Get started
Work

Application of Precision Agriculture Technologies for Sustainable Crop Management in the Southern High Plains

Year: 2024
Type: dissertation
Abstract: Precision agriculture (PA) technologies are essential for advancing sustainable crop management by optimizing resource use and crop productivity. This study investigated the application of remote sensing, machine learning (ML), and deep learning (DL) technologies for enhancing nitrogen (N) management, crop yield prediction, and site-specific crop management. The objectives were to 1) review critical principles of N management in PA and identify challenges and opportunities for optimizing nitrogen use efficiency (NUE) through site-specific practices; 2) develop and evaluate a hybrid CNN-LSTM model that integrates spatial and temporal characteristics of satellite imagery to predict within-field cotton yield variability; 3) evaluate vegetation indices (VIs) derived from UAV and satellite imagery for estimating N content in cotton, comparing the performance of ML models in identifying the most reliable approach for N management; 4) assess the impact of N rates, soil physiochemical properties, and topographic attributes on cotton yield and fiber quality, and delineate management zones for precision N management; and 5) examine the relationship between grapevine vigor and soil physicochemical properties and topographic attributes, contributing to within-vineyard variability, and delineate management zones for site-specific vineyard management. Nitrogen is a critical nutrient for plant growth, but conventional N management practices often result in inefficient use, leading to environmental issues such as nitrate leaching and nitrous oxide emissions. Precision N management, supported by remote sensing and real-time crop monitoring, optimizes NUE and reduces environmental impacts by matching fertilizer application to the crop's specific requirements. Despite these benefits, challenges such as high technological costs and data integration hinder wider adoption. The study reviews advanced technologies for detecting crop N status and explores the agronomic, environmental, and economic impacts of precision N management. Accurate crop yield prediction is crucial for improving precision crop management. The hybrid CNN-LSTM model, incorporating time-series PlanetScope imagery, demonstrated superior performance than conventional ML models such as SVM, RFR, GB, and PLSR. The study, conducted over two years, showed that the CNN-LSTM model outperformed the standalone CNN and LSTM models in capturing the spatial and temporal features of the data. Specifically, the 4-band scenario (including the NIR band) achieved an R² of 0.97 and an RMSE of 55 kg/ha, highlighting the importance of NIR in yield prediction. These findings suggest that modern DL techniques hold significant potential for further advancements in agricultural yield forecasting, especially in enhancing decision-making for resource allocation. In cotton N estimation, various VIs derived from UAV and Sentinel-2 imagery were evaluated, with the Canopy Chlorophyll Content Index (CCCI) proving most effective when paired with high-resolution UAV plant canopy data. Among the ML models tested, RFR performed best in capturing the complex relationships between VIs and N content, achieving an R² of 0.90 and RMSE of 0.27%. The study demonstrated that integrating UAV data with advanced ML models enhances precision in N management, providing accurate, site-specific recommendations. This approach reduces fertilizer input costs while maximizing crop yield and minimizing environmental impact, supporting sustainable farming practices. The cotton yield and fiber quality research emphasized the importance of soil texture and elevation in delineating management zones (MZ) for precision N management. By delineating zones based on these factors, cotton producers can apply N more efficiently, improving NUE and productivity. The study utilized historical yield data to validate these delineated MZ, revealing that site-specific management strategies based on soil texture and elevation can capture within-field variability and improve resource use efficiency, enhancing the sustainability of cotton production. The vineyard research demonstrated the critical role of topographic attributes and soil properties in influencing grapevine vigor variability. By integrating soil texture, elevation, and slope data, three management zones delineated having unique characteristics related to vine vigor. This site-specific approach to vineyard management allows for better resource allocation, improving grape quality and sustainability. The findings highlight the importance of considering both topography and soil in precision vineyard management to optimize production efficiency and long-term sustainability. This research underscores the potential of PA technologies in improving N management, crop yield prediction, and site-specific management across diverse agricultural systems. Future research should focus on refining these technologies and expanding their application to a broader range of crops and agroecological zones, ultimately contributing to the global advancement of sustainable PA.
Author: Karn, Rupak
Language: English
Subfield: Ecology
Open Access status: green
OpenAlex ID: W7026699553
Cites:
Cited by:
Related to: