This PDF file contains the front matter associated with SPIE Proceedings Volume 11856, including the Title Page, Copyright information, and Table of Contents.

The MORERA program has recently been selected as one of the “Missions Science and Innovation” from the Spanish CDTI, an innovative program targeting solutions for deep social problems through innovation. The main Spanish industry is Agriculture (11% GDP), but this sector is threatened by climate change, as 34% of the Spanish irrigated surface is considered out of balance. Difficulty of providing useful and fully processed information to the end-users for supporting their decisions severely affect the optimization of the resources. Well informed decisions optimize resources and costs, maximizing productivity. To solve this problem, MORERA involves in a unique project the complete value chain, from sensor to user, thanks to a solid consortium, and it is based on three pillars: - Final personalized irrigation requirements that will be directly provided to the user using a mobile device. - Artificial intelligence techniques will be used to combine all relevant data to build a final watering recommendation. - A compact and highly specific freeform optical instrument will be used to estimate evapotranspiration data at farm level with required TIR bandwidth and spatial resolution. Since no present instrument fulfills these requirements, it will be developed in the framework of the project. The MORERA concept can be extrapolated to many remote sensing applications, and to take advantage of this, it has been conceived as a modular system, where each module may be adapted with minor impact. This first system is focused on providing precise irrigation and fertilization recommendations, as well as self-learning yield estimations.

The 2020+ Common Agricultural Policy encourages the use of Copernicus remote sensing data for the monitoring of agricultural parcels. In this work, a procedure for automatic identification of land use from remote sensing data is proposed. The approach includes the use of spectral information of Sentinel-2 time series over the Valencia province (Spain) during the agronomic year 2027/2018, and deep learning recurrent networks. In particular, a bi-directional Long Short Term Memory (Bi-LSTM) network was trained to classify active land uses and abandoned lands. A comparison exercise was undertaken to assess the classification power of the Bi-LSTM as compared to the random forest (RF) algorithm. The Bi-LSTM network outperformed the RF, and provided and overall accuracy of 97.5% when discriminating eleven land uses including abandoned lands. The results suggest the proposed methodology could potentially be implemented in an automated procedure to supervise the CAP requirements to access subsidies. In addition, the classification process also supports the continuous update of the Land Parcel Identification System (LPIS), which allows paying agencies to uniquely identify land parcels in space, store records of land uses (and assess its evolution), and ultimately ease the declaration procedure to both farmers and paying agencies.

Experiments were carried out to investigate the use of Land Use and Coverage Area frame Survey (LUCAS) dataset and Sentinel-2 imagery to produce a land cover map in Portugal through automated supervised classification. LUCAS is a free land cover land use (LCLU) dataset based in Europe, while Sentinel-2 satellites provide also free images with short revisit frequency. The goal was to evaluate if LUCAS dataset from 2018 can be used as a single reference dataset for land cover classification at national level. The Random Forest (RF) algorithm was used. Some processing steps were undertaken to use LUCAS as reference dataset. The original LUCAS LCLU nomenclature was modified into a new nomenclature composed of 12 and 6 level-2 and level-1 map classes, respectively. Filtering was performed on LUCAS metadata, reducing the initial number of LUCAS points over Portugal from 7168 to 4910. Monthly composites of Sentinel-2 images acquired between October 2017 and September 2018 were used. To reduce the imbalance in LUCAS training points, an oversampling technique based on Synthetic Minority Over-Sampling Technique (SMOTE) was used. An independent validation dataset was produced with 600 points. RF shows an overall accuracy (OA) of 57% for level-2 and 72% for level-1 nomenclatures. When using the oversampling technique, the OA accuracy increases by 3% for level2 and 2% for level-1. The preliminary results of this experiment show that LUCAS dataset used in supervised machine learning classification has potential to produce a reliable land cover map at national scale.

The launch of the Copernicus Sentinel-2 mission offered new insights for the management of the European Common Agrarian Policy (CAP). Sentinel-2 provides information at a spatial and temporal resolution of 10 m and 5 days, respectively. However, this unprecedented time series of high resolution satellite imagery requires from approaches to extract meaningful agronomical information and reduce dimensionality. This could be the case of land surface phenology, which consists in estimating key phenometrics related to agronomical events from time series of vegetation indices (VIs). Knowing the dynamics of crop phenology is essential for the correct monitoring of CAP. We used EVI2 (Enhanced Vegetation Index 2) time series of Sentinel-2 data for the period 2018-2020. EVI2 is a VI widely used as an indicator of plant vigour, that avoids saturation in regions with high biomass. Double Logistic smoothing method was used to fill the gaps caused by the lack of images due to cloud presence or sensor failures. We selected plots of durum and common wheat, sorghum, barley and triticale according to the Geographical Information System for the CAP (GISCAP-CAP) declarations in Andalusia, Spain. The phenometrics extracted were start of the season (SOS), middle of the season (MOS), end of the season (EOS), their respective values of EVI2, and length of the season (LOS) (EOS-SOS). The aim of this study is to characterise the phenology of different winter cereals, through the extraction of phenometrics, and to evaluate whether these latter measures can serve to distinguish them. Results show that the response is quite similar between all of them, except sorghum. Common wheat shows the earliest SOS, followed by barley, durum wheat, triticale and sorghum. Common wheat shows the earliest EOS, followed by durum wheat, barley, triticale and sorghum.

