In the monitoring and assessment of landscapes, randomly located sampling plots are often used to minimize sampling bias and enable inference to larger landscape units. In particular, spatially balanced random designs are more robust to spatial autocorrelation and therefore produce data usable for answering multiple management questions. The Generalized Random Tessellation Stratified (GRTS) approach is one technique for creating spatially balanced random designs in natural resources and has been widely adopted by multi-scale terrestrial and aquatic monitoring programs. While the parameters for a GRTS design are relatively simple to specify with stratification polygons and per-stratum sample sizes, generating a design from those inputs has historically required the ability to code using the R package spsurvey. As a result, technical skills have been a bottleneck for this statistical approach in resource monitoring programs.
To remove this technological barrier, we have developed an R package and a web application, the “Balanced Design Tool”, which uses it to create GRTS designs through a graphical interface. The tool walks the user through the process of specifying their design, prompting them for the critical components, then generates a random, spatially balanced set of points as well as the necessary files and R script to reproduce the design at a later date.
Results/Conclusions
The R package, sample.design, is publicly available and has been used internally to generate dozens of designs for Bureau of Land Management short-term and multi-year monitoring efforts, including sage grouse habitat assessment and land use plan effectiveness monitoring. The web tool has been maintained for over five years and has been accessed by hundreds of users from federal, non-profit, and private entities to generate designs for a wide range of natural resource monitoring and inventory use cases. The latest version has brought significant improvements to stability, reproducibility, and usability and has added support for additional approaches to effort allocation.