A combination of field data and models have been used to estimate long-term carrying capacity (LTCC) of domestic livestock in Queensland grazing lands. These methods have been synthesised and coupled with recent developments in science and information technology to provide a fully-automated approach of modelling LTCC through the FORAGE online system. In this study, the GRASP model was used to simulate pasture growth with parameter sets and safe pasture utilisation rates defined for 225 land types across Queensland. Distance to water points was used to assess the accessibility of pastures to livestock. Spatial analysis classified the property into unique areas based on paddock, land type and distance to water points, which estimated pasture growth, pasture utilisation and accessibility at a sub-paddock scale. Thirteen foliage projective cover (FPC) classes were used in modelling the pasture system to deal with the non-linear relationship between tree and grass interactions. As ‘proof of concept’, remotely-sensed individual-date green ground cover data were used to optimise the GRASP model parameters to improve the model performance, and a Monte Carlo analysis provided uncertainty estimates for model outcomes. The framework provides an efficient and standardised method for estimating LTCC. To test the system, LTCCs from 43 ‘benchmark’ properties were compared with simulated LTCCs, and 65% of the modelled LTCCs were within ± 25% of the benchmark LTCCs. Due to uncertainties in model inputs at the property scale and in model simulation, the modelled LTCC should be used as a starting point for further refinement of actual property LTCC.

结合现场数据和模型已被用于估计昆士兰牧场家畜的长期承载能力 (LTCC)。这些方法已被综合并与科学和信息技术的最新发展相结合，以提供一种通过 FORAGE 在线系统对 LTCC 建模的全自动方法。在这项研究中，GRASP 模型用于模拟牧场生长，其中包含为昆士兰州 225 种土地类型定义的参数集和安全牧场利用率。到取水点的距离被用来评估牧场对牲畜的可达性。空间分析根据围场、土地类型和到取水点的距离将财产划分为独特的区域，从而在子围场范围内估计牧场的生长、牧场利用和可达性。使用 13 个树叶投影覆盖 (FPC) 类对牧场系统进行建模，以处理树和草相互作用之间的非线性关系。作为“概念证明”，遥感个体日期绿色地面覆盖数据用于优化 GRASP 模型参数以提高模型性能，蒙特卡罗分析为模型结果提供了不确定性估计。该框架为估计 LTCC 提供了一种高效且标准化的方法。为了测试系统，将来自 43 个“基准”特性的 LTCC 与模拟 LTCC 进行了比较，并且 65% 的建模 LTCC 在基准 LTCC 的 ± 25% 范围内。由于模型输入在属性尺度和模型模拟中的不确定性，建模的 LTCC 应作为进一步细化实际属性 LTCC 的起点。