Verticillium wilt (VW) represents the most formidable challenge in cotton cultivation, critically impairing both fiber yield and quality. Conventional resistance assessment techniques, which are largely reliant on subjective manual evaluation, fail to meet the demands for precision and scalability required for advanced genetic research. This study introduces a robust evaluation framework utilizing feature selection and optimization algorithms to enhance the accuracy and efficiency of the severity assessment of cotton VW. We conducted comprehensive time-series UAV hyperspectral imaging (400 to 995 nm) on the cotton canopy in a field environment on different days after sowing (DAS). After preprocessing the hyperspectral data to extract wavelet coefficients and vegetation indices, various feature selection methods were implemented to select sensitive spectral features for cotton VW. By leveraging these selected features, we developed machine learning models to assess the severity of cotton VW at the canopy scale. Model validation revealed that the performance of the assessment models responded dynamically as VW progressed and achieved the highest R² of 0.5807 at DAS 80, with an RMSE of 6.0887. Optimization algorithms made a marked improvement for SVM in severity assessment using all observation data, with R² increasing from 0.6986 to 0.9007. This study demonstrates the potential of feature selection and machine learning methods based on hyperspectral data in enhancing VW management, promising advancements in high-throughput automated disease assessment, and supporting sustainable agricultural practices.

In mountainous citrus orchards, the application of conventional ground sprayers for the control of citrus red mite (Panonychus citri) is often constrained by complex terrain and low operational efficiency. The Unmanned Aerial Spraying System (UASS), due to its low-altitude, low-volume, and high-maneuverability characteristics, has emerged as a promising alternative for pest management in such challenging environments. To evaluate the spray performance and field efficacy of different UASS types in controlling P. citri, five representative UASS models (JX25, DP, T1000, E-A2021, and T20), four mainstream pesticide formulations, and four novel tank-mix adjuvants were systematically assessed in a field experiment conducted in a typical hilly citrus orchard. The results showed that T20 delivered the best overall spray deposition, with upper canopy coverage reaching 10.63%, a deposition of 3.01 μg/cm², and the highest pesticide utilization (43.2%). E-A2021, equipped with a centrifugal nozzle, produced the finest droplets and highest droplet density (120.3–151.4 deposits/cm²), but its deposition and coverage were lowest due to drift. Nonetheless, it exhibited superior penetration (dIPR 72.3%, dDPR 73.5%), facilitating internal canopy coverage. T1000, operating at higher flight parameters, had the weakest deposition. Formulation type had a limited impact, with microemulsions (MEs) outperforming emulsifiable concentrates (ECs) and suspension concentrates (SCs). All adjuvants improved spray metrics, especially Yimanchu and Silwet, which enhanced pesticide utilization to 46.8% and 46.4% for E-A2021 and DP, respectively. Adjuvant use increased utilization by 4.6–11.9%, but also raised ground losses by 1.5–4.2%, except for Yimanchu, which reduced ground loss by 2.3%. In terms of control effect, the rapid efficacy (1–7 days after application, DAA) of UASS spraying was slightly lower than that of ground sprayers—electric spray gun (ESG), while its residual efficacy (14–25 DAA) was slightly higher. The addition of adjuvants improved both rapid and residual efficacy, making it comparable to or even better than ESG. E-A2021 with 5% abamectin·etoxazole ME (5A·E) and Yimanchu achieved 97.4% efficacy at 25 DAA. Among UASSs, T20 showed the rapid control, while E-A2021 outperformed JX25 and T1000 due to finer droplets effectively targeting P. citri. In residual control (14–25 DAA), JX25 with 45% bifenazate·etoxazole SC (45B·E) was most effective, followed by T20. 5A·E and 45B·E showed better residual efficacy than abamectin-based formulations, which declined more rapidly. Adjuvants significantly extended control duration, with Yimanchu performing best. This study demonstrates that with optimized spraying parameters, nozzle types, and adjuvants, UASSs can match or surpass ground spraying in P. citri control in hilly citrus orchards, providing valuable guidance for precision pesticide application in complex terrain.

Recent advancements in precision agriculture have been significantly bolstered by the Uncrewed Aerial Vehicles (UAVs) equipped with multispectral sensors. These systems are pivotal in transforming sensor-recorded Digital Number (DN) values into universal reflectance, crucial for ensuring data consistency irrespective of collection time, region, and illumination. This study, conducted across three regions in China using Sequoia and Phantom 4 Multispectral cameras, focused on examining the effects of radiometric correction on data consistency and accuracy, and developing a conversion model for data from these two sensors. Our findings revealed that radiometric correction substantially enhances data consistency in vegetated areas for both sensors, though its impact on non-vegetated areas is limited. Recalibrating reflectance for calibration plates significantly improved the consistency of band values and the accuracy of vegetation index calculations for both cameras. Decision tree and random forest models emerged as more effective for data conversion between the sensors, achieving R² values up to 0.91. Additionally, the P4M generally outperformed the Sequoia in accuracy, particularly with standard reflectance calibration. These insights emphasize the critical role of radiometric correction in UAV remote sensing for precision agriculture, underscoring the complexities of sensor data consistency and the potential for generalization of models across multi-sensor platforms.