Choosing an appropriate sample size and sample spacing is fundamental to the success of all soil studies; however, sampling strategies are rarely justified in research papers and presentations. Indeed, in many studies, decisions relating to the sampling strategy appear to be based on a combination of subjective opinion, repeating previously used sampling designs, and available resources. When the aim of a study is to estimate the spatial mean of a soil property the sample size should be large enough to meet the accuracy objectives, but not so large that unnecessary (and costly) additional samples are collected. Similarly, sample spacing should be large enough to minimize correlation among samples (i.e., maximizing sample independence and information), but close enough to constrain costs. Here, we present a simple tool that provides quantitative estimates of the sample size and sample spacing required to accurately estimate the spatial mean value of soil temperature, soil moisture, or related properties at the 1 ha scale. The model is based on empirical measurements of the spatial structure of variation in soil temperature and soil moisture at 60 representative NEON candidate sites throughout the USA, which were used to create semivariograms for each site.
Results/Conclusions
Based on the semivariograms from the 60 sites, the proportion of sites where samples spaced x meters apart were at least y % independent increased as sample spacing increased and sample independence decreased. For example, samples spaced 5 m apart were 100 % spatially independent at only ~10 % of sites, whereas samples spaced 20 m apart were 80 % spatially independent at ~75 % of sites. This information can be used to inform sample spacing decisions. Using the expected sample independence estimated above, as well as the user’s desired accuracy and confidence, the tool calculates the necessary sample size based on the semivariogram sill and mean soil temperature and moisture at each site. For example, at 75% of sites 8 samples would be sufficient to measure soil temperature to within 10% of the spatial mean with 90% confidence when samples were 75% independent. In contrast, based on the same parameters approximately 150 samples would be required to estimate the spatial mean of soil moisture. This tool provides a robust and quantitative basis for choosing a sampling strategy to estimate the spatial mean of soil properties at the 1 ha scale.