Recent advancements in hyperspectral satellite technology, including sensors like EnMAP, are promising for monitoring grassland ecosystems at the landscape scale. These developments include detailed vegetation analysis capabilities, crucial for understanding plant traits and species composition. However, technical restrictions of recent hyperspectral satellite missions can hinder comprehensive coverage of research areas resulting in a temporal mismatch of field measurements and satellite data. Here we utilize hyperspectral data from the DESIS, PRISMA, and EnMAP mission, along with field measurements from 74 grassland plots of the German Biodiversity Exploratories in Schorfheide-Chorin and Hainich, collected in May 2020 and 2023. We focus on the impact of different satellite sensors and their acquisition timing grouped within five phenological seasons (early April to August) on the accuracy of biomass and species composition models using Partial Least Squares Regression (PLSR) and Procrustes randomization tests. Additionally, we are evaluating the effectiveness of two spectral transformations in improving model accuracy and reliability. Our findings reveal significant differences in the relationship of hyperspectral satellite data with grassland biomass and species composition. Even though comparison of DESIS biomass models indicated that sensor data from the beginning of April achieved best results for biomass (R2 = 0.48), sensor data covering the SWIR from late April and June showed slightly better modeling results (EnMAP: R2 = 0.51, PRISMA: R2 = 0.53). Species composition was significantly related to spectral composition, with sensor data from late April showing the strongest relationships. The performance of sensors, including VNIR and VNIR-SWIR, was almost equal (e.g., DESIS: R2 = 0.54, PRISMA: R2 = 0.59). Overall, the results highlight the benefit of SWIR bands for biomass modeling, while their importance was minor in relation to species composition. A trend of improved model performance was observed with mean normalization for EnMAP and PRISMA data, while continuum removal led to a decrease in performance. Our study underscores the critical role of temporal and spectral data selection in improving grassland models, suggesting potential pathways for refining remote sensing approaches in ecological monitoring and management.

Earth observation (EO) provides a powerful tool for evidence-based policy-making in the European Union (EU) and globally. We are entering a golden age of EO science, in which the availability of high-quality, everincreasing data converges with the growing demand for monitoring land-use changes and evaluating the impact of policies. This comes at a crucial time, as the EU has recently adopted ambitious environmental and agricultural policies, including the Common Agricultural Policy (CAP), the Regulation on Land Use, Land Use Change and Forestry (EU LULUCF), and the Regulation on Deforestation-free Products (EUDR), whose implementation and effectiveness will rely on robust EO-based monitoring tools. Encouragingly, existing EO capacities, particularly through the Copernicus Programme, already provide a strong foundation for policy support. However, our synthesis reveals that current information products only partially meet the monitoring and compliance needs of these policies. Fully unlocking the potential of EO will require advancements in temporal resolution (approaching near-real-time), spatial resolution (<= 5 m), and rigorous uncertainty quantification. The forthcoming Copernicus Expansion and Next Generation missions can help close these gaps, particularly if new products are co-designed with stakeholders and end-users. By strategically aligning technical innovations with policy priorities, EO stands to become a transformative enabler of the EU's sustainability ambitions.

Increases in the frequency and intensity of droughts and heat waves are threatening forests around the world. Climate-driven tree dieback and mortality is associated with devastating ecological and societal consequences, including the loss of carbon sequestration, habitat provisioning, and water filtration services. A spatially finegrained understanding of the site characteristics making forests more susceptible to drought is still lacking. Furthermore, the complexity of drought effects on forests, which can be cumulative and delayed, demands investigation of the most appropriate meteorological indicators. To address this research gap, we investigated the drivers of drought-induced forest damage in a particularly drought-affected region of Central Europe using SHapley Additive exPlanations (SHAP) values, an explainable artificial intelligence (XAI) method which allows for the relevance of predictors to be quantified spatially. To develop a reproducible approach that facilitates transferability to other regions, open-source data was used to characterize the meteorological, vegetation, topographical, and soil drivers of tree vulnerability, representing 41 predictors in total. The forest drought response was characterized as a binary variable ("damaged" or "unchanged") at a 30-m resolution based on the Normalized Difference Moisture Index (NDMI) anomaly (%) between a baseline period (2013-2017) and recent years (2018-2022). We revealed critical tipping points beyond which the forest ecosystem shifted towards a damaged state: <81 % tree cover density, <4% of broadleaf trees, and < 24 m canopy height. Our study provides an enhanced understanding of trees' response to drought, which can support forest managers aiming to make forests more climate-resilient, and serves as a prototype for interpretable early-warning systems.

