Chlorophyll is a key biochemical component as it is integral to the photosynthetic mechanism. Chlorophyll content (Chl-ab) is therefore widely used to track vegetation dynamics and assess the health status of canopies. In this paper, we present an innovative approach to accurately estimate chlorophyll content at the canopy level by integrating radiative transfer models with deep learning algorithms. Our methodology utilizes a U-Net deep learning model trained with Sentinel-2 imagery and chlorophyll content obtained from resampled sentinel-2 reflectance simulations conducted with the Pro4SAIL model. We evaluated the resulting Chl-ab predictions against leaf chlorophyll content measured in Lanžhot forest, Czechia, during six field campaigns in 2019 and 2020. Our models effectively captured the complexities of forest canopies in Lanžhot forest, where they achieved an RMSE of 5.32 μg cm-2and MAE of 4.56 μg cm-2.

Vegetation monitoring requires tools to efficiently track key plant traits, such as chlorophyll content. Satellite or aircraft-based Earth Observation is crucial to regularly collect spectral information over extensive vegetated areas. To infer plant traits from these spectral observations, radiative transfer model inversions are needed, but deploying them can be challenging. To address this, we introduce an online radiative transfer (RT)-simulator, to model canopy-level reflectance in crops and forests based on plant traits and canopy properties. The RT-simulator can also resample reflectance as measured by multiple existing and upcoming satellite sensors. It can help both novices and experts in the remote sensing community to easily explore the response of different spectral domains to specific combinations of physiological plant traits. At the same time, the RT-simulator supports efforts to operationally monitor forest health and stress by estimating plant traits through the systematic combination of biophysical models and satellite data.

Cotton significantly contributes to global agriculture and provides livelihoods for approximately 100 million farmers in 80 countries. Therefore, new approaches are needed to better inform producers, in near-real time, for optimising crop management practices, increasing profitability and sustainability. Here, we investigated the potential of proximal sensing metrics, derived from multispectral and thermal bands onboard an Unmanned Aerial Vehicles (UAVs), to estimate variability in cotton production due to different agronomic practices. We employed three main approaches, including (i) multilinear regression (MR), (ii) random forest (RF) and (iii) partial least square (PLS). All methods showed significantly strong relationship with lint yield. Specifically, the MR approach explained around 88% (R2= 0.88, RMSE = 322 kg/ha) of the variance in final yield across all plots. Further research is currently underway to explore the ability of multi-temporal, hyperspectral and radiative transfer models (RTM) to understand variability across different phenological stages in cotton management.

Remote sensing offers great potential to monitor crop performance, which could help to improve water and nitrogen (N) management. The aim of this study is to assess the nutritional and water status of two wheat (Triticum aestivum L.) genotypes (Cellule and Nogal) to determine their performance by means of vegetation indices, plant traits retrieved by a radiative transfer model and thermal data. To this end, two field experiments were conducted in central Spain during 2018–2021. The results showed that the best differentiation between genotype performance was achieved by predicted chlorophyll (Chl) and leaf area index retrieved through the PROSAIL model and the canopy Chl content index (CCCI), showing that the Cellule genotype had a stronger response than Nogal to N application. Similarly, the water deficit index and canopy–air temperature difference showed that Cellule suffered lower water stress than Nogal.

In this paper, we discuss the integration of systems such as multi-dimensional radiative transfer models (RTM) with deep learning (DL) algorithms to estimate plant biochemical, physiological, and morphological traits at canopy level using high-resolution hyperspectral imagery (361 bands in the 400-1000 nm spectral range). We applied the approaches to two case studies for dryland cropping in Australia (i.e., wheat and sorghum). Crop type averages for the early flight for leaf area index (LAI) varied between 2, for Canola, to as high as 4.3 for Lentils. Wheat and Barley had LAI of 4.1 and 3.8 (m2/m2), respectively. Chlorophyll a+b (Ca+b) averages for emerged crops were 18, 41, 44, 51 and 59 μg/cm2for Faba beans, Wheat, Canola, Barley and Oats, respectively. The pigment Anthocyanin varied from 4.9 to 15.9 μg/cm2for Lentils and Canola, respectively. Similar patterns were observed in the Carotenoid (Cx+c) levels (as high as 16.5 μg/cm2for Oats). For sorghum plots, the integrated DL approaches showed significant high correlation in predicting sorghum LAI (R2= 0.84, RMSE = 0.65 m2/m2) and Ca+b(R2= 0.94, RMSE = 4.94 µgcm-2). The maximum velocity carboxylation rates (Vcmax) varied between 45-75 µmol m-2s-1. For both studied periods, we yielded a R2> 0.78 and RMSE-2s-1, being the RMSE lower when using the modelled fluorescence emission for retrieving the Vcmax. In addition, we derived the solar induced fluorescence emission hyperspectral narrowband (5.8 nm) sensing and radiative transfer models (RTM).