Land surface phenology (LSP), the study of seasonal dynamics of vegetation analysing phenological metrics -phenometrics- derived from vegetation indices time series (VI), has emerged as an important research focus in recent decades as LSP patterns have been considered as an important ecological indicator for understanding the functioning of terrestrial ecosystems. LSP from high-spatial-resolution satellite imagery in ecosystems with significant heterogeneity of plant species, such as Macaronesian ecosystems, are needed for a better understanding on how these ecosystems function. The objective of this study was to monitor LSP dynamics of representative species of the Canary Islands: Olea Cerasiformis, Pistacia atlantica, Juniperus turbinata, Pinus canariensis, Myrica Faya and Erica arborea. NDVI (Normalised Difference Vegetation Index) Sentinel-2 time series at a spatial and temporal resolution of 10 meters and 5 days were generated for the 2018-2020 period. Atmospheric disturbances and noise were reduced using a double-logistic function. SOS (start of the growing season), EOS (end of the growing season) and LOS (length of the growing season) were extracted using a threshold-based method. Thermophilus species, such as Olea Cerasiformis and Pistacia atlantica had the SOS in October-November and the EOS between June and July. Juniperus turbinata showed double seasonality in La Palma, being the first growing season between November-December and April-May and the second growing season between May-June and September-October. Growing season of Pinus canariensis started in September-October and ended in April-June, nevertheless a double seasonality was observed in some locations of Pinus canariensis, probably associated to the understory. Subtropical laurel forest composed by different plant species, such as Myrica Faya and Erica arborea, did not show a clear seasonality. The species-specific LSP patterns for the Canary Islands can contribute to stablishing a baseline to monitor future impacts of climate change in Macaronesian biogeographical region.

Slag consists mostly of mixed oxides of elements such as silicon, and recycled slag may be used for cultivating. The relationship between slag fertilization and plant growth rate can be expected to change depending on the volume of slag applied. Quantifying the chlorophyll content, an effective indicator of disease as well as nutritional and environmental stresses on plants, will enable optimal slag fertilization and then monitoring chlorophyll content using field measurements would enable the determination of optimal slag fertilization rates. In this study, radish plants (Raphanus sativus L), which belongs to the family Brassicaceae and is popular root vegetable in both tropical and temperate regions, were cultivated with slag fertilization and the potential use of hyperspectral reflectance was evaluated. Some preprocessing techniques were effective for retrieving chlorophyll contents in radish leaves from hyperspectral reflectance and then the regression model based on random forest and continuum-removed reflectance had the highest performance with a root mean square error of 5.141 μg cm^-2 and RPD values of 1.858 for the test data set.

QLM are developing a novel remote gas imaging sensor for the detection, imaging, and quantification of methane and other greenhouse gas emissions. The sensor combines aspects of Tunable Diode Laser Absorption Spectroscopy (TDLAS) with Differential Absorption Lidar (DIAL) and Time Correlated Single Photon Counting (TCSPC) to enable remote spectroscopy and ranging with low power semiconductor diode lasers. We directly measure the shape of a gas absorption line by continuously sweeping the output wavelength of a diode laser across the line, and simultaneously modulate the laser output to encode the light signal and then use a digital time-domain correlation algorithm between the transmitted and detected light to identify the returned light. By simultaneously tuning the laser wavelength and modulating the amplitude it is possible to determine both the range the laser light has traveled, as with typical Lidar, and the amount of a particular gas that the laser light has passed through, as with typical TDLAS. Our methane sensors operate around the CH4 absorption line at 1650.9 nm. Tests with calibrated gas cells and controlled gas releases have demonstrated quantification of leak rates as low as 0.01 g/s with accuracy around 25% and detection at distances over 100 m. The accuracy, speed, and practicality of the sensor, combined with an expectation of low-cost in volume, offers the potential that these sensors can be effectively applied for widespread continuous monitoring of industrial methane emissions. We are developing a cloud-based server that automates the collection, analysis and reporting of gas data and provides an autonomous leak monitoring system that can accurately identifying emissions by position, size, and duration.

The challenge of classifying and locating Phoenix palm trees in different scenes with different appearances and varied ages has been addressed with deep learning object detection over aerial images. Nevertheless, an explicit limitation hereof is that palms should be visually identifiable in the image—i.e., palm crowns should be larger than the pixel size. Unfortunately, high-spatial resolution imagery is not widely and directly available in the Phoenix palm growing regions of the Mediterranean, Middle East, and North Africa. This study, therefore, presents the re-implementation of a semantic segmentation architecture to train a model able to classify Phoenix palm pixels. This is applied to freely available medium resolution space-borne Sentinel-2 images over the Spanish island of La Gomera (Canary Islands). At the study site, a total of 116,330 Phoenix palms had been inventoried by the local government. Palms appear in multiple, heterogeneous environments, which implies a background variability that is a persistent challenge for palm pixel classification. The re-implemented architecture is a novelty in deep semantic segmentation and density estimation initially developed for counting objects of sub-pixel size. And it proved to be successful for creating a model of palm classification, thereby compensating for the limited spatial resolution of the Sentinel-2 images. The palm tree sub-pixel classification model achieved an overall accuracy of 0.921, with a recall and precision of 0.438 and 0.522. These results demonstrate the potential of remote sensing data of medium-spatial resolution for vegetation mapping in applications where trees are scattered over extensive areas.

Albedo has long been recognized as a relevant bio-geophysical variable to model Earth surface and it was involved in all the climate simulation models. Therefore, the correct modelling of albedo is essential to reduce the error propagation in the prediction algorithms. To meet such a purpose, different methods have been developed over the past years. Among them, the simplified approach proposed by Liang in 2000 and the corrected algorithm introduced by Silva et al. (2016) are commonly used. To the best of our knowledge, the outcomes produced by applying such techniques have not been investigated yet. The present paper is intended to explore the potentialities of Google Earth Engine (GEE) platform in estimating land surface albedo from three medium-resolution geospatial data gathered by different Landsat sensors in diverse acquisition periods. Java-script code was developed to numerically implement the above-mentioned algorithms in GEE environment. Their performances were compared and the error committed adopting the simplified method was quantified. As a result, the corrected algorithm reported more accurate values. Nevertheless, its complexity implies a high implementation difficulty and, consequently, a higher processing time is required to handle the data. Conversely, the simplified approach allowed to estimate land surface albedo in a short time. Quantifying the error committed using the simplified approach allows us to correct its results, improving their accuracy. Although obtained results are preliminary, this research enhanced the possibility to model the albedo by adopting the simplified algorithm after correcting it. This implies to reduce error propagation and, simultaneously, to speed up the data handling.

