CONTEXT

Mechanistic sorghum models have been mostly used to estimate sorghum yield for grain sorghum for a range of genotype, management, and environmental conditions. There is a lack of model testing for crop growth and development responses for forage genotypes and information for phenological parameterization under sub-optimal water and nitrogen stress conditions in forage systems.

OBJETIVE

The aims of this study were to (i) use NDVI to parametrize APSIM model to estimate forage sorghum phenology, (ii) calibrate APSIM to simulate green cover, intercepted solar radiation and aboveground biomass, and (iii) quantify the variance of inter-annual aboveground biomass and the effect of water availability on forage sorghum biomass under sub-optimal environment × management combinations.

METHODS

We used climate, soil, management records and sorghum crop observations collected from farm and field experiments in Argentina and Australia. NDVI values were gathered from Sentinel-2 and a handheld optical sensor and then related to fAPAR measurements. Phenological stages were derived from fAPAR seasonal dynamics and implemented as input in the APSIM calibration. Finally, we assessed the temporal AGB variability through long-term simulations analysis.

RESULTS AND CONCLUSIONS

NDVI seasonal dynamics accurately represented the fraction of the absorbed photosynthetically active radiation (R²=0.92) and then, the remote-sensing parametrized APSIM model satisfactorily simulated crop phenology (CCC=0.75-0.92, NRMSE=9-22%). The model was also able to satisfactorily simulate crop growth (CCC=0.89 and NRMSE=24.8% for green cover; CCC=0.81 and NRMSE=34.6% for intercepted solar radiation; CCC=0.91 and NRMSE=37% for aboveground biomass). APSIM simulations during 22 years across 5 contrasting locations showed high inter-annual variability of aboveground biomass (CV=47%), mainly driven by inter-annual variation of soil water availability (CV=20%). Our study demonstrated that (i) remote sensing data was a reliable source for APSIM phenology parametrization, (ii) the model was able to satisfactorily simulate crop growth and development of forage sorghum under sub-optimal conditions across several genotype × environment × management combinations and (iii) water availability was the main driver of aboveground biomass inter-annual variance.

SIGNIFICANCE

Given the pressure of the global human population to satisfy an increasing demand for food, our results provide a new path for the combined use of remote sensing and mechanistic modelling to improve forage sorghum biomass estimations in marginal environments.

中文翻译：

---

在次优生长条件下利用遥感和田间数据校准高粱的 APSIM

语境

机械高粱模型主要用于估算一系列基因型、管理和环境条件下高粱的高粱产量。缺乏针对牧草基因型的作物生长和发育响应的模型测试以及牧草系统在次优水和氮胁迫条件下物候参数化的信息。

目标

本研究的目的是 (i) 使用 NDVI 参数化 APSIM 模型来估计牧草高粱物候，(ii) 校准 APSIM 以模拟绿化覆盖、截获的太阳辐射和地上生物量，以及 (iii) 量化年际变化次优环境×管理组合下地上生物量及水分利用率对饲用高粱生物量的影响。

方法

我们使用了从阿根廷和澳大利亚的农场和田间试验中收集的气候、土壤、管理记录和高粱作物观察结果。NDVI 值是从 Sentinel-2 和手持式光学传感器收集的，然后与 fAPAR 测量相关。物候阶段来自 fAPAR 季节性动态，并作为 APSIM 校准的输入实施。最后，我们通过长期模拟分析评估了时间 AGB 变异性。

结果和结论

NDVI 季节动态准确地代表了吸收的光合有效辐射的比例（R ²=0.92），然后，遥感参数化APSIM模型令人满意地模拟了作物物候（CCC=0.75-0.92，NRMSE=9-22%）。该模型还能够令人满意地模拟作物生长（绿色覆盖的 CCC=0.89 和 NRMSE=24.8%；截获的太阳辐射的 CCC=0.81 和 NRMSE=34.6%；地上生物量的 CCC=0.91 和 NRMSE=37%）。APSIM 在 22 年间对 5 个对比地点的模拟显示，地上生物量的年际变化很大（CV=47%），这主要是由土壤水分可用性的年际变化（CV=20%）驱动的。我们的研究表明（i）遥感数据是 APSIM 物候参数化的可靠来源，

意义

鉴于全球人口满足日益增长的粮食需求的压力，我们的研究结果为结合使用遥感和机械建模来改进边缘环境中的草料高粱生物量估计提供了一条新途径。