Modeling long-term soil water dynamics in response to land-use change in a semi-arid area

Soil water content (SWC) is critical for the health and sustainability of the ecosystem in water limited areas. Characterized by low precipitation and deep groundwater, the Chinese Loess Plateau (CLP) has experienced intensive agricultural and revegetation activities for a period long enough to alter the state of SWC to the extent that can affect plant growth and water cycle. Knowledge of long-term SWC dynamics under land-use change is therefore important for optimal water management and vegetation restoration in this water-scarce region. This study simulated and analyzed long-term (1970-2060) variations in SWC in the top 4-m soil layer in response to land-use change on CLP using the Hydrus-1D model. Soil and plant parameters required for the model were validated using in situ SWC data for different land-use types - exotic grass (M. sativa), exotic shrub (C. korshinskii), arable crop (S. italica or V. radiata) and natural grass (S. bungeana). Five phases of SWC dynamics driven by land-use change were identified: i) fluctuation of SWC with rainfall during arable cropping period; ii) rapid drop of SWC after exotic plant establishment; iii) period of no change for which SWC remained low and stable; iv) gradual recovery of SWC after conversion of exotic vegetation into natural grass; and v) fluctuation of SWC with rainfall under natural grass. Particularly, SWC decreased rapidly with the planting of M. sativa and C. korshinskii and water depletion depth increased with plant growth after revegetation. Dry soil layer (DSL) began to develop below 1-m after 3 years of growth of M. sativa or C. korshinskii. Also, SWC at depth below 1 m remained low and unavailable for plant growth. Compared with arable cropland, SWC in the 1-4 m soil layer decreased by 33% under both exotic grass and shrub during vegetation restoration period. However, it took 7-8 years for soil water to recover in the 0-4 m soil layer after 13-year (2004-2016) exotic grass or shrub was converted to natural grass. This suggested that DSL in the 4-m soil depth can fully recover by changing vegetation type. We concluded that planting M. sativa and C. korshinskii on arable cropland could enhance DSL formation in deep soil profile, but can gradually disappear upon land-use change. The results provided useful information for the prediction of long-term dynamics of soil water with land-use change. This is critical for the management of soil water in water limited areas.