Snow is a ubiquitous natural material that plays an important role in Earth’s climatological system and energy resource budget. Its insular and reflective properties are key factors contributing to the radiation budget of the cryosphere. Due to its prevalence at extreme latitudes, the monitoring of snowpack quantities is often performed via remote observation. These data are acquired using either satellite readings or by fixing instruments to the underside of aircraft. When acquiring data remotely, it is important to account for the angular configuration of the source illumination and the location of the instrument relative to the surface since reflection is affected by the geometry of the observation. In other words, the bidirectional reflectance distribution function (BRDF) depends on the angle of incidence of the solar illumination and the angle of observation in addition to the wavelength of the incident light. It has been recognized that the granular properties of a snowpack markedly influence its BRDF. Unfortunately, works examining the effects of snow grain characteristics, such as size and facetness, on BRDF outputs are still scarce. Moreover, measured BRDF values from field studies presented in the literature are limited to specific target samples. This further hinders a more comprehensive understanding of the effects of changes in snow characterization parameters on the bidirectional reflectance of snowpack. The measured datasets often do not provide a detailed characterization of the target samples either, which also reduces their usefulness for elucidating these effects. To address these limitations and enhance the current understanding about the sensitivity of snow BRDF to variations in grain characteristics, we have conducted controlled experiments employing a first-principles in silico experimental framework supported by measured data. Our findings unveil the qualitative effects that snow granular properties have on bidirectional reflectance of snowpack, and highlight the importance of accounting for snow granular properties in remote sensing applications. In addition, our in silico experiments provide a high-fidelity assessment of snow BRDF with respect to key wavelengths particularly relevant for remote sensing applications. More broadly, our investigation demonstrates how remote observations of snow-covered terrains can be significantly improved by the correct incorporation of snow grain characteristics into the bidirectional reflectance models used to assess snowpack properties.

Remote sensing methods allow obtaining important information from the Earth's surface to effectively evaluate agricultural processes. The present research work proposes to carry out the comparative analysis of evapotranspiration through the methodologies of the SEBAL model and pan evaporimeter in the Huancané basin, Peru. The specific objectives were to estimate evapotranspiration using the SEBAL model from Landsat 8 images, to estimate real evapotranspiration using the pan evaporimeter method from meteorological data from the Huancané and Muñani stations, and to compare and validate evapotranspiration results. obtained from the pan evaporimeter method with the SEBAL model. The methodological stages that are considered to achieve the objectives are: the collection of meteorological data from National Meteorology and Hydrology Service (SENAMHI) of Peru and Landsat 8 satellite images and processing of existing information. For the SEBAL model, the lowest evapotranspiration values correspond to areas with soils without crops or low vegetation cover (NDVI <0.21) and for areas covered with vegetation or grasslands (NDVI> 0.41) represent values between 1.50 to 4.20 mm / day. The pan evaporimeter method allowed to determine the real evapotranspiration (ETR), on average they are 2.10, 2.44, 1.76 and 2.04 mm / day of areas under analysis. The comparison and validation of the evapotranspiration values observed (ETR pan evaporimeter) and estimated (ETR SEBAL), for the analysis areas near the Huancané station present a mean square error of 0.26 and 0.25, coefficient of determination of 0.97 and a Nash-Sutcliffe efficiency of 0.81 and 0.83. Likewise, for the areas near the Muñani station where they show a mean square error of 0.13 and 0.14, coefficient of determination of 0.97 and 0.93; and a Nash-Sutcliffe efficiency of 0.81 and 0.82. The results obtained with the SEBAL model are very satisfactory, which shows that its use is feasible.

In this paper we used two methodologies of remote sensing data and its derived variables for the estimation of Evapotranspiration (ET). In the first method, the sensible heat flux was calculated by combining air temperature and the remotely sensed surface temperature using Thornthwaite method. We applied and evaluated the ET successfully in the AlAin area of United Arab Emirates. In the Second method, vegetation index derived using Landsat 8 OLI was used for the determination of surface resistance for latent heat. To derive the predicted ET using Normalized Difference Vegetation Index (NDVI), regression analyses were conducted between data derived from satellites, published field meteorological stations data and ET values. From the collected variables of interest, we have also studied the bivariant density estimation curves. It is evident from the patterns of multimodal data that the data belong to different locations with different ET status. It was also observed that wind velocity (U) seems to be decreasing with increasing ET and rest all variable were increasing with increasing ET, which depend on the saturation vapor pressure (SVP). From this approach, we confirmed that the prediction of ET is achievable from the remote sensing data. It is also confirmed that the predicted ET results gained from the NDVI regression functions were comparable to the ET values obtained by the previously published field data. The results showed that indirect application of remotely sensed vegetation index could be used for the ET determination.

Estimating actual crop evapotranspiration (ETc) is a critical component in tracking soil water availability and managing near real-time irrigation scheduling. Energy and water balance models are two common approaches for estimating daily crop ETc. The Spatial EvapoTranspiration Modeling Interface (SETMI) hybrid model combines these two approaches and has been used to increase the accuracy of modeled ETc and soil water content by assimilating actual ET values to update the soil water balance. In this study, modeled daily ETc from the two-source energy balance (TSEB), root zone water balance, and the hybrid modeling approach were compared to measured ETc from eddy covariance flux tower systems to quantify model accuracy. The TSEB model used the Priestly-Taylor approximation for estimating ETc and the water balance model was updated with reflectance-based crop coefficients. The models were informed with UAS acquired multispectral reflectance and thermal infrared imagery collected over irrigated and rainfed maize and soybean fields during the 2018-2020 growing seasons.

