Early detection of within‐field yield variability for high‐value commodity crops, such as cotton (Gossypium spp.), offers growers potential to improve decision‐making, optimize yields, and increase profits. Over recent years, publicly available datasets have become increasingly available and at a resolution where within‐field yield prediction is possible. However, the viability of using these datasets with machine learning to predict within‐field cotton lint yield at key growth stages are largely unknown. This study was conducted on two cotton fields, located near Mungindi, New South Wales, Australia. Three years of yield data, soil, elevation, rainfall, and Landsat imagery were collected from each field. A total of 12 models were created using: (a) two machine learning algorithms: random forest (RF) and gradient boosting machines (GBM); (b) three growth stages: squaring, flowering, and boll‐fill; and (c) two different amounts of variables: all variables and the optimal variables determined by a recursive feature elimination (RFE). Results showed a strong agreement between predicted and observed yields at flowering and boll‐fill when more information was available. At flowering and boll‐fill, root mean square error (RMSE) ranged between 0.15 and 0.20 t ha⁻¹ and Lin's concordance correlation coefficient (LCCC) ranged between 0.50 and 0.66, with RF providing superior results in most cases. Models created using the optimal variables determined by the RFE provided similar results compared to using all variables, allowing greater model accuracy and resolution with targeted sampling. Overall, these findings indicate significant potential of publicly available datasets to predict within‐field cotton yield and guide decision‐making in‐season.

Simulation models are an important tool to predict how farming practices influence utilisation and loss of nitrogen (N). However, many simulation exercises lack sufficient validation of N dynamics from both soil and fertiliser sources and rely on single or a few measurable N pools, potentially shifting bias from one pool to another. This study evaluated the capacity of Agricultural Production Systems sIMulator (APSIM) to simulate N fertiliser budgets in dryland sorghum and wheat systems of eastern Australia using an extensive ¹⁵N dataset from 18 field trials with up to four N rates. Key metrics included ¹⁵N fertiliser recoveries at harvest in soil and plant, and ¹⁵N fertiliser losses. Fertiliser N in APSIM was calculated as the balance between simulated N in the non-fertilised control and the respective applied N rate, and compared to relevant ¹⁵N data. APSIM’s performance for fertiliser N budget simulation was tested with (i) conventional calibration fitting plant N uptake and (ii) improved calibration including the denitrification and mineralisation parameters to fit fertiliser N loss. Simulations with conventional calibration showed good agreement with phenology, soil water and plant N uptake data but largely underestimated the measured fertiliser N loss (RMSE: 23 kg N ha⁻¹, PBIAS: − 75%). The simulated fertiliser N loss with improved calibration agreed better with the observation (RMSE: 18 kg N ha⁻¹, PBIAS: -34%). The identified inconsistencies in the default model’s representation of N cycling highlight opportunities for combined efforts of experimental research and model development to improve the reliability and utility of the model.