Mechanistic modelling is gradually replacing empiricism in crop models, focusing on leaf-level physiological processes. This shift necessitates simulating crop surface temperature at infra-canopy sub-daily scales but many crop models still rely on empirical formulations for canopy temperature estimation, typically on a daily basis. We developed MONTPEL, a multi-component Penman-Monteith model that allows simulating the crop energy balance with flexible canopy representations (“BigLeaf” vs. “Layered”, “Lumped” vs. “Sunlit-Shaded”) and accounts for atmospheric stability conditions. We analyzed the model behavior, sensitivity and accuracy, using measurements from four wheat (Triticum aestivum L.) experiments conducted under varying pedoclimatic and water stress conditions. Measurements included hourly energy balance terms (total net radiation, soil heat flux, sensible and latent energy fluxes), hourly temperature of the canopy surface or of leaves at different depths inside the canopy, and sunlit and shaded leaf temperatures around solar noon at different dates. MONTPEL reproduced the measured energy balance terms with a root mean square error (RMSE) between 21 and 87 Wm^-2 and a coefficient of determination (R²) exceeding 0.65. The model's accuracy in simulating canopy temperature, with RMSE ≤ 2.2 °C and R² ≥ 0.92, remained consistent regardless of measurement scale. Adjusting the aerodynamic resistance for atmospheric stability minimized simulated canopy temperature errors, notably in semi-arid conditions. Crop latent energy flux and temperature were most sensitive to the maximal stomatal conductance ($g_{\mathrm{s},\text{max}}$$g_{\mathrm{s},\text{max}}$) parameter. However, using a single $g_{\mathrm{s},\text{max}}$$g_{\mathrm{s},\text{max}}$ value across the simulated experiments yielded satisfactory results, suggesting a weak sensitivity to the temporal and site-to-site variability of $g_{\mathrm{s},\text{max}}$$g_{\mathrm{s},\text{max}}$. Distinguishing sunlit from shaded canopy fractions systematically resulted in lower latent energy fluxes compared to “Lumped” canopy representation results. Analysis identified limitations in the multi-component approach, particularly an unrealistic uniform temperature shift across leaf layers when soil surface temperature changes.

中文翻译：

---

MONTPEL：多分量 Penman-Monteith 能量平衡模型


机理建模正在逐渐取代作物模型中的经验主义，专注于叶片水平的生理过程。这种转变需要在树冠下次日尺度上模拟作物表面温度，但许多作物模型仍然依赖经验公式来估计树冠温度，通常是每天一次。我们开发了 MONTPEL，这是一个多分量 Penman-Monteith 模型，它允许使用灵活的冠层表示（“BigLeaf”与“分层”、“集总”与“Sunlit-Shaded”）来模拟作物能量平衡，并考虑大气稳定性条件。我们使用在不同土壤气候和水胁迫条件下进行的四种小麦 （Triticum aestivum L.） 实验的测量值分析了模型行为、灵敏度和准确性。测量包括每小时能量平衡项（总净辐射、土壤热通量、显热和潜能通量）、冠层表面或冠层内不同深度的叶子的每小时温度，以及不同日期太阳正午前后阳光照射和遮荫的叶子温度。MONTPEL 再现了测得的能量平衡项，均方根误差 （RMSE） 在 21 和 87 ^Wm-2 之间，决定系数 （R²） 超过 0.65。无论测量尺度如何，该模型模拟冠层温度的准确性都保持一致，RMSE ≤ 2.2 °C，R² ≥ 0.92。调整空气动力学阻力以实现大气稳定性，最大限度地减少了模拟的冠层温度误差，特别是在半干旱条件下。 作物潜能通量和温度对最大气孔导度最敏感（g s，max“>$g_{\mathrm{s}，\text{max}}$） 参数。但是，使用单个 g s，max“>$g_{\mathrm{}s，max}$模拟实验中的值产生了令人满意的结果，表明对 g s，max 的时间和位点间变异性具有较弱的敏感性>$g_{\mathrm{}s，max}$. 与 “集总 ”树冠表示结果相比，系统地区分阳光照射和阴影的树冠分数导致较低的潜在能量通量。分析确定了多组分方法的局限性，特别是当土壤表面温度变化时，叶层之间的均匀温度变化是不切实际的。