The purpose of the study was the development of GIS models implementing two empirical methods of reference evapotranspiration (ETo), namely FAO-56 Penman-Monteith (FAO-56 PM) and Hansen, and two empirical methods of actual evapotranspiration (ETa), namely Turc and Corrected Turc (CT). GIS software (ArcMap 10.6) model builder environment was used for model-implementation, though the models could be easily transferred in relative software types (e.g. QGIS, ERDAS). The affordances of the presented models are flexibility and applicability to any study area (using the corresponding raster images of remote-derived or interpolated data as inputs), cell-by-cell calculations (estimates for the continuum of space) and dual representation of the empirical formulae; as optical models and as python scripts.Models’ application has been made on Peloponnese, Greece, a complex Land Use Land Cover (LULC) Mediterranean area, engaging MODIS LST Terra day and night inputs which have been acquired from USGS Earth Explorer and NASA EARTH DATA platforms. MODIS LST day and night products have been proved satisfactory proxies of maximum and minimum values (respectively) of local near surface air temperature (Tair). Alternatively, considering overpassing local time for each study-area Aqua or any other satellite products could be used as inputs. Furthermore, 8-day composites of MODIS net evapotranspiration (ET) (MOD16A2V6) have been acquired and compared to model outputs for different time scales and ET types. Corrected Turc and Hansen formulae exhibited estimates satisfactorily close to MODIS ET for different time scales. Overall, the proposed models have been proved time-saving and useful tools for water management applications that can be utilized by inexperienced to advanced users.

The pressure on vegetation, whether forests, meadows or cultivated areas, is becoming increasingly greater. Climate change, extreme weather and ever higher yields taking place at the same time are creating enormous challenges for areas under cultivation. Drought stress, heavy rains and cultivation of monocultures stress both, the soil and the crops themselves. Regular monitoring of the crops or trees as well as soil condition is essential for a sustainable land use. The use of unmanned aerial vehicles (UAVs) for aerial structural surveys, the recording of soil parameters such as soil temperature, soil moisture and gas exchange have so far mostly been carried out independently of each other. Combining these measurement techniques, a holistic picture of the state of these ecosystems becomes possible. The Fraunhofer-Institute for Physical Measurement Techniques IPM presents a coherent process chain for the fully comprehensive recording of ecosystems. A recording by means of LiDAR systems from the ground, multispectral aerial images, terrestrial laser scans and the recording of nitrous oxide emission. Thus, we obtain a full structural image of the ecosystem enriched with metadata on plant condition and soil parameters. This forms the basis of an analysis of the overall condition of the full ecosystem. We present the results of the different sensors and the fused data of a first measurement campaign.

Olive groves have high socioeconomic impacts in Spain. Global warming is leading to changes in agrarian systems, affecting phenology and productivity. Remote Sensing data is useful to study the current trends in olive groves. The MEDGOLD project, funded by the European Union's H2020 programme (No. 776467), aims to develop climate services for the Mediterranean agri-food sector (grape, olives and durum wheat). In this work, images from the MODIS sensor have been used to study the status of olive groves in 2000-2020 in Southern Spain. Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI) and Normalized Multi-band Drought Index (NMDI) were computed and correlated with precipitation and temperature using datasets provided by the Environmental Information Network of Andalusia (REDIAM) and scenarios based on the MED-GOLD outcomes. Additionally, some annual indicators were derived from NDVI to study trends: maximum (MAX), minimum (MIN), mean (MEAN), relative range (RREL), date of the maximum (DMAX) and date of the minimum (DMIN). Reported good and bad production years were reflected on vegetation indices. Trends were observed in annual indicators. The average photosynthetic activity increased, especially in MIN (Δ_year = 0.0025; r2 = 0.68). The trend of RREL (Δ_year = -0.012; r2 = 0.52) indicated that vegetation is moving to a more constant seasonal behaviour. The beginning and end of the season tend to occur earlier each year, as showed by DMAX (Δ_year = -2.7 days) and DMIN (Δ_year = -1.7 years). Management practices may require adaptations to the new seasonal behaviour of herbaceous vegetation, which could affect soil properties.

Predicting species’ suitable habitats is critical to biodiversity conservation planning and implementation. Species habitat distribution is closely linked to environmental and bioclimatic variables which are widely used for estimating habitat suitability (HS) from species distribution models (SDMs). Integration of environmental parameters derived from satellite remote sensing, bioclimatic variables, and edaphic properties has created an advanced way to improve the SDM performance. The objective of this study is to assess the performance for predicting the potential HS of the arid plant species using Maximum Entropy (MaxENT) species distribution model based on an ecological niche machine-learning algorithm. Prosopis cineraria (Ghaf) in the United Arab Emirates (UAE) was selected for the model simulation. The Ghaf tree is a keystone species to prevent desertification and improve soil fertility in arid environments. We have selected 33 environmental parameters, including satellite remote sensing data (MODIS NDVI, LST, and PET), WorldClim bioclimate variables and static edaphic properties (topography, elevation, soil quality) along with 100 field observations. Collinearity within the bioclimatic variables was eliminated using Pearson correlation. The variables with zero percentage contribution were eliminated for the final model simulation. To evaluate the contribution of environmental parameters to the performance of MaxEnt, we used three scenarios: a) All key predictor variables, b) Only bioclimatic variables, and c) without remote sensing variables. With scenario a) model simulation has substantially improved the potential HS prediction with mean AUC value 0.98, indicating a better predictive accuracy in the integration of satellite remote sensing data. MaxENT results showed that elevation, precipitation of coldest quarter, NDVI, and precipitation of warmest quarte had a significant contribution to the potential HS of Ghaf trees in the UAE. Model results showed that the spatial proportions of the potential HS in the UAE consisted of high (2%), medium (3.7%) and low (94.3%) suitability classes.

