**1. Introduction**

Vegetation is essential for regional carbon sequestration, soil and water conservation, and climate regulation [1,2]. Arid areas, which account for 40% of the world's land area, are characterised by water shortages and uneven spatiotemporal distributions of water resources [3]. Changes in vegetation and related management practices (e.g., irrigation) in arid areas may lead to the redistribution of regional water resources, which can intensify the competition between ecosystems and humans for water resources [1,4]. In this context, the water demand and water consumption characteristics of vegetation change in arid areas are of particular concern [5,6].

Nowadays, physically based distributed (or semi-distributed) hydrological models can clearly reflect the spatial variability of hydrological processes in a basin, and these models are playing an important role in simulations and predictions of the hydrological cycle in basins [7–9]. Notably, SWAT (Soil and Water Assessment Tools) is a typical distributed hydrological model with a strong physical foundation [10]. It is suitable for simulating surface hydrological processes in a complex basin with a variety of soil types, land use types, slopes, and management practices, and it can be used in data-poor regions [11–13]. Currently, SWAT is a key component of the USDA-Conservation Effect Assessment Project and the USEPA-Hydrologic and Water Quality System [14]. Nevertheless, SWAT has a weak ability to simulate groundwater processes, thereby limiting its application in arid areas with strong surface-water–groundwater exchange [15–17].

The ability of SWAT to simulate groundwater processes can be improved by replacing the groundwater module of SWAT with a well-established groundwater model [15,16].

**Citation:** Jin, X.; Jin, Y.; Mao, X.; Zhai, J.; Fu, D. Modelling the Impact of Vegetation Change on Hydrological Processes in Bayin River Basin, Northwest China. *Water* **2021**, *13*, 2787. https://doi.org/ 10.3390/w13192787

Academic Editors: Dengfeng Liu, Hui Liu and Xianmeng Meng

Received: 14 August 2021 Accepted: 29 September 2021 Published: 8 October 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

A relatively well-established practice for this approach is to couple SWAT with a MOD-FLOW model by using the same temporal and spatial scales for both models, thereby allowing SWAT to calculate and input hydrological response unit (HRU)-based groundwater recharge data to the MODFLOW model and then allowing the MODFLOW model to calculate and return the groundwater flow between the aquifer and river to SWAT [15,16]. The SWAT-MODFLOW code developed by Bailey et al. couples the most recent SWAT code with the MODFLOW-NWT code, which improves the solution of unconfined groundwater flow problems [16,18]. This version of the SWAT-MODFLOW model has recently been developed and is the most widely used. Semiromi and Koch [19] modelled complex interaction of surface–groundwater interactions by MODFLOW in the Gharehsoo River basin, located in Northwest Iran. Mosase et al. [20] used SWAT-MODFLOW to assess the spatial distribution of annual and seasonal groundwater recharge and interactions with surface water in the Limpopo River basin, an arid basin in Africa. Jafari et al. [21] developed a calibration tool for SWAT-MODFLOW and used the model to simulate the runoff and groundwater in Shiraz catchment, located in southwestern Iran. The authors used SWAT-MODFLOW to model the natural water cycle of 'atmosphere–slope–underground–river' components. In this process, the impact of human activities, such as land use/land cover change, is generalised [17]. However, in view of increasingly intense human activities, full consideration of both the impact of human activities and natural factors on the water cycle process in a basin is paramount to ensure that distributed hydrological models can accurately describe the water cycle process [22,23]. Intensive vegetation change is one of the final results of human activities [4]. Vegetation growth in SWAT is a key process to consider in the quantitative modelling of eco-hydrological processes, as it directly affects evapotranspiration (ET), water interception, and soil erosion [23]. Therefore, accurate determination of vegetation change in different HRUs is a key to modelling hydrological processes [24]. SWAT can reflect vegetation changes in a basin by using a land-use update module [23]. However, HRUs, the basic computational units of SWAT, are virtual units, each of which is treated as a lumped unit to achieve the same soil type, land use/cover type, and slope at different spatial sites. This makes it infeasible for SWAT to effectively reflect partial land cover type conversions or land cover types converted to multiple other landcovers within the same HRU. To the best of our knowledge, only a few studies have overcome this limitation of HRUs in SWAT-MODFLOW.

Given the above context, in this study, we developed a LU-SWAT-MODFLOW model by integrating a coupled SWAT-MODFLOW model with dynamic HRUs, which can overcome the limitation of considering the vegetation change compared to the original HRUs for the middle and lower reaches of the Bayin River basin, a typical arid endorheic river, where there are frequent surface-water–groundwater interactions and evident vegetation changes. With the advancement of remote sensing technology, data products with high spatiotemporal resolution such as leaf area index (LAI) and ET, combined with observed hydrological data, were used to calibrate the model [25–27]. The performance of SWAT-MODFLOW and LU-SWAT-MODFLOW were compared first. Later, the hydrological effects of revegetation were analysed based on the simulation results of LU-SWAT-MODFLOW. This study can provide assistance for ensuring revegetation sustainability and rationally allocating water resources in arid areas.

#### **2. Materials and Methods**
