Study on Optimal Sampling Analysis of Soil Moisture at Field Scale for Remote Sensing Applications

Abstract

With the rapid development of soil moisture estimation techniques involving remote sensing technology, the sampling designs used in soil moisture research are very important. To estimate the rational sample number for measuring near-surface soil moisture (0–20 cm), a random combination method was used to study the relationship between the average measured soil moisture contents and the true values at given scales. Compared to classic statistics and stratified sampling, the random combination method easily obtained precision estimates from a small number of samples. Moreover, the random combination method was upscaled to further discuss the influence of the coefficient of variation and study-region scale on the rational sample numbers at different scales (2, 10, 20, 40, 80, and 160 m). The results showed that the rational sample numbers for measuring near-surface soil moisture at the 2, 10, 20, 40, 80, and 160 m scales were 2, 5, 5, 8, 20, and 42, respectively, under the relative error of 10% at the 95% confidence level. The rational sample numbers at different scales were proportional to the coefficient of variation and the regional scale.