Semiarid grasslands are sensitive to drought and are one of the most threatened ecosystems by climate change. In this work, we proposed to assess the performance of two different vegetation indices (VIs) in two zones of a semiarid area with different agro-ecological characteristics. We calculated the VIs and climate anomalies and then attempted to identify and characterize their dynamics with recurrence techniques, a nonlinear method. In this study, the Normalised Difference vegetation index (NDVI) and the Modified Soil Adjusted Vegetation Index (MSAVI) were used. The minimum temperature and precipitation series in both areas were also extracted, as they are the key driving factors of the system. The original series was seasonally adjusted by subtracting the average per date to obtain the anomalies series. On this new set, recurrence plots (RPs), cross recurrence plots (CRPs), and recurrence quantification analysis (RQA) were computed to achieve this goal. RPs are proposed as a method to reveal the periodic or chaotic behaviour of a system. CRPs are a bivariate extension of RPs, and they are computed to analyse the relationships between two variables of the same system. Both are quantified by the RQA, obtaining measures of complexity. We have found that RPs allow visualising different VIs anomalies patterns in each zone. Furthermore, the CRPs revealed the VIs sensitivity to detect and differentiate local conditions. Overall, we have characterised and measured the dynamics of the VIs anomalies and we have shown that recurrence techniques are a valuable tool to explore drought events in semiarid areas.

Cover crop in olive orchards is an increasingly applied soil and water conservation strategy, supported by European policies due to its multiple environmental benefits. To quantify these benefits, supervise and encourage the adoption of this practice, robust and affordable monitoring indicators of the cover crop dynamic and its biomass are required. This work represents the first attempt to estimate the biomass produced by olive grove cover crops based on remotely sensed data and an adaptation of the Monteith efficiencies approach. Ten olive tree fields were selected, distributed in three zones of Southern Spain. They comprised a high environmental variability and differed in the herbaceous layer management: cover crop in strips; non-tillage without strips (full coverage); and conventional tillage. An adaptation of the LUE (Light Use Efficiency)- model was applied to estimate Net Primary Production (NPP) using meteorological and Sentinel-2 data and subtracting the contribution of the wooded vegetation from the ground spectral response. The results showed an uneven adjustment in different fields. RMSD was equal to 650 kg ha-1, with an MBD of -17 kg ha-1, indicating a moderately high error (around 39%) but not too much bias. This error suggests that the model requires further refining, including the adjustment of model parameters to better represent this agrosystem. However, the evolution of biomass accumulation throughout the cover crop growing season and the behaviour of the daily biomass production provided interesting keys about the cover crops’ phenology and management, supporting the discrimination between management practices.

New techniques for agriculture science are widely explored since several decades in order to improve production yield. Measurements of optical properties at different scales of the crop are investigated and exploited to assess different parameters of interest such as state of stress. For instance, nowadays, there exists acquisition systems embedded in drones, mobile machines and satellites that are able to collect huge amount of hyperspectral imaging data. Identification of optical signature extracted from these techniques can help agronomist with adapting irrigation or distinguishing different plant varieties. These techniques allow to improve greatly the agricultural management, however they do not provide information about the internal structure of the plant leaf and their interaction with electromagnetic fields. Knowing precisely the plant leaf structure can bring critical information that can lead to the development of new techniques for phenotyping and precocious stress detection. To do this it is necessary to probe the plant at the leaf scale using THz instead of optical frequencies because the scattering sensitive phenomenon for plants is more drastic at optical frequencies. To find out how the light interact with the leaf, in a deterministic way, we can model the vegetal tissue as a stack of different physical layers characterized by the thickness and the optical index. In this study, funded by ANR project OptiPAG, we use a well-known reverse engineering technique to retrieve leaf architecture from the reflection data. In time domain, a short Terahertz pulse illuminates a multilayer sample that reflects a part of the signal carrying information about the sample structure. Using a numerical fit in the frequency domain allows to identify each layer and deduce the respective optical index over the input frequency range. We use a few classical (inorganic) etalon samples and analyze the echoes to reveal their thicknesses under the assumption of negligible absorption. Then, we use reverse engineering technique to fit the data in the THz range by taking into account the absorption, making an excellent agreement with the previous results with more accuracy. The measured thickness of the samples correspond very well with the manufacturing specifications. And finally we use this technique with vegetal tissues (sunflower leaves), that poses a much more complex situation. Results emphasize a 8-layer stack including trichomes, cuticules, epidermis and mesophyll layers and for each layer we extract the thickness and the complex index. To our knowledge this is the first time that the leaf multilayer structure is extracted with accuracy using a non-contact techniques.

The high-fidelity estimation of the light penetration depths of dry and wet sand-textured soils is of considerable interest for applied remote sensing and geoscience research initiatives involving a wide range of landscapes, from deserts and arable fields to coastal habitats. These initiatives include the restoration of vegetation in arid regions and the mitigation of weed dissemination in agricultural areas covered by wind-transported layers of these soils. Similarly, the remote detection and analysis of hyperspectral signatures from subsurface targets located in sandy landscapes also requires a sound understanding about the light penetration properties of the covering particulate materials under dry and wet conditions. Despite their relevance, however, there is a noticeable lack of data on the light penetration depths of sand-textured soils, notably accounting for their sensitivity to distinct patterns of water presence, either in their pore space or forming films around their grains. In this work, we aim to make inroads, both qualitatively and quantitatively, toward the understanding of key aspects associated with these interconnected processes. In order to achieve this goal without being constrained by laboratory and logistics limitations, we performed an array of controlled in silico experiments to systematically evaluate the effects of distinct water saturation states on the light penetration depths of representative samples of sand-textured soils. Our investigation was centered at the 400 to 1000 nm spectral domain, relevant for studies involving the mineralogy and morphology of natural sands, and it was carried out employing a first-principles simulation framework supported by actual measured data. By advancing the current knowledge in this area, our findings are expected to contribute to the development of new technologies aimed at the cost-effective monitoring and management of landscapes covered by natural sand deposits, and at the acquisition of more precise data on fundamental biophysical phenomena (e.g., seed germination) with a direct impact on crop yield and the recovery of ecosystems affected by the expansion of arid terrains.

