Although herbicide drifts are known worldwide and recognized as one of the major risks for crop security in the agriculture sector, the traditional assessment of damage in cotton crops caused by herbicide drifts has several limitations. The aim of this study was to assess proximal sensor and modelling techniques in the detection of phenoxy herbicide dosage in cotton crops. In situ hyperspectral data (400-900 nm) were collected at four different times after ground-based spraying of cotton crops in a factorial randomized complete block experimental design with dose and timing of exposure as factors. Three chemical doses: nil, 5% and 50% of the recommended label rate of the herbicide 2,4-D were applied to cotton plants at specific growth stages (i.e. 4-5 nodes, 7-8 nodes and 11-12 nodes). Results have shown that yield had a significant correlation (p-values <0.05) to the green peak (~550 nm) and NIR range, as the pigment and cell internal structure of the plants are key for the assessment of damage. Prediction models integrating raw spectral data for the prediction of dose have performed well with classification accuracy higher than 80% in most cases. Visible and NIR range were significant in the classification. However, the inclusion of the green band (around 550 nm) increased the classification accuracy by more than 25%. This study shows that hyperspectral sensing has the potential to improve the traditional methods of assessing herbicide drift damage.

Yield loss in crops is often associated with plant disease or external factors such as environment, water supply and nutrient availability. Improper agricultural practices can also introduce risks into the equation. Herbicide drift can be a combination of improper practices and environmental conditions which can create a potential yield loss. As traditional assessment of plant damage is often imprecise and time consuming, the ability of remote and proximal sensing techniques to monitor various bio-chemical alterations in the plant may offer a faster, non-destructive and reliable approach to predict yield loss caused by herbicide drift. This paper examines the prediction capabilities of partial least squares regression (PLSR) models for estimating yield. Models were constructed with hyperspectral data of a cotton crop sprayed with three simulated doses of the phenoxy herbicide 2,4-D at three different growth stages. Fibre quality, photosynthesis, conductance, and two main hormones, indole acetic acid (IAA) and abscisic acid (ABA) were also analysed. Except for fibre quality and ABA, Spearman correlations have shown that these variables were highly affected by the chemical. Four PLS-R models for predicting yield were developed according to four timings of data collection: 2, 7, 14 and 28 days after the exposure (DAE). As indicated by the model performance, the analysis revealed that 7 DAE was the best time for data collection purposes (RMSEP = 2.6 and R² = 0.88), followed by 28 DAE (RMSEP = 3.2 and R² = 0.84). In summary, the results of this study show that it is possible to accurately predict yield after a simulated herbicide drift of 2,4-D on a cotton crop, through the analysis of hyperspectral data, thereby providing a reliable, effective and non-destructive alternative based on the internal response of the cotton leaves.