In semi-arid countries like South Africa, commercially produced mango ( L.) are grown entirely under irrigation. However, there is little accurate quantitative information on the water use of mango orchards, and few accurate water use models currently exist. In this study, we evaluated the performance of the FAO 56 method of estimating crop evapotranspiration (ET) and a dual source water use model against measurements of actual consumptive water use of a commercial mango orchard. The FAO approach calculates ET as the product of a crop coefficient (K) and the reference evapotranspiration (ET). We derived K for well-watered orchards from readily available data including the fraction of vegetation cover, average tree height, and a stomatal control coefficient. The dual-source model calculated orchard evapotranspiration as the sum of tree transpiration and orchard floor evaporation derived from the modified Shuttleworth and Wallace model. Actual evapotranspiration was measured using an open path eddy covariance system, while tree transpiration was measured using the heat ratio method of monitoring sap flow. Data were collected in a mature ‘Tommy Atkins’ mango orchard grown with micro-sprinkler irrigation in north-eastern South Africa. Results show that the reference evapotranspiration explained most of the variation in orchard transpiration (R ∼ 0.78) compared to the solar radiation (R ∼ 0.66) and the vapour pressure deficit of the air (R ∼ 0.66). Compared with field measured water use data, the FAO method gave relatively accurate estimates of transpiration (R = 0.74, NRMSE = ±18.00 %, NMAE = ±15.00 %, NSE = 0.70) and evapotranspiration (R = 0.58, NRMSE = ±19.00 %, NMAE = ±16.00 %, NSE = 0.55). NSE is the Nash-Sutcliffe Efficiency. The dual-source model yielded slightly less accurate estimates of transpiration (R = 0.61, NRMSE = ±22.00 %, NMAE = ±20.00 %, NSE = 0.39) and evapotranspiration (R = 0.52, NRMSE = ±23.00 %, NMAE = ±25.00 %, NSE = 0.51). The basal crop coefficient (K) showed small seasonal fluctuations, varying between 0.40 and 0.60 throughout the year. In contrast, K showed clear seasonality varying between 0.59–0.64 at flowering and fruit set and rising to 0.81–0.90 during the fruit growing phase. Estimates of the annual total transpiration (686 mm) from the FAO method were within 5 % of the actual measured values (677 mm) while the dual source model estimated 612 mm of transpiration, which was almost 10 % lower than the measured values. The results highlight that even though the FAO method requires fewer input parameters, it is potentially more accurate in estimating water use in mango orchards than the dual-source model.

中文翻译：

---

使用植被覆盖率和双源蒸散模型估算南非东北部一个满果芒果园的作物系数和用水量


在南非等半干旱国家，商业化生产的芒果 (L.) 完全在灌溉下种植。然而，关于芒果园用水的准确定量信息很少，目前也很少有准确的用水模型。在本研究中，我们根据商业芒果园实际消耗水量的测量结果，评估了估算作物蒸散量 (ET) 的 FAO 56 方法和双源用水模型的性能。粮农组织的方法将蒸散量计算为作物系数 (K) 和参考蒸散量 (ET) 的乘积。我们根据现成的数据（包括植被覆盖率、平均树高和气孔控制系数）推导出浇水良好的果园的 K。双源模型将果园蒸散量计算为树木蒸腾量和果园地面蒸发量之和，该总和源自修正的沙特尔沃斯和华莱士模型。实际蒸散量使用开路涡流协方差系统测量，而树木蒸腾量使用监测树液流的热比法测量。数据是在南非东北部一个采用微喷灌技术种植的成熟“汤米·阿特金斯”芒果园中收集的。结果表明，与太阳辐射 (R ∼ 0.66) 和空气蒸气压差 (R ∼ 0.66) 相比，参考蒸散量解释了果园蒸腾量 (R ∼ 0.78) 的大部分变化。与实地测量的用水数据相比，FAO 方法对蒸腾量（R = 0.74，NRMSE = ±18.00 %，NMAE = ±15.00 %，NSE = 0.70）和蒸散量（R = 0.58，NRMSE = ±19.00 %）给出了相对准确的估计。 ，NMAE = ±16.00%，NSE = 0.55）。 NSE 是纳什-萨克利夫效率。 双源模型对蒸腾量（R = 0.61，NRMSE = ±22.00 %，NMAE = ±20.00 %，NSE = 0.39）和蒸散量（R = 0.52，NRMSE = ±23.00 %，NMAE = ±25.00）的估计精度稍差％，NSE = 0.51）。基作系数（K）季节性波动较小，全年变化在0.40～0.60之间。相比之下，K 表现出明显的季节性，在开花和坐果期为 0.59-0.64，在果实生长阶段升至 0.81-0.90。根据FAO方法估算的年总蒸腾量（686毫米）与实际测量值（677毫米）的误差在5%以内，而双源模型估算的蒸腾量为612毫米，比测量值低了近10%。结果强调，尽管粮农组织方法需要的输入参数较少，但它在估算芒果园用水量方面可能比双源模型更准确。