In areas with extensive, nomadic, or transhumant livestock farming, it is important to access regular information on the location of ephemeral surface water bodies. Existing near-real time methods for high-resolution surface water mapping are mainly based on the use of optical satellite imagery. However, the use of optical data restricts the water detection to cloud-free conditions. To overcome this limitation SAR data are used for water bodies mapping. Nevertheless, the implemented techniques are usually not fully automated or are not applicable in hilly landscapes. Indeed, surface roughness, hill shadows, and presence of vegetation are known to affect the backscatter and lead to false alarms. In this study, a SAR-based method was used to map surface water from a set of Sentinel-1 images using the Otsu Valley Emphasis method to automatically detect a threshold for water in the histogram of backscatter. In order to reduce the false alarm rate in the steep areas, five different water masks using terrain and drainage information with different thresholds are compared in the mountainous province of KwaZulu-Natal (KZN) in South-Africa. The quantitative assessment shows that the overall accuracy ranged between 0.865 and 0.958 with the highest value obtained with the HAND (Height Above the Nearest Drainage)-based mask with a threshold of 10m. This mask also minimized the false detection of water with the lowest specificity of 0.037. The visual inspection over two reservoirs (Midmar Dam and Wagendrift Dam) shows that there is high agreement between the produced map and the reference data despite differences in their spatial and temporal coverage. Besides, radiometrically terrain corrected SAR data, which could be advantageous in such landscapes were recently made available by the ASF vertex platform. Even though they are not available in NRT, the potential of using such data for water detection is investigated.

Long-term reconstruction or prediction of spatial and temporal information on land or terrestrial water storage (TWS) dynamics globally is critical and highly challenging. A hybrid approach was proposed to combine the strengths of physically-based modeling and deep learning for estimating global TWS anomalies (TWSA). Specifically, we developed a spatiotemporal attention-based deep learning model (STAU-Net), integrating the U-Net architecture with ConvLSTM layer and convolutional block attention module (CBAM) to learn the spatiotemporal patterns of TWSA observed by GRACE, driven under different predictor combinations, e.g., meteorological forcings, soil properties, and modeled TWSA. Once trained and validated, the model can estimate long-term global TWS dynamics without requiring GRACE TWSA as inputs. The evaluation results suggest that the hybrid approach can provide improved predictions of global TWSA compared to others. This study demonstrates the unique ability of the hybrid approach in global freshwater availability monitoring and prediction.

Estimation of leaf area index (LAI) is of vital importance to improve the prediction accuracy of crops quality and yield. However, it is more difficult to precisely assess LAI at the late growth stages of crops due to the influences of leaf senescence and soil background. Unmanned aerial vehicles (UAVs), with hyperspectral sensors onboard, can acquire high spatial and spectral resolution images and provide detailed information of fields, and consequently, are widely used for monitoring the biophysical parameters of crops in precision agriculture. The aim of this study was to evaluate the potential of UAV-based hyperspectral data in LAI estimation for sunflower and maize at the milk-filling stage, with machine learning regression algorithms (MLRA) for data analyses. Three algorithms including linear regression (LR), partial least square regression (PLSR) and kernel ridge regression (KRR) were used with the individual vegetation index (VI), VI-combination and spectral reflectance of full wavelengths as input variables. Results indicate that from the perspective of accuracy of estimation models, the PLSR based on VI-combination derived from hyperspectral images outperformed the LR based on individual VI and KRR based on VI-combination or spectral reflectance, which was proven to be the most suitable for the LAI estimation for both maize and sunflower at late growth stage, with 68% and 64% of the variation in LAI were explained, respectively. From the perspective of VIs tested, the modified triangular vegetation index (MTVI1) and improved soil-adjusted vegetation index (MSAVI) were found to be the best LAI estimators for maize and sunflower. Meanwhile, the contributions of the two VIs were also superior over other VIs tested in developing estimation models based on the PLSR method.

Iron chlorosis in soybean is a nutrient deficiency condition with general symptoms of chlorosis (yellowing) of soybean foliage and stunting of the plant, in turn impacting crop yield. Identifying, selecting and advancing varieties offering resistance to iron chlorosis is a critical component of soybean breeding. Genetic characterization of various soybean varieties is carried out using phenotypic measurements that are collected manually. Such measurements are extremely subjective confounded with rater variability, compromising measurement quality. Furthermore, manual data collection is labor intensive and expensive. In this study, we propose an automatic scoring system employing an analytical framework that applies image processing and machine learning (ML) techniques on red-green-blue (RGB) color channel images collected via Unmanned Aerial Vehicle (UAV) for quantifying iron chlorosis severity. Results from the machine learning model indicate that the ML-based scores yielded good correlation with the manual scores. Additionally, ML scores demonstrated higher heritability/repeatability compared to those obtained from the manual scores, suggesting the use of UAV imagery in conjunction with machine learning approaches for field assessments of iron chlorosis, reducing long and tedious manual data collection efforts. Moreover, such approaches provide a scalable and high-throughput scoring system, enabling efficient breeding practices.

To determine the degree of degradation of agricultural lands for a key values (humus content, mobile potassium, mobile phosphorus, PH), the use of multispectral UAV materials synchronized with ground-based spectrometric imagery is proposed. Spectroradiometer HandHeld 2, soil acidity (pH) meter, satellite GLONASS-GPS receiver of geodetic class were used for field survey. Multispectral orthophoto obtained at the time of ground surveys using multispectral cameras Tetracam Micro-MCA 4 and Tetracam ADC-micro installed on board of the Supercam-S350F UAV. In parallel with the spectrometric work, samples of soils of different soil varieties and washout degree were taken, in representative sites of elementary soil areas. Laboratory studies were carried out with the selected samples, in order to determine the main agrochemical parameters: humus (%), mobile phosphorus (mg), mobile potassium (mg), pH (H₂O). The work was tested on two field sites located in the Chuvash Republic (Russia), on cultivated (arable land) forest-steppe zonal soils (leached chernozems, dark gray forest soils). As a result of mathematical data processing, statistically significant relationships were obtained between certain groups of agrochemical indicators and spectral data in different channels of UAV images for specific soil varieties. In the course of the study, relationships were found between the green, NDVI, NIR, red channels obtained using the Supercam-S350F unmanned aerial vehicle and laboratory data: humus, phosphorus, potassium and soil pH. In general, the results of the experiment prove the fundamental possibility of using multispectral UAV materials, together with ground spectrometric imagery for automated express determination of agrochemical indicators of agricultural lands.

