Soil hydrology and crop evapotranspiration modeling in a dry agricultural region of the Tibet Plateau

Vapor flow plays major role in soil–atmosphere water exchange in arid regions, and can be partly driven by airflow, but its impact is often neglected. In this study, a two-year field experiment was conducted in a dry agricultural region on the Tibet Plateau (TP) to investigate the effect of airflow on soil hydro–thermal dynamics and evapotranspiration modeling under three cultivation patterns: ridge-furrow planting with black-film mulching (RM), flat planting with black-film mulching (FM), and flat planting with no mulching (FN). An airflow-incorporated, based on Philip and de Vries (PdV) model, STEMMUS (Simultaneous Transfer of Energy, Mass and Momentum in Unsaturated Soil) was adopted. Considering objective’s (Lycium barbarum L.) sparse canopy, excluding Penman-Monteith (P-M) algorithm which already employed in STEMMUS, Shuttleworth-Wallace (S-W) model was incorporated into STEMMUS model to simulate evapotranspiration rate. Validation results showed that STEMMUS reliably captured the behaviors of observed soil moisture, soil temperature, and evapotranspiration (index of agreement d = 0.4, 0.6 and 0.5 for soil moisture under FN, FM and RM; 0.9 for soil temperature under three treatments; 0.6, 0.6 and 0.8 for evapotranspiration under FN, FM and RM; 0.6, 0.5 and 0.5 for evaporation under FN, FM and RM). Incorporating airflow extended the 0-1 m soil profile temperature modeling precision (d value improved 1%), led to the maximum 5% gap of soil moisture at 20 cm depth, and 3.7 mm d^-1 gap of daily evapotranspiration compared to pattern without airflow under non-mulched treatment. However, the impacts of airflow are weak under mulching treatments (the gaps between including/excluding airflow modeling were within 0.1% for soil moisture, 0.1 ℃ for soil temperature and 0.1 mm d^-1 for evapotranspiration). Furthermore, the effect of coupling airflow became significant when water inputs (precipitation/irrigation) were higher than 18 mm. Incorporating S-W model successfully improved evapotranspiration modeling precision, with d values increased by 0.5% and 1% for FM and FN respectively in evapotranspiration simulation, increased by 0.5%, 6.4% and 2.2% for RM, FM and FN respectively in evaporation simulation. The results here provide insights into the role of airflow in soil hydrology modeling in arid and semi-arid regions.