Aims It is more difficult to determine soil moisture in root zones (RSM) than soil moisture in the top-most layer (TSM). The relationship between TSM and RSM based on point A is useful for acquiring RSM at point B from its TSM if the relationship between TSM and RSM is general. The general relationships established so far have been based on only one crop or one ecosystem. Our aims are to determine the general relationship between TSM and RSM in China over various crops and ecosystems and to explore the effects of ecosystem, soil property, precipitation, vegetation, soil thickness and the order of magnitude of soil moisture on the relationship.
Methods We used 3 437 pairs of TSM and DSM data from 31 stations at 109 observation sites. The data represent all the terrestrial ecosystems, including forest, grass, agriculture, desert and mire, in 2006 from the China Ecosystem Research Network. We used linear regression in mathematics and effect analysis in physics to study the general relationship between TSM and RSM and to determine the influence of ecosystem, soil property, precipitation, vegetation, soil thickness, the order of magnitude of soil moisture on the relationship. The data are divided into calibration and verification groups for producing the relationship.
Important findings There is generally a linear relationship between TSM and RSM. The coefficient of determination (R2) between RSM and TSM is high for forest and mire ecosystems (R2 > 0.79 for both calibration and verification), intermediate for agricultural (R2 > 0.80 for calibration and 0.70 for verification) and grass ecosystems (R2 > 0.80 for calibration and 0.50 for verification) and low for desert ecosystem (R2 = 0.62 for calibration and 0.49 for verification). The effect of soil properties on the relationship is similar to that of ecosystem, with mire ecosystem corresponding to meadow soil with white pulp, desert ecosystem corresponding to aeolian soil, grass ecosystem corresponding to sandy-loam and sandy soil, forest ecosystem corresponding to loam and agricultural ecosystem corresponding to various soil types. The spatial variability of the parameters for the relationship is very high for soil moisture within semi-humid, semi-arid and arid zones and low for humid zone. The relationship is similar among very-humid zone, loam and forest ecosystem. The effect of vegetation on the relationship can be divided into four categories: poor relationship due to the influence of vegetation itself, poor regional relationship due to the difference between the location sites of vegetation, good relationship at single point and good relationship over a region. It is not recommended to derive the RSM from TSM based on the relationship for the first two categories. It is suitable to use the relationship only at single point for the third category. And it is practicable to use the regional relationship to derive RSM from TSM for the fourth category. Correlation is higher between TSM and RSM where the soil layer is thicker. The R2 can remain as high as 0.79 when the soil
layer is as thick as 100 cm. By scaling the original soil moisture data with maximum value, it is found there is no effect of the order of the magnitude of soil moisture on the relationship.