Heat stress is the leading climate-related cause of premature deaths in Europe. Major heatwaves have struck Europe recently and are expected to increase in magnitude and length. Large cities are particularly threatened due to the urban morphology and imperviousness. Green spaces mitigate heat, providing cooling services through shade provision and evapotranspiration. However, the distribution of green cooling and the population most affected are often unknown. Here we reveal environmental injustice regarding green cooling in 14 major European urban areas. Vulnerable residents in Europe are not concentrated in the suburbs but in run-down central areas that coincide with low-cooling regions. In all studied areas, lower-income residents, tenants, immigrants and unemployed citizens receive below-average green cooling, while upper-income residents, nationals and homeowners experience above-average cooling provision. The fatality risk during extreme heatwaves may increase as vulnerable residents are unable to afford passive or active cooling mitigation.

In this study, hyperspectral data of two geographically distant study areas in Germany were analyzed to find the most effective spectral wavelengths for red-edge-based vegetation indices to differentiate bark beetle-infested Norway spruce into three infestation classes, including green-attack. Considering that the relationship between red-edge position (REP) and reflection shoulder position (RESP) is crucial for the relative quantification of chlorophyll concentration and LAI and their changes due to bark beetle infestation, we tested normalized difference red-edge indices (NDRE) formed from these parameters. Using combinations of 75 spectral bands in the range 685 nm to 850 nm, we analyzed 1187 different NDREs, and two additional vegetation indices, and show that the sorted, normalized Kruskal-Wallis H-value, followed by a frequency analysis of the RESP and REP bands of the 10 % of the highest ranked NDREs is an effective way to identify the global optima of these parameters. The optimal REP and RESP estimated using this approach were determined at 714 nm and 758 nm, respectively. The derived normalized difference red-edge index NDRE758_714 could be successfully applied to both test sites. Moreover, we found that aggregating hyperspectral bands into broader bands - 703 nm to 729 nm for a REP band and 741 nm to 767 nm for a RESP band - did not degrade model accuracy when used in a multispectral NDRE, suggesting that a hyperspectral camera is not required to perform the task. Multinomial logistic regression (MNLR) and random forest (RF) classification models were successfully transferred between the two geographically distant study sites. The predictive accuracies of the models for the test data indicated MNLR to be more robust than the RF. With the MLNR models applied, overall accuracies 77.4 % were achieved in the first study area imaged during the green-attack core period, and 88.2 % in the second study area imaged at the final green-attack stage. Our results contribute to forest health monitoring with imaging spectroscopy data and provide practical recommendations for sensor design with broader bands.

ABSTRACTIndividual approaches to observe water dynamics across our landscape, from the land surface to groundwater, are many though they individually only provide glimpses into the real world due to their specific space–time scales. Comprehensive integration across all available observations is still largely lacking, limiting both our ability to reduce scientific knowledge gaps, and to guide land and water management using the best available scientific evidence. We argue that a stronger focus on integration of observational products, while utilising machine learning and accounting for current perceptual understanding is urgently needed to overcome this limitation. Since Europe is warming faster than any other continent, central Europe is undergoing a dramatic hydroclimatic transition about which such integrated observations would provide timely and valuable insights. Here, we present potential and gaps of current and planned observational methods. We argue that hyperresolution (sub km) integrated estimates of landscape water dynamics are feasible, which could significantly improve our ability to simulate vadose zone and groundwater dynamics, ultimately closing gaps in our current perception of hydrological processes in a temperate region under strong influence from climate change. We close by arguing that an interdisciplinary effort of various scientific communities is needed to enable this advancement.

