Leaf area index (LAI) is a crucial biophysical variable for agroecosystems monitoring. Conventional vegetation indices (VIs) based on red and near infrared regions of the electromagnetic spectrum, such as the normalized difference vegetation index (NDVI), are commonly used to estimate the LAI. However, these indices commonly saturate at moderate-to-dense canopies (e.g., NDVI saturates when LAI exceeds three). Modified VIs have then been proposed to replace the typical red/green spectral region with the red-edge spectral region. One significant and often ignored aspect of this modification is that the reflectance in the red-edge spectral region is comparatively sensitive to chlorophyll content which is highly variable between different crops and different phenological states. In this study, three improved indices are proposed combining reflectance both in the red and red-edge spectral regions into the NDVI, the modified simple ratio index (MSR), and the green chlorophyll index (CI_green) formula. These improved indices are termed NDVI_{red& RE} (red and red-edge NDVI), MS_{Rred& RE} (red and red-edge MSR index), and CI_{red& RE} (red and red-edge CI). The indices were tested using RapidEye images and in-situ data from campaigns at Maccarese Farm (Central Rome, Italy), in which four crop types at four different growth stages were measured. We investigated the predictive power of nine VIs for crop LAI estimation, including NDVI, MSR, and CIgreen; the red-edge modified indices: NDVI_Red-edge, MSR_Red-edge, and CI_Red-edge (generally represented by VIRed-edge); and the newly improved indices: NDVI_{red& RE}, MSR_{red& RE}, and CI_{red& RE} (generally represented by VI_{red& RE}). The results show that VI_{red& RE} improves the coefficient of determination (R²) for LAI estimation by 10% in comparison to VI_Red-edge. The newly improved indices prove to be the powerful alternatives for the LAI estimation of crops with wide chlorophyll range, and may provide valuable information for satellites equipped with red-edge channels (such as Sentinel-2) when applied to precision agriculture.

This study focuses on the comparison of hybrid methods of estimation of biophysical variables such as leaf area index (LAI), leaf chlorophyll content (LCC), fraction of absorbed photosynthetically active radiation (FAPAR), fraction of vegetation cover (FVC), and canopy chlorophyll content (CCC) from Sentinel-2 satellite data. Different machine learning algorithms were trained with simulated spectra generated by the physically-based radiative transfer model PROSAIL and subsequently applied to Sentinel-2 reflectance spectra. The algorithms were assessed against a standard operational approach, i.e., the European Space Agency (ESA) Sentinel Application Platform (SNAP) toolbox, based on neural networks. Since kernel-based algorithms have a heavy computational cost when trained with large datasets, an active learning (AL) strategy was explored to try to alleviate this issue. Validation was carried out using ground data from two study sites: one in Shunyi (China) and the other in Maccarese (Italy). In general, the performance of the algorithms was consistent for the two study sites, though a different level of accuracy was found between the two sites, possibly due to slightly different ground sampling protocols and the range and variability of the values of the biophysical variables in the two ground datasets. For LAI estimation, the best ground validation results were obtained for both sites using least squares linear regression (LSLR) and partial least squares regression, with the best performances values of R² of 0.78, rott mean squared error (RMSE) of 0.68 m² m⁻² and a relative RMSE (RRMSE) of 19.48% obtained in the Maccarese site with LSLR. The best results for LCC were obtained using Random Forest Tree Bagger (RFTB) and Bagging Trees (BagT) with the best performances obtained in Maccarese using RFTB (R² = 0.26, RMSE = 8.88 μg cm⁻², RRMSE = 17.43%). Gaussian Process Regression (GPR) was the best algorithm for all variables only in the cross-validation phase, but not in the ground validation, where it ranked as the best only for FVC in Maccarese (R² = 0.90, RMSE = 0.08, RRMSE = 9.86%). It was found that the AL strategy was more efficient than the random selection of samples for training the GPR algorithm.

A better comprehension of soil properties and processes permits a progress in agricultural management effectiveness, together with a diminution of environmental damage and more beneficial use of resources. This research investigated the usage of multispectral (Sentinel-2 MSI) satellite data at the farm/regional level, for the identification of agronomic bare soil presence, utilizing bands of the spectral range from visible to shortwave infrared. The research purpose was to assess the frequency of cloud-free bare soil time-series images available during the year in typical agricultural areas, needed for the development of digital soil mapping (DSM) approaches for agricultural applications, using hyperspectral satellite missions such as current PRISMA and the planned EnMAP or CHIME. The research exploited the Google Earth Engine platform, by processing all available cloud-free Sentinel-2 images throughout a time span of four years. Two main results were obtained: (i) bare soil frequency, indicating where and when a pixel (or an agricultural field) was detected as bare surface in three representative agricultural areas of Italy, and (ii) a temporal sensitivity analysis, providing the acquisition frequency of useful bare soil images applicable for the retrieval of soil variables of interest. It was shown that, in order to provide for an effective agricultural soil monitoring capability, a revisit frequency in the range of five to seven days is required, which is less than the planned specifications e.g., of PRISMA or CHIME hyperspectral missions, but could be addressed by combining data from the two sensors.