植被冠层导度参数化改进对SWAT模型蒸散发及径流模拟的影响

Effects of Improved Parameterization of Vegetation Canopy Conductance on Simulation of Evapotranspiration and Runoff in SWAT Model

摘要: 高寒地区(如青藏高原地区)水循环过程复杂，蒸散发是决定水资源平衡的关键环节，但传统SWAT (Soil and Water Assessment Tool)模型在蒸散发模拟过程中，受植被冠层导度参数化精度限制，导致水循环过程模拟存在较大不确定性，影响流域水资源评估与管理的可靠性。针对该问题，本研究耦合冠层导度模型——Irmak模型至SWAT模型，构建改进的SWAT-I模型，并在拉萨河流域开展模拟及验证。结果表明：1) SWAT-I模型显著提升了水文过程模拟性能，率定期与验证期的径流Nash-Sutcliffe效率系数较传统SWAT模型分别提高8.11%和4.23%；2) SWAT-I模型中Irmak模型的冠层导度的多因子协同参数化方案改善了蒸散发的模拟，R²提升57.14%，PBIAS改善28.71%；3) SWAT-I模型表明拉萨河流域蒸散发表现出明显的空间异质性，2000~2017年间年均蒸散发量以9.94 mm/a的速率显著上升，其中下游低海拔地区蒸散发量显著高于上游高海拔地区；4) 降水–水文要素关系呈现海拔梯度特征：流域上游受冰川融雪影响，降水与流域径流的相关性较弱(相关系数约为0.50)，而中下游降水主导区相关系数达0.70以上，符合水源分割理论。本研究提出的耦合Irmak冠层导度模型的SWAT-I模型可有效提升流域蒸散发、径流等水循环过程的模拟精度，为气候变化下拉萨河流域水资源精准管理提供了模型支撑。

Abstract: In alpine regions such as the Qinghai-Xizang Plateau, the water cycle is highly dynamic and complex, with evapotranspiration playing a pivotal role in regulating regional water availability. However, the traditional SWAT (Soil and Water Assessment Tool) model is constrained by the limited parameterization accuracy of vegetation canopy conductance, introducing significant uncertainty into water cycle simulations and undermining the reliability of water resource assessments and management. To address this limitation, this paper integrated the Irmak canopy conductance model into SWAT model, developing a modified SWAT-I model, which was applied and validated in the Lhasa River Basin. The results demonstrate that: 1) SWAT-I model significantly enhances hydrological process simulations, with Nash-Sutcliffe efficiency coefficients for runoff improving by 8.11% and 4.23% during the calibration and validation periods, respectively, compared to the original SWAT model; 2) The multi-factor parameterization scheme of canopy conductance in SWAT-I improves ET simulations, increasing R² by 57.14% and reducing PBIAS by 28.71%; 3) SWAT-I model reveals strong spatial heterogeneity in ET, with annual ET increasing at a rate of 9.94 mm/a from 2000 to 2017, and ET in the downstream low-elevation areas significantly exceeding that in upstream high-elevation regions; 4) The relationship between precipitation and runoff shows an altitudinal gradient, with weak correlation in the glacier-fed upper reaches (correlation coefficient is about 0.50), while in the precipitation-dominated middle and lower reaches, the correlation exceeds 0.70, consistent with water source partitioning theory. These findings demonstrate that the SWAT-I model, incorporating the Irmak canopy conductance model, substantially improves the accuracy of evapotranspiration and runoff simulations, providing a robust modeling framework for the precise management of water resources in alpine basins under climate change.