Abstract. The major challenge of multiscale measurement methods beyond the point scale is their complex interpretation in the light of landscape heterogeneity. For example, methods like cosmic-ray neutron sensing, remote sensing, or hydrogravimetry are all able to provide an integral value on the water storage, representative for their individual measurement volume. A rigorous assessment of their performance is often hindered by the lack of knowledge about the truth at their corresponding scale, given the high complexity and detail of natural landscapes. In this study we suggest a synthetic virtual landscape that allows for an exact definition of all variables of interest and, consequently, constitutes the so-called "virtual truth" free of knowledge gaps. Such a landscape can be explored in various "virtual field campaigns" using "virtual sensors" that mimic the response and characteristics of actual devices. We use dedicated physically-based models to simulate the signal a sensor would receive. These model outputs termed "virtual observations" can be explored and also allow the reconstruction of water storage, which can then readily be compared to the "virtual truth". Insights from this comparison could help to better understand real measurements and their uncertainties, and to challenge accepted knowledge about signal processing and data interpretation. The "Virtual Joint Field Campaign" is an open collaborative framework for constructing such landscapes. It comprises data and methods to create and combine different compartments of the landscape (e.g. atmosphere, soil, vegetation). The present study demonstrates virtual observations with Cosmic Ray Neutron Sensing, Hydrogravimetry, and Remote Sensing in three exemplary landscapes. It enables unprecedented opportunities for the systematic assessment of the sensor’s strengths and weaknesses and even their combined use.

Central Europe experienced severe droughts from 2018 to 2020, which led to bark beetle infestations in Norway spruce (Picea abies L.). While recent studies have identified these droughts as a major factor in the dieback of spruces, the exact mechanism through which drought stress influences spruce vitality and increases vulnerability to infestations is not well understood. To gain insights into the spatial and temporal relations between infestations and drought stress, this study compared soil moisture index (SMI) data for the entire soil layer from the German Drought Monitor and Sentinel-2 satellite time series for healthy and bark beetle infested spruce stands preceding the infestation. The study used t-tests to compare pre-infestation monthly sums of SMI and vegetation indices (VIs) for healthy and infested stands. Yearly trajectories of six VIs were used to compare the separability between infested and healthy plots. The results showed that the monthly sum values of SMI for healthy spruce plots were statistically significantly higher than the infested plots from May 2017 to January 2018, while the monthly sum values of Inverted Red Edge Chlorophyll Index of healthy spruce plots were significantly higher than infested plots during growing seasons. With these findings, the study provides useful information for the understanding of the relation of soil moisture and remote sensing VIs patterns in relation to bark beetle infestations.

Urban Green Infrastructure (UGI) provides ecosystem services such as cooling of temperatures and is majorly important for climate change adaptation. Green Volume (GV) describes the 3-D space occupied by vegetation and is highly useful for the assessment of UGI. This research uses Sentinel-2 (S-2) optical data, vegetation indices (VIs), Sentinel-1 (S-1) and PALSAR-2 (P-2) radar data to build machine learning models for yearly GV estimation on large scales. Our study compares random and stratified sampling of reference data, assesses the performance of different machine learning algorithms and tests model transferability by independent validation. The results indicate that stratified sampling of training data leads to improved accuracies when compared to random sampling. While the Gradient Tree Boost (GTB) and Random Forest (RF) algorithms show generally similar performance, Support Vector Machine (SVM) exhibits considerably greater model error. The results suggest RF to be the most robust classifier overall, achieving highest accuracies for independent and inter-annual validation. Furthermore, modelling GV based on S-2 features considerably outperforms using only S-1 or P-2 based features. Moreover, the study finds that underestimation of large GV magnitudes in urban forests constitutes the biggest source of model error. Overall, modelled GV explains around 79% of the variability in reference GV at 10 m resolution and over 90% when aggregated to 100 m resolution. The research shows that accurately modelling GV is possible using openly available satellite data. Resulting GV predictions can be useful for environmental management by providing valuable information for climate change adaptation, environmental monitoring and change detection.

