Modelling the effects of climate change on transpiration and evaporation in natural and constructed grasslands in the semi-arid Loess Plateau, China

Investigating the effects of climate change on transpiration (T) and evaporation (E) in natural and constructed grasslands is of theoretical and practical significance for vegetation restoration and water resource management in the Loess Plateau, China. In this study, a two-year field experiment was conducted in one natural (Imperata cylindrica plot) and two constructed grasslands (Pennisetum giganteum and Medicago sativa plots) in the semi-arid Loess Plateau. Hydrus-1D models were then established and validated to simulate T and E processes under scenarios combining climate characteristics in both future periods and different hydrological years. The results showed that under all climate scenarios, T in the natural grassland, and P. giganteum and M. sativa plots ranged from 63.7-179.3, 126.8-245.5, and 96.9-243.1 mm, respectively, while E ranged from 90.6-131.9, 68.7-92.8, and 70.4-92.4 mm, respectively. Both T and E showed a decreasing trend in the future periods, exhibiting the highest values in wet years and the lowest values in dry years. The effects of climate change on T were greater than that on E in all three grasslands. Of the studied grasslands, T of M. sativa plot exhibited the strongest response, followed by that in the natural grassland and P. giganteum plot. However, E of the natural grassland responded to the strongest degree, followed by that of the P. giganteum and M. sativa plots. Precipitation during the growing-season was the most dominant climatic factor affecting T and E, and was positively linearly related with T and E (P < 0.01) in all three grasslands. The conversion thresholds of precipitation for T-E dominance were 435.8, 149.4, and 253.3 mm for the natural grassland, and P. giganteum and M. sativa plots, respectively. In constructed grasslands, more water was lost through T than through E; hence, the T/ET ratios were high ( > 0.55). In contrast, T was high in the natural grassland only in conditions of sufficient precipitation; hence, the T/ET ratio was relatively lower (around 0.5). Precipitation distribution strongly affected the soil water supply, and the natural grassland was more efficient in maintaining soil water than the constructed grasslands. Our study not only improves our understanding of the hydrological processes and water budget under different grass restoration measures, but also provides valuable guidelines for the management of water resources and the restoration of ecological environment in the Loess Plateau.