We study the matching problem of regular tree languages, that is, "$\exists \sigma:\sigma(L)\subseteq R$?" where $L,R$ are regular tree languages over the union of finite ranked alphabets $\Sigma$ and $\mathcal{X}$ where $\mathcal{X}$ is an alphabet of variables and $\sigma$ is a substitution such that $\sigma(x)$ is a set of trees in $T(\Sigma\cup H)\setminus H$ for all $x\in \mathcal{X}$. Here, $H$ denotes a set of "holes" which are used to define a "sorted" concatenation of trees. Conway studied this problem in the special case for languages of finite words in his classical textbook "Regular algebra and finite machines" published in 1971. He showed that if $L$ and $R$ are regular, then the problem "$\exists \sigma \forall x\in \mathcal{X}: \sigma(x)\neq \emptyset\wedge \sigma(L)\subseteq R$?" is decidable. Moreover, there are only finitely many maximal solutions, the maximal solutions are regular substitutions, and they are effectively computable. We extend Conway's results when $L,R$ are regular languages of finite and infinite trees, and language substitution is applied inside-out, in the sense of Engelfriet and Schmidt (1977/78). More precisely, we show that if $L\subseteq T(\Sigma\cup\mathcal{X})$ and $R\subseteq T(\Sigma)$ are regular tree languages over finite or infinite trees, then the problem "$\exists \sigma \forall x\in \mathcal{X}: \sigma(x)\neq \emptyset\wedge \sigma_{\mathrm{io}}(L)\subseteq R$?" is decidable. Here, the subscript "$\mathrm{io}$" in $\sigma_{\mathrm{io}}(L)$ refers to "inside-out". Moreover, there are only finitely many maximal solutions $\sigma$, the maximal solutions are regular substitutions and effectively computable. The corresponding question for the outside-in extension $\sigma_{\mathrm{oi}}$ remains open, even in the restricted setting of finite trees.

Adjusting nitrogen (N) fertilization and accounting for the legacy of past N fertilizer application (i.e., residual N) based on remote sensing estimation of crop nutritional status may increase resource efficiency and promote sustainable management of cropping systems. Our main goal was to evaluate the potential of hyperspectral airborne imagers and ground-level sensors for identifying N fertilizer rates and the residual N effect from the previous crop fertilization in a maize/wheat rotation. A two-season field trial that provided various combinations of N rates and residual N response was established in central Spain. Ground-level sensors and aerial hyperspectral images were used to calculate vegetation indices (VIs). In addition, the solar-induced chlorophyll fluorescence (SIF760) was estimated by the Fraunhofer line-depth method using high-resolution hyperspectral imagery, and together with biophysical modeling, biochemical and biophysical constituents at canopy scales were retrieved. N uptake, N output, grain N concentration, and proximal sensors discriminated between different N fertilizer rates and identified the residual effect when it was relevant. Structural, photosynthetic pigments and short-wave infrared region (SWIR)-based VIs, together with SIF760and the chlorophyll$a + b$($C_{{\mathrm {ab}}}$), biomass, and the leaf area index (LAI), performed similarly on N rate detection. However, the residual effect of nitrification inhibitors was only detected by the structural (NDVI and OSAVI), chlorophyll (CCCI and NDRE), blue/green, NIR-SWIR ($\text{N}_{850,1510}$) indices, SIF760,$C_{{\mathrm {ab}}}$, biomass, and the LAI. This study confirmed the ability of remote sensing to identify N rates at early growth stages and highlighted its potential to detect residual N in crop rotation.

This work focuses on understanding the effects caused by the within-tree structural heterogeneity on the Crop Water Stress Index (CWSI). We present an assessment of the CWSI variability and its relationship with stomatal conductance (Gs) using different automatic object-based tree-crown detection algorithms based on temperature quartile thresholds. The study was carried out in an almond orchard cultivated under three irrigated regimes. High-resolution (25 cm) thermal imagery was acquired by an aircraft on summer 2015. The tree crowns were segmented into 4 classes using the 25th, 50th and 75th percentiles via automatic object-based methods. Results showed that CWSI was linearly and inversely correlated with Gs in all thermal classes. However, the relationship with Gs was heavily affected by the crown segmentation levels applied, and improved remarkably when CWSI values where those corresponding to the coldest and purest vegetation pixels (r2=0.78 from pure vegetation pixels vs. r2=0.52 when warmer pixels were used).