As globalization continues, the spread of invasive species is accelerating, posing a severe threat to native biodiversity. To manage such species, reduce their negative impact on native biota and utilize management costs efficiently, a profound understanding of their geographical distribution pattern is mandatory. In this study, the species distribution model Maxent was used to predict the potential spatial distribution of U. europaeus. To account for sampling bias, three bias correction methods were applied, including a novel approach to increase the number of presence points by sampling occurrences based on satellite images. Furthermore, a decision structured process was used to evaluate and select optimal Maxent parameterization and account for limitations of single evaluation criteria. The currently suitable area of U. europaeus is primarily distributed in the coastal and central regions of Chilean natural region Zona Sur in south-central Chile. Annual mean temperature (bio1), annual precipitation (bio12), and precipitation seasonality (bio15) were the most important environmental variables that affected the distribution of U. europaeus. The sampling of additional presence points could effectively correct for sampling bias in species occurrence data. The use of a decision structured process for model evaluation proved to be useful in determining optimal model parameterization for decreased model complexity. This study highlights the importance of optimized Maxent calibrations to yield results as accurately as possible. The predicted suitable habitats can inform nature conservation planners and landscape managers to guide and prioritize conservation measures.

Airborne and spaceborne platforms are the primary data sources for large-scale forest mapping, but visual interpretation for individual species determination is labor-intensive. Hence, various studies focusing on forests have investigated the benefits of multiple sensors for automated tree species classification. However, transferable deep learning approaches for large-scale applications are still lacking. This gap motivated us to create a novel dataset for tree species classification in central Europe based on multi-sensor data from aerial, Sentinel-1 and Sentinel-2 imagery. In this paper, we introduce the TreeSatAI Benchmark Archive, which contains labels of 20 European tree species (i.e., 15 tree genera) derived from forest administration data of the federal state of Lower Saxony, Germany. We propose models and guidelines for the application of the latest machine learning techniques for the task of tree species classification with multi-label data. Finally, we provide various benchmark experiments showcasing the information which can be derived from the different sensors including artificial neural networks and tree-based machine learning methods. We found that residual neural networks (ResNet) perform sufficiently well with weighted precision scores up to 79 % only by using the RGB bands of aerial imagery. This result indicates that the spatial content present within the 0.2 m resolution data is very informative for tree species classification. With the incorporation of Sentinel-1 and Sentinel-2 imagery, performance improved marginally. However, the sole use of Sentinel-2 still allows for weighted precision scores of up to 74 % using either multi-layer perceptron (MLP) or Light Gradient Boosting Machine (LightGBM) models. Since the dataset is derived from real-world reference data, it contains high class imbalances. We found that this dataset attribute negatively affects the models' performances for many of the underrepresented classes (i.e., scarce tree species). However, the class-wise precision of the best-performing late fusion model still reached values ranging from 54 % (Acer) to 88 % (Pinus). Based on our results, we conclude that deep learning techniques using aerial imagery could considerably support forestry administration in the provision of large-scale tree species maps at a very high resolution to plan for challenges driven by global environmental change. The original dataset used in this paper is shared via Zenodo (https://doi.org/10.5281/zenodo.6598390, Schulz et al., 2022). For citation of the dataset, we refer to this article.

