Exploring Sentinel‐2‐Based Spectral Variability for Enhancing Grassland Diversity Assessments Across Germany

Farming systems of high natural and cultural value represent approximately 30% of farmlands in the European Union and are associated with a high species and habitat diversity and/or the presence of species of European conservation concern. This study aims to synthesize the existing knowledge on the assessment of biodiversity and ecosystem services, the social‐ecological drivers of change and the innovations developed around high nature value (HNV) farming systems. We performed a systematic review of scholarly publications in English reporting empirical studies of HNV farmland in Europe. Information was extracted from 178 articles. We employed a multiple correspondence analysis and a hierarchical cluster analysis to uncover patterns in the assessment of ecosystem services and biodiversity in the HNV farming literature. A qualitative analysis of open text on challenges provided insight into drivers and their interactions. Similarly, open texts on main lessons learnt and recommendations allowed us to explore innovations. Most of the studies (n = 122, 69%) assessed the dimensions related to biodiversity in which grassland and bird diversity dominated. Six clusters were disentangled. The predominant biophysical assessment of plant diversity was generally grouped into clusters 1 and 2, while less common approaches were found in the remaining clusters, which tackled social dimensions (cluster 3), considered a wider set of biodiversity groups (cluster 5), or ecosystem services beyond provisioning services, either from a sociocultural perspective (cluster 4) or from a biophysical perspective (cluster 6). Direct drivers of change in land use in HNV farming systems are mostly related to abandonment and intensification processes. Among the indirect drivers of change, policy and institutional drivers were addressed in one‐third of the studies, with the Common Agricultural Policy being the most common driver in this category. Innovations were mostly related to technological innovations, while a limited number of articles were related to social, institutional and marketing innovations. Advisory services and knowledge transfer can have transformative potential to avoid the exclusion of marginalised farmers from being subsidised. Synergies can be promoted in policy mixes to jointly address landscape conservation and product innovation by supporting niche markets.

Local, adaptive traditional grassland management systems have played a fundamental role in the creation, maintenance and conservation of high nature value (HNV) grasslands. The state of diverse HNV grasslands has deteriorated across Europe in conjunction with changes in various management factors, such as management type and management intensity. To conserve the species-rich vegetation of HNV grasslands and to avoid undesirable shifts in plant functional type dominance, it is important to explore the effects of management factors crucial for nature conservation and to adapt them to local circumstances. In our study, we focus on three of the main factors in the management of valuable meadow steppes in the Great Hungarian Plain region (Central Hungary). We studied management types (mowing, grazing and combined), different levels of herbage removal intensity (low, medium, high) and spatio-temporal complexity (low, medium and high) of grassland management. Altogether 172 plots (1 m × 1 m) were designated in 17 sites. Plant diversity indexes and plant functional types were calculated according to the presence and percentage cover of plant species in the plots. Regarding plant diversity and the dominance of plant functional types, herbage removal intensity and spatio-temporal complexity of management had, for the most part, stronger effects than the type of management. Higher spatio-temporal complexity of management resulted in higher plant diversity, while higher intensity of management led to significantly lower diversity. Proper application of type, intensity and spatio-temporal complexity of management practices (separately and in combination) proved to be determining factors in the long-term maintenance and conservation of diversity and species composition of HNV grasslands.

Vertically resolved meteorological adjustments of aerosols and trace gases in Beijing, Taiyuan, and Hefei by using RF model

How High Nature Value (HNV) farmland is related to bird diversity in agro-ecosystems – Towards a versatile tool for biodiversity monitoring and conservation planning

Reservoir fluid PVT properties are measured in the laboratory for various use in reservoir engineering evaluation and estimation. Despite the indispensability of these PVT parameters, PVT lab data are seldomly available and if available may be unreliable. Instead, various empirical models have been developed and used in the industry. These empirical models are inherently inaccurate when used to predict PVT properties of fluid from different geological region with different depositional environment and fingerprint. Artificial Intelligence (AI) has evolved over the years and provided some algorithms with potentials to develop accurate predictive model for the prediction of bubblepoint pressure. This work tested some AI algorithms, compared performances and choose random forest regression algorithm in developing a robust predictive model for the estimation of bubblepoint pressure. Two thousand five hundred and twenty-two datasets obtained from oil reservoirs in different geographical locations were used for the feature scaling of input data, training and testing of the models. The independent variables, gas-oil ratio, temperature, oil density and gas density were confirmed to have key influence on the dependent variable Bubblepoint pressure The random forest model developed uses ensemble learning approach, combines predictions from multiple machine learning algorithms by averaging all predictions to make a more accurate prediction. The ‘forest’ generated by the random forest algorithm was trained through bootstrap aggregating. This is an ensemble meta-algorithm that improves the accuracy of machine learning algorithms. Percentage data split was 70% training