Soil moisture content (SMC) is a key parameter of environmental processes. Remote sensing provides effective methods for mapping SMC at different spatial resolutions. Using UAS-borne hyperspectral observations enables a SMC retrieval at sub-meter scales. Radiative transfer models (RTMs) such as ProSAIL or Scope include a SMC specific input variable and are thus a potential tool to derive SMC and avoiding extensive reference SMC measurements. The inverse application of RTMs supplies information on SMC and plant traits. Scope and ProSAIL involve SMC data of the root zone and at the surface, respectively. The combined use of both models offers the possibility to derive SMC at two vertical depths. Moreover, SMC relevant vegetation proxies such as leaf water content can be retrieved and alternatively used as indicator for SMC. Such plant traits are highest correlated to SMC at depths of major water uptake. However, their response can have a significant time-lag. We analyze the derivation of SMC at the soil surface and at the root zone using the SMC parameters within existing RTMs. As a first step, we investigate on the sensitivity of ProSAIL and Scope to their soil moisture parameters. We apply these findings on UAS-borne hyperspectral and TIR imagery acquired over a pre-alpine TERENO grassland area. The site is equipped with a SoilNet that measures SMC at different depths. Using this data, we assess the vertical extent of both soil moisture content parameters. By inverse modelling of the vegetation parameters and the use of the temporally continuous SoilNet data at root zone level, we analyze the time-lag between changes in SMC and the corresponding plant trait response to optimize the retrieval of SMC.

This article aims to analyze agronomic drought in a highly anthropogenic semi-arid region. This is the western Mediterranean region. The study uses satellite data from the Moderate Resolution Imaging Spectroradiometer (MODIS) and Advanced Scatterometer (ASCAT) describing the dynamics of vegetation cover and soil water content through the Normalized Difference Vegetation Index (NDVI) and the Soil Water Index (SWI). An analysis of the vegetation anomaly index (VAI) highlights the difference between agricultural and natural areas. Thus, two land use classes are considered for the analysis of drought indices, agricultural areas and natural areas. The contribution of vegetation cover (VAI) was combined with the effect of soil water content using the moisture anomaly index (MAI) through a new drought index called the global drought index (GDI). This index considers the seasonal effect of the development of vegetation cover and soil water content with variable weightings over time for the two indices VAI and MAI.

In the present study, we evaluate the potential of multi-incidence L-band and C-band data to retrieve soil moisture. In -situ measurements were acquired during satellite acquisitions over cereal fields in the Kairouan plain in central Tunisia (semi-arid area). Analysing radar data, L-band Advanced Land Observing Satellite-2 multi-incidence data (28°, 32.5° and 36°) in HH (L-HH) and HV (L-HV) polarizations and C-band like-polarization Sentinel-1data, with an incidence angle of approximately 39°, (C-VV) are strongly impacted by soil roughness. In addition, results highlight the sensitivity of L-band data to soil moisture in dense cover class where Normalized Vegetation Difference Index (NDVI) values are higher than 0.6. Two options of Water Cloud Model (WCM) were used (with and without the integration of soil-vegetation interaction component) to simulate radar signal over cereal fields. Each option of WCM was coupled to the best performance bare soil backscattering models. By inverting WCM, results underline the important contribution of soil-vegetation interaction component to estimate soil moisture with L-HV data compared to a neglected impact on C-band data inversion accuracy and stable accuracy in L-HH.

Quantification of Root-Zone Soil Moisture (RZSM) is crucial for agricultural applications. It impacts processes like vegetation transpiration and water percolation. The surface soil moisture (SSM) can be assessed through active and passive microwave remote sensing, but no current sensor enables direct retrieval of RZSM. Spatial maps of RZSM can be retrieved via proxy observations (vegetation stress, water storage change, surface soil moisture) or from land surface model predictions. Recently, more interest has risen in the use of data-driven methods to predict RZSM. In this study, we investigated the use of physical-process based features in the context of Artificial Neural Networks (ANN). We integrated the infiltration process information into an ANN model through the use of the recursive exponential filter. We also used a remote sensing-based evaporative efficiency as an input feature. It is important to note that these two processes depend on surface soil moisture which can be assessed through remote sensing. The impact of the use of geophysical variables was also assessed through the use of surface soil temperature and Normalized Difference Vegetation Index (NDVI). At each step of the study, the ANN models were trained using either only in-situ surface soil moisture data provided by the International Soil Moisture Network (ISMN) or an additional geophysical or processbased feature. The results show that the use of more features in addition to SSM information improves the prediction accuracy in specific cases when compared to an ANN model that predicts RZSM based on only SSM. The ability of the developed models to predict RZSM over larger areas will be assessed in the future.

Desertification and land degradation have severe negative effects on land-use, water resources, soil stability, agriculture and biodiversity. Especially, drylands cover 33.8% of northern Mediterranean countries: approximately 69% of Spain and 66% of Cyprus. The European Environment Agency (EEA) indicated that 8% of the territory of the European Union (mostly in Bulgaria, Cyprus, Greece, Italy, Spain and Portugal) experience a ‘very high’ or ‘high sensitivity’ to desertification. For Cyprus Island, 9.68% of the land area was found to be susceptible to land degradation. The objective of this literature review is to provide a detailed synthesis of the main contributions of the global vegetation phenology research to the development of environmental knowledge, based on land degradation/ desertification and Earth observation (EO)-based science and technology. The study identifies the current fields of research and possible research gaps. To achieve this, more than 700 scientific papers were screened from which approximately 549 papers were reviewed, identifying and the state of land surfaces and vegetation phenology with remote sensing data. Most of the studies have as a central research object direct human-induced land degradation or the degradation of anthropogenic-modified landscapes, without having considered long-term un-altered natural vegetation, in order to assess the impact and the level of climate change. Hence, a detailed EO-based time-series monitoring and analysis of un-altered natural vegetation could provide indicators that may serve as early warning signals for the scale and level of climate change induced effects on vegetation and ecosystems that might lead to land degradation and even to desertification.

