Evapotranspiration (ET) is a critical climate and ecosystem variable that interconnects water, energy, and carbon cycles. Breathing Earth System Simulator (BESS) is one of the state-of-the-art biophysical models capable of producing spatio-temporal continuous ET results. However, we found that since the BESS model does not use an explicit constraint on soil moisture (SM), it has a relatively lower performance under drier conditions. Given that changes in land surface temperature (LST) are closely associated with surface water status and sensible heat energy, we hypothesize that integrating LST changes could explicitly add the soil moisture constraints and thus enhance BESS’s ability to estimate ET. Here we used the morning rise rate of LST (Trate) as a proxy of LST change because of the low noise level in Trate as well as Trate’s close relationship with daily mean sensible heat. To test the hypothesis, this study first assessed whether the performance of BESS ET can be explained by the LST change, targeting grassland sites of the AmeriFlux network in the US Midwest and Great Plains. Specifically, the ET deviation (i.e., the difference between BESS-modeled ET and field-measured ET) and Trate deviation, as well as their relationships, were investigated under different conditions of precipitation, SM, and vapor pressure deficit at the AmeriFlux sites. Results indicated that BESS ET exhibited consistently higher performance under well-watered conditions than water-deficit conditions. Also, the deviations of ET and Trate became more negatively correlated under water-deficit conditions. Leveraging the empirical relationship between ET and Trate deviations, this study developed a new way to calibrate BESS ET based on Trate calculated from LST diurnal observations, particularly under soil or atmospheric water-deficit conditions. After calibrating BESS ET, the statistical indicators between the calibrated ET and the ground measurements showed meaningful improvements relative to those before calibration. Specifically, in the Midwest (Great Plains), R2 increased from 0.42 to 0.51 (from 0.45 to 0.46), and RMSE and absolute bias decreased by 12% and 42% (11% and 45%), respectively. This study highlights that the morning rise rate of LST can effectively constrain the ET models that have no SM constraints under water-deficit conditions and also sheds lights on improved ET estimation for crop, biofuel, and pastureland production in dryland and semi-dryland ecosystems.

中文翻译：

---

在缺水条件下，结合地表温度的变化可以提高 BESS 蒸散量估计：美国中西部和大平原草原的案例研究


蒸散 （ET） 是水、能源和碳循环相互关联的关键气候和生态系统变量。呼吸地球系统模拟器 （BESS） 是最先进的生物物理模型之一，能够产生时空连续 ET 结果。然而，我们发现，由于 BESS 模型没有对土壤水分 （SM） 使用显式约束，因此它在较干燥的条件下性能相对较低。鉴于地表温度 （LST） 的变化与地表水状况和感热能密切相关，我们假设整合 LST 变化可以明确增加土壤水分约束，从而增强 BESS 估计 ETS 的能力。在这里，我们使用 LST 的晨升率 （Trate） 作为 LST 变化的代理，因为 Trate 的噪声水平较低，并且 Trate 与日平均显热密切相关。为了检验这一假设，本研究首先评估了 BESS ET 的性能是否可以用 LST 变化来解释，目标是美国中西部和大平原的 AmeriFlux 网络的草原站点。具体来说，在 AmeriFlux 站点的不同降水、SM 和蒸气压不足条件下研究了 ET 偏差（即 BESS 建模的 ET 和现场测量的 ET 之间的差异）和 Trate 偏差以及它们之间的关系。结果表明，BESS ET 在浇水良好的条件下表现出始终高于缺水条件下的性能。此外，在缺水条件下，ET 和 Trate 的偏差变得更加负相关。 利用 ET 和 Trate 偏差之间的经验关系，本研究开发了一种基于 LST 昼夜观测计算的 Trate 校准 BESS ET 的新方法，特别是在土壤或大气缺水条件下。校准 BESS ET 后，校准后的 ET 与地面测量值之间的统计指标相对于校准前显示出有意义的改进。具体来说，在中西部（大平原），R2 从 0.42 增加到 0.51（从 0.45 增加到 0.46），RMSE 和绝对偏差分别下降了 12% 和 42%（11% 和 45%）。本研究强调，LST 的早晨上升速率可以有效约束在缺水条件下没有 SM 约束的 ET 模型，并为改进旱地和半旱地生态系统中作物、生物燃料和牧场生产的 ET 估计提供了启示。