimal allocations). Simulations were used to estimate three landscape metrics—Shannon’s diversity index, forest edge length, and number of forest patches. Larger sample sizes and optimal allocation led to better estimates. We discovered that some metrics, like forest edge length and the number of forest patches, were strongly related to the forest area. This means that simple indicators like forest cover can be used in post-stratification schemes. This can offer more flexibility, especially in landscapes that change over time. In conclusion, these findings can help governments and organizations design better monitoring programs that are both scientifically robust and cost-effective.