Drought events and environmental disturbances related to water scarcity have become more severe and frequent, affecting food security and endangering vulnerable biomes. Reliable soil moisture content (SMC) estimations at the landscape scale are therefore essential to understand patterns in drought events and vegetation response to such occurrences. Accurate soil moisture predictions can support actions to mitigate water scarcity effects in vegetation, for instance, by precisely managing crops to avoid further depleting limited water resources. However, most available SM products derived from remote sensing (RS) or meteorological data are supplied at a coarse spatial scale and are unsuitable for heterogeneous landscapes in terms of topography and land cover. The gaps between significant changes in SMC levels at the root zone and the vegetation response during the dry and wet seasons are still unknown. Before defining whether up-scaling (or modelling) in situ data using RS or downscaling coarse images to a landscape scale would resolve this research gap, a better understanding of temporal and spatial contributions and uncertainties of different technologies to SMC products is needed. Despite the advance in sensors and processing capacity, a combination of spatial and temporal resolution required for SMC retrieval is unlikely to be available soon globally. Satellite sensors (e.g. microwave, optical, thermal) present different limitations and rely on proxies and assumptions to indirectly derive SMC at the root zone. Moreover, the relationships across time can be biased by weather conditions, masked by land cover type and clouds, or misled by spurious correlations between meteorological and plant trait variables (phenology). For instance, microwave signals can be affected over dense forests, snow cover, or steep topography. Furthermore, optical data are often unavailable due to cloud cover or have their reflectance drastically change from living vegetation to bare soil between two acquisitions in non-permanent crop fields. Therefore, multi-platform approaches, combining technologies and resolutions to derive a versatile and accurate SMC product, should be prioritized. As the model relies on indirect relationships with plant traits or moisture from the topsoil rather than the underlying hydrological processes, the spatial-temporal patterns (and autocorrelation) should not be neglected as they carry crucial information about water balance. In this study, we analyze 38 soil moisture probes installed in landscapes with different vegetation cover, topography, and soil type in Germany. The SMC measurements are provided by cosmic-ray neutron sensors (CRNSs), a non-invasive technology that provides measurements at a field scale (130 to 240m radius). The CRNS time-series measurements are compared to RS and meteorological products. Auxiliary variables such as precipitation, evapotranspiration, and vegetation parameters (e.g. LAI) are also aligned with the SMC and RS-derived products. The similarity and mismatching of the explanatory time-series patterns compared to the reference SMC for different vegetation cover (forest, grassland, and crops), season, regional characteristics (climate, soil properties, and topography), and resolutions (temporal and spatial) are presented. The results can support the development of a soil moisture retrieval approach at a medium to high spatial resolution based on a data cube combining different RS platforms and auxiliary variables.

Modeling the past or future spread patterns of invasive plant species is challenging and in an ideal case requires multi-temporal and spatially explicit data on the occurrences of the target species as well as information on the habitat suitability of the areas at risk of being invaded. Most studies either focus on modeling the habitat suitability of a given area for an invasive species or try to model the spreading behavior of an invasive species based on temporally or spatially limited occurrence data and some environmental variables. Here we suggest a workflow that combines habitat suitability maps, occurrence data from multiple time steps collected from remote sensing data, and cellular automata models to first reconstruct the spreading patterns of the invasive shrub Ulex europaeus on the island Chiloé in Chile and then make predictions for the future spread of the species. First, U. europaeus occurrences are derived for four time steps between 1988 and 2020 using remote sensing data and a supervised classification. The resulting occurrence data is combined with occurrence data of the native range of U. europaeus from the GBIF database and selected environmental variables to derive habitat suitability maps using Maxent. Then, cellular automata models are calibrated using the occurrence estimates of the four time steps, the suitability map, and some additional geo-layer containing information about soils and human infrastructure. Finally, a set of calibrated cellular automata models are used to predict the potential spread of U. europaeus for the years 2070 and 2100 using climate scenarios. All individual steps of the workflow where reference data was available led to sufficient results (supervised classifications Overall Accuracy > 0.97; Maxent AUC > 0.85; cellular automata Balanced Accuracy > 0.91) and the spatial patterns of the derived maps matched the experiences collected during the field surveys. Our model predictions suggest a continuous expansion of the maximal potential range of U. europaeus , particularly in the Eastern and Northern part of Chiloé Island. We deem the suggested workflow to be a good solution to combine the static habitat suitability information—representing the environmental constraints—with a temporally and spatially dynamic model representing the actual spreading behavior of the invasive species. The obtained understanding of spreading patterns and the information on areas identified to have a high invasion probability in the future can support land managers to plan prevention and mitigation measures.

