Sampling frequency significantly influenced surface soil moisture dynamics but not its prediction accuracy in an arid mountain forest

Abstract

Study region

a typical arid mountain region of northwestern China.

Study focus

Soil water content (SWC) is the key factor regulating patchy vegetation patterns in arid/semiarid areas. However, accurately determining the regional SWC status remains a challenge due to the time and labor-intensive nature of manual sampling methods. Furthermore, a thorough understanding of the influence of different sampling frequencies (SFs) on SWC spatio-temporal dynamics in arid mountain forests is lacking.

New hydrological insights for the region

SFs had a distinct effect on mean SWC, and temporal stability characteristics under lower (15–45 days, LSFs) and higher SFs (within 7 days, HSFs). SF influenced mean SWC for 0–20 cm under HSFs only but had a significant influence for 0–20 and 40–60 cm under LSFs. SF did not influence Spearman’s rank correlation coefficient (r_s) for the 0–20 cm layer, but had a significant effect on the standard deviation of mean relative difference (SDRD) under HSFs; however, SF had a significant effect on r_s for the deep layer (80–100 cm), but did not influence SDRD under LSFs. Although the number of representative locations (RLs) was significantly higher under HSFs than LSFs, no RLs were found at 100–120 cm. The mean SWC for all soil depths except 40–100 cm under HSFs was predicted accurately for each SF. This indicated that HSFs were not conducive to the identification of deep soil RLs, and had a significant impact on the prediction accuracy of SWC for deep layers. LSFs were not conducive to the identification of surface soil RLs but they can accurately estimate mean SWC, and prediction accuracy improved when SF was reduced. These results have important implications for optimizing water sampling schemes and promoting sustainable ecological development in water-deficient regions.