Irrigation enterprises who manage the irrigated water distribution need to anticipate farmer’s demands to minimize the evaporation amounts from irrigation pools, as well as storing enough water to accomplish crop’s water needs at any time and farmer’s management approach. Crop coefficient Kc has proved to be essential when estimating evapotranspiration with FAO56 procedure, but it varies locally. The purpose of this study is to estimate the local corn crop coefficient with remote sensing to estimate the water crop needs in 164 plots in an area of the northwest of the Iberian Peninsula and to identify different farmer’s ways to manage the land. For this purpose, 25 images from Sentinel-2 were analyzed to create their NDVI images. Therefore, the temporal Kc values were estimated and a Kc-curve for each corn field was calculated. Results allowed to differentiate the four crop growth stages and their corresponding Kc values for the study area. Besides, the 164 corn fields were clustered into 31 groups according to their different Kc curves as a result of farmer’s management. Therefore, the method has proved to help in the future to anticipate the local irrigation needs of the corn crops and to improve the farmer’s assessment to reduce their water demands without diminishing their crop production.

The intensity of flowering of the holm oak trees is important for the annual phenological monitoring and as a predictive index of final acorn production. Their male flowers present in long catkins of intense yellow color and the estimation of their abundance in the field is a time-consuming task that becomes unfeasible at a large scale. In this work, a methodology to estimate the intensity of flowering of oak trees using RGB (Red Green Blue) images, provided by an unmanned aerial vehicle, was tested. During the spring of 2019, three aerial zenith images of 3 cm spatial resolution were taken in three selected dehesa sites, together with simultaneous ground digital photographs per tree (50 at each site). The intensity of flowering was visually estimated using the ground digital photographs in three categories, ranging from 1 (little or no flowering) to 3 (high flowering). A simple flowering intensity index, based on the closeness to pure yellow within a Cartesian RGB space, was developed to check the relationship between the drone images and the visually analyzed photographs. The results showed that those trees with lower flowering intensity were grouped in higher yellow distances and the high flowering intensity trees in the lower ones. As a result, it can be concluded that this index was able to identify qualitatively the flowering intensity of holm oaks at the farm level and could be useful for future phenological or productivity applications.

Pollution of the Arctic territories with garbage dumps provides the general warming in the northern latitudes and cooling in the southern latitudes of the Earth. This article examines the state of the cryosphere of the studied territories and the impact on the constituent elements of solid domestic and industrial waste. The necessary information of medium, high spatial resolution for further study can be obtained using technologies for remote sensing of the Earth from spacecraft with hyperspectral measurements. We propose a method for detecting leachate elements in unauthorized dumpsites in the Arctic using space vehicles. This task is relevant for the implementation of geo-ecological monitoring of the Arctic territories covered with snow. An algorithm for finding the creation of leachate under the influence of solid household and industrial waste has been developed. The article examines the consequences of climate change on forming the biomedical component of the process under consideration. We present a comparison of the proposed processing algorithm on the space images of the Arctic and subarctic territories of the Russian Federation.

Due to depletion of natural resources, climate change and their impact on the land-production systems, farmers are facing more and more challenges related to the practical application of the sustainable development paradigm. These problems result in rapid development of precision agriculture as a management strategy, taking advantage of state-of-theart technologies. In precision agriculture, Variable Rate Application (VRA) technology is focused on the automated application of materials (such as fertilizers, herbicides, and irrigation water) to a given crop field. It involves different approaches, including sensor-based systems for monitoring and assessment of crop status and field environmental conditions. For operational success of VRA reliable data is needed to indicate the variety of processes taking place in the farm field. In the present research, we present spectral signature data for the status of winter wheat (Triticum aestivum L.) development in different growth stages. Spectral signatures vary depending on environmental conditions and related effects for the agroecosystems such as drought stress, crop diseases, and crop nutrient deficiencies. The generated spectral signature profiles are based on the Sentinel-2 satellite data, acquired in three consecutive growing seasons, distinguished with different ecological conditions. Spectral vegetation indices, indirectly representing the manifestation of biophysical processes and drought stress are calculated for each profile. Field climatic data is used for differentiation of the ecological conditions and validation of the results. The present research supports the creation of spectral library and can be used to create machine learning algorithms for monitoring of winter wheat status and application of variable rate technology.

Agricultural and forestry satellite for agriculture and forestry monitoring are scheduled to be launched in the Republic of Korea in 2025. The Agricultural and Forestry Satellite CAS500(Compact Advanced Satellite 500)-4 is a multi-spectral satellite with a spatial resolution of 5 m and with a revisit cycle of 3 days. Prior to launch, this study intends to develop a NDVI composite technique to minimize the effect of clouds. A high-altitude Korean cabbage field (<67ha), which has a relatively large area as a single crop field in Korea, was selected as the study area. Sentinel-2A/B (10m spatial resolution, 5-day revisit cycle) acquired from May 2019 to July 2021 for the study area was used. For monthly compositing, the MaxNDVI technique, which is a representative composite technique, and the recently suggested score-based composite technique were applied and compared. The score-based method calculates the fitness score for compositing for each pixel by assigning various factors and weights to minimize the effect of clouds during NDVI composite and maximize temporal representativeness. Therefore, the reflectance of the pixel with the highest score is used for compositing. The reflectance composite image produced in this way is converted to NDVI. Although both composite techniques minimize the effect of clouds, both results show that MaxNDVI shows high NDVI at the end of the month at the time of early growth after sowing, whereas the score-based technique shows NDVI at the middle of the month. Compared to the MODIS composite data from 2019 to 2021, the monthly composite data of Sentinel-2 NDVI showed various growth patterns by site in more detail.