Cosmic-ray neutron sensing (CRNS) allows for the estimation of root-zone soil water content (SWC) at the scale of several hectares. In this paper, we present the data recorded by a dense CRNS network operated from 2019 to 2022 at an agricultural research site in Marquardt, Germany - the first multi-year CRNS cluster. Consisting, at its core, of eight permanently installed CRNS sensors, the cluster was supplemented by a wealth of complementary measurements: data from seven additional temporary CRNS sensors, partly co-located with the permanent ones; 27 SWC profiles (mostly permanent); two groundwater observation wells; meteorological records; and Global Navigation Satellite System reflectometry (GNSS-R). Complementary to these continuous measurements, numerous campaign-based activities provided data by mobile CRNS roving, hyperspectral im-agery via UASs, intensive manual sampling of soil properties (SWC, bulk density, organic matter, texture, soil hydraulic properties), and observations of biomass and snow (cover, depth, and density). The unique temporal coverage of 3 years entails a broad spectrum of hydro-meteorological conditions, including exceptional drought periods and extreme rainfall but also episodes of snow coverage, as well as a dedicated irrigation experiment. Apart from serving to advance CRNS-related retrieval methods, this data set is expected to be useful for vari-ous disciplines, for example, soil and groundwater hydrology, agriculture, or remote sensing. Hence, we show exemplary features of the data set in order to highlight the potential for such subsequent studies. The data are available at doi.org/10.23728/b2share.551095325d74431881185fba1eb09c95 (Heistermann et al., 2022b).

Bark beetle infestations are among the most substantial forest disturbance agents worldwide. Moreover, as a consequence of global climate change, they have increased in frequency and in the size and number of affected areas. Controlling bark beetle outbreaks requires consistent operational monitoring, as is possible using satellite data. However, while many satellite-based approaches have been developed, the full potential of dense, multi-sensor time series has yet to be fully explored. Here, for the first time, we used all available multispectral data from Landsat and Sentinel-2, Sentinel-1 SAR data, and combinations thereof to detect bark beetle infestations in the Bavarian Forest National Park. Based on a multi-year reference dataset of annual infested areas, we assessed the separability between healthy and infested forests for various vegetation indices calculated from the satellite data. We used two approaches to compute infestation probability time series from the different datasets: Bayesian conditional probabilities, based on the best-separating index from each satellite type, and random forest regression, based on all indices from each satellite type. Five different sensor configurations were tested for their detection capabilities: Landsat alone, Sentinel-1 alone, Sentinel-2 alone, Landsat and Sentinel-2 combined, and data from all satellite types combined. The best overall results in terms of spatial accuracy were achieved with Sentinel-2 (max. overall accuracy: 0.93). The detections of Sentinel-2 also were the closest to the onset of infestation estimated for each year. Sentinel-2 detected infested areas in larger contiguous patches with higher reliability compared to smaller patches. The results achieved with Landsat were somewhat inferior to those of Sentinel-2 (max. accuracy: 0.89). While yielding similar results, the combination of Landsat and Sentinel-2 did not provide any advantages over using Landsat or Sentinel-2 alone (max. accuracy: 0.87), while Sentinel-1 was unable to detect infested areas (max. accuracy: 0.62). The combined data of all three satellite types did not achieve satisfactory results either (max. accuracy: 0.67). Spatial accuracies were typically higher for Bayesian conditional probabilities than for random forest-derived probabilities, but the latter resulted in earlier detections. The approach presented herein provides a flexible disturbance detection pipeline well-suited for the monitoring of bark beetle outbreaks. Furthermore, it can also be applied to other disturbance types.

