The advent of high-spatial-resolution hyperspectral imagery from unmanned aerial vehicles (UAVs) made a breakthrough in the detailed retrieval of crop traits for precision crop-growth monitoring systems. Here, a hybrid approach of radiative transfer modelling combined with a machine learning (ML) algorithm is proposed for the retrieval of the leaf area index (LAI) and canopy chlorophyll content (CCC) of wheat cropland at the experimental farms of ICAR-Indian Agricultural Research Institute (IARI), New Delhi, India. A hyperspectral image captured from a UAV platform with spatial resolution of 4 cm and 269 spectral bands ranging from 400 to 1000 nm was processed for the retrieval of the LAI and CCC of wheat cropland. The radiative transfer model PROSAIL was used for simulating spectral data, and eight machine learning algorithms were evaluated for hybrid model development. The ML Gaussian process regression (GPR) algorithm was selected for the retrieval of crop traits due to its superior accuracy and lower associated uncertainty. Simulated spectra were sampled for training GPR models for LAI and CCC retrieval using dimensionality reduction and active learning techniques. LAI and CCC biophysical maps were generated from pre-processed hyperspectral data using trained GPR models and validated against in situ measurements, yielding R² values of 0.889 and 0.656, suggesting high retrieval accuracy. The normalised root mean square error (NRMSE) values reported for LAI and CCC retrieval are 8.579% and 14.842%, respectively. The study concludes with the development of optimized GPR models tailored for UAV-borne hyperspectral data for the near-real-time retrieval of wheat traits. This workflow can be upscaled to farmers’ fields, facilitating efficient crop monitoring and management.

This study aimed to demonstrate the efficacy of drone-assisted crop monitoring in precision agriculture by evaluating the relationships between the NDVI, leaf area index (LAI), and leaf nitrogen content (LNC) in three wheat varieties (DBW-187, HD-3086, PBW-826) under eight nitrogen treatments (N0–N210). The NDVI was derived from drone-based multispectral imagery at the flowering (90 DAS) and grain-filling (108 DAS) stages. Strong correlations were observed between the NDVI, LAI, and LNC, with the R² values improving from 0.78–0.86 at flowering to 0.88–0.90 at grain filling. These findings highlight the potential of drone-derived indices for efficient crop monitoring, resource use optimization, and yield prediction in precision agriculture.

Soil moisture is an essential environmental parameter affecting hydrological cycles, agricultural productivity, and climate systems. Conventional in situ measurements are precise but do not provide the spatiotemporal coverage for large applications. This research provides an extensive framework for estimating and mapping surface soil moisture by integrating Sentinel-1 Synthetic Aperture Radar (SAR) data with machine learning in the Google Earth Engine (GEE) cloud platform. The study area is the agricultural region of Perambalur district in Tamil Nadu State, India. The research took place between September 2018 and January 2019. The dual-polarized (VV and VH) Sentinel-1 C-band images were collected in tandem with ground truth soil moisture data collected through the gravimetric method. A set of SAR indices and engineered features were extracted from the backscattering coefficients (σ°). A random forest (RF) machine learning model was used in this study to estimate soil moisture. The RF model incorporating the complete set of engineered features showed a coefficient of determination (R²) of 0.694 and a root mean square error (RMSE) of 1.823 (Soil moisture %). The complete processing and modeling workflow was encapsulated in the GEE-based software tool (version 1) providing an accessible, user-friendly platform for generating near-real-time maps of soil moisture. This research proves that the combination of Sentinel-1 data with clever machine-learning algorithms in the GEE cloud platform provides a scalable, efficient, and potent tool for operational soil moisture mapping serving applications in precision agriculture and in the management of the water resource.

The advent of hyperspectral remote sensing represented a breakthrough in the accurate, fast, and non-invasive estimation of important soil fertility parameters. The present study utilizes non-imaging hyperspectral data in the spectral range of 350–2500 nm for estimating soil organic carbon (SOC) content. When partial least squares (PLS) scores were taken as independent variables, support vector machine (SVM) outperformed artificial neural network (ANN) and partial least squares regression (PLSR), achieving an R² value of 0.83. After pre-processing, the proximal spectral values were spatially interpolated to construct a synthetic hyperspectral image of the experimental fields. By applying the regression model to this synthetic hyperspectral imagery, a high-resolution SOC map showing the variability of organic carbon content in the soil was generated.