Temporal stability analysis for estimating spatial mean soil water storage and deep percolation in irrigated maize crops

• The temporal variation of components of the soil–water balance was evaluated.
• Soil water storage (SWS) of irrigated croplands was temporally stable.
• Deep percolation (DP) was high in the initial, development and senescence stages.
• Representative location could estimate the spatial means of SWS and DP.

The temporal stability of soil moisture in irrigated cropland may have profound implications for precision agriculture in arid regions. This study evaluated the temporal variation of components of the soil water balance in irrigated cropland in northwestern China and identified representative locations to reliably estimate the profiles (0–1 and 1–2 m) of spatial mean soil water storage (SWS) and deep percolation below a depth of 2 m. Soil water storages were determined with a neutron probe at 48 locations (31 and 17 in the northern and southern croplands, respectively) on a total of 18 occasions in 2011 and 2012. Crop evapotranspiration, SWS variations and deep percolation were analyzed during the maize growing seasons. Soil water storages in the northern and southern croplands were temporally stable. The location with the lowest standard deviation of relative differences could accurately estimate the spatial mean SWS with a high coefficient of determination (R2 > 0.91, P < 0.001) and prediction accuracy (PE > 0.76) and near-zero mean absolute relative error (MARE = 0). The most representative locations were different for the 0–1 and 1–2 m soil layers, but the location for the 0–1 m layer could also generally represent SWS in the 1–2 m layer in the northern and southern croplands. From 21 May to late September in both years, approximately 39 and 22% of the irrigation and rainwater were lost as deep percolation in the northern and southern croplands, respectively. Deep percolation at the most representative locations of spatial mean SWS for the 0–1 m layers could generally estimate the spatial mean deep percolation below a depth of 2 m. This study provides an alternative approach for estimating spatial mean SWS and deep percolation, which is essential for the rational management of scarce resources of irrigation water in a large arid region of northwestern China.