Calibrating land surface phenology (LSP) with tree rings is important to model spatio-temporal variations in forest productivity. We used MODIS (resolution: 250 m) NDVI, WDRVI and EVI series 2000–2014 to derive LSP metrics quantifying phenophase timing and canopy photosynthetic rates of 26 European beech forests covering a large thermal gradient (5–16 °C) in Italy. Average phenophase timing changed greatly with site temperature (e.g. growing season 70 days longer at low- than high-elevation); average VI values were affected by precipitation. An annual temperature about 12 °C (c. 1100 m asl) represented a bioclimatic threshold dividing warm from cold beech forests, distinguished by different phenology-BAI (basal area increment) relationships and LSP trends. Cold forests showed decreasing VI values (browning) and delayed phenophases and had negative BAI slopes. Warmer forests tended to increase VI (greening), and positive BAI slopes. NDVI peak, commonly used in global trend assessments, changed with elevation in agreement with changes in wood production. A cross-validation modelling approach demonstrated the ability of LSP to predict average BAI and its interannual variability. Merging sites into bioclimatic groups improved models by amplifying the signal in growth or LSP. NDVI had highest performances when informing on BAI trends; WDRVI and EVI were mostly selected for modelling mean and interannual BAI. WDRVI association with tree rings, tested in this study for the first time, showed that this VI is highly promising for studying forest dynamics. MODIS LSP can quantify forest functioning changes across landscapes and model interannual spatial variations and trends in productivity dynamics under climate change.

Transpiration is a fundamental biophysical process, directly measured in plants by dividing sap flow by total leaf area. Under non-limiting conditions, transpiration and reference evapotranspiration (ETo) are hypothesized to be equal when ETo is normalized by the leaf area index of the reference crop, i.e., LAI = 2.88. Known as the E-2.88 model, it has only been tested with ET(o )derived from Penman-Monteith FAO56. Phenological influences on canopy conductance potentially decouple transpiration from atmospheric evaporative demand and lower the accuracy of E2.88. This study tested the accuracy of 20 E-2.88 models in predicting apple (Malus domestica (Suckow) Borkh. var. Granny Smith) and pear (Pyrus communis L. var. Beurre Bosc Pear) transpiration over the 2020-2021 austral growing season. For apple, the Penman-Monteith ASCE-EWRI model had the highest predictive power with 7% error and r(2) = 0.89; whereas for pear the Valiantzas (2018, Eq. (7)) showed 2% error and r(2) = 0.96 evaluated via linear regression. Generally, models that included a humidity parameter had stronger predictive power than models excluding humidity. Yet, the predictive power of the E-2.88 models decreased considering the phenological phases for each crop. For apple, early and late season E(2.88 )models underestimated transpiration by at least 27%. For pear, late season error increased to 7% as the E-2.88 models overestimated transpiration. Canopy conductance and the atmospheric decoupling factor were significantly greater in early and late season for apple and significantly lower in late season for pear. Therefore, phenology decreased the predictive power of the E-2.88 model in early and late season by decoupling physiological processes from atmospheric evaporative demand.

The monitoring of soil moisture content (SMC) at very high spatial resolution (<10m) using unmanned aerial systems (UAS) is of high interest for precision agriculture and the validation of large scale SMC products. Data-driven approaches are the most common method to retrieve SMC with UAS-borne data at water limited sites over non-disturbed agricultural crops. A major disadvantage of data-driven algorithms is the limited transferability in space and time and the need of a high number of ground reference samples. Physically-based approaches are less dependent on the amount of samples and are transferable in space and time. This study explores the potential of (1) a hybrid method targeting the soil brightness factor of the PROSAIL model using a variational heteroscedastic Gaussian Processes regression (VHGPR) algorithm, and (2) a data-driven method employing VHGPR for the retrieval of SMC over three grassland sites based on UAS-borne VIS-NIR (399-1001 nm) hyperspectral data. The sites were managed by mowing (Fendt), grazing (Grosses Bruch) and irrigation (Marquardt). With these distinct local pre-conditions we aimed to identify factors that favor and limit the retrieval of SMC. The hybrid approach presented encouraging results in Marquardt (RMSE = 1.5 Vol_%, R2 = 0.2). At the permanent grassland sites (Fendt, Grosses Bruch) the thatch layer jeopardized the application of the hybrid model. We identified the complex canopy structure of grassland as the main factor impacting the hybrid SMC retrieval. The data-driven approach showed high accuracy for Fendt (R2 = 0.84, RMSE = 8.66) and Marquardt (R2 = 0.4, RMSE = 10.52). All data-driven models build on the LAI-SMC relationship. However, this relationship was hampered by mowing (Fendt), leading to a lack of transferability in time. The alteration of plant traits by grazing prevents finding a relationship with SMC in Grosses Bruch. In Marquardt, we identified the timelag between changes in SMC and plant response as the main reason of decrease in model accuracy. Yet, the model performance is accurate in undisturbed and water-limited areas (Marquardt). The analysis points to challenges that need to be tackled in future research and opens the discussion for the development of robust models to retrieve high resolution SMC from UAS-borne remote sensing observations.

Deriving soil moisture content (SMC) at the regional scale with different spatial and temporal land cover changes is still a challenge for active and passive remote sensing systems, often coped with machine learning methods. So far, the reference measurements of the data-driven approaches are usually based on point data, which entails a scale gap to the resolution of the remote sensing data. Cosmic Ray Neutron Sensing (CRNS) indirectly provides SMC estimates of a soil volume covering more than 1 ha and vertical depth up to 80 cm and is thus able to narrow this scale gap. So far, the CRNS-based SMC has only been used as validation source of remote sensing based SMC products. Its beneficial large sensing volume, especially in depth, has not been exploited yet. However, the sensing volume of the CRNS, which is changing with hydrological conditions, bears challenges for the comparison with remote sensing observations. This study, for the fist time, aims to understand the direct linkage of optical (Sentinel 2) and SAR (Sentinel 1) data with CRNS-based SMC. Thereby, the CRNS-based SMC is obtained by an experimental CRNS cluster that covers the high temporal and spatial SMC variability of an entire pre-alpine subcatchment. Using different Random Forest regressions, we analyze the potentials and limitations of both remote sensing sensors to follow the CRNS-based SMC signal. Our results show that it is possible to link the CRNS-based SMC signal with SAR and optical remote sensing observations via Random Forest modelling. We found that Sentinel 2 data is able to separate wet from dry periods with a R2 of 0.68. It is less affected by the changing soil volume that contributes to the CRNS-based SMC signal and it is able to assign a land cover specific SMC distribution. However, Sentinel 2 regression models are not accurate (R2 < 0.21) in mapping the CRNS-based SMC for the frequently mowed grassland areas of the study site. It requires soil type and topographical information to accurately follow the CRNS-based SMC signal with Random Forest regression. Sentinel 1 data instead is affected by the changing soil volume that contributes to the CRNS-based SMC signal. It has reasonable model performance (R2 = 0.34) when the CRNS data correspond to surface SMC. Also for Sentinel 1 the retrieval is impacted by the mowing activities at the test site. When separating the CRNS data set into dry and wet periods, soil properties and topography are the main drivers of SMC estimation. Sentinel 1 or Sentinel 2 data add the existing temporal variability to the regression models. The analysis underlines the need of combining optical and SAR observations (Sentinel 1, Sentinel 2) as well as soil property and topographical information to understand and follow the CRNS-based SMC signal for different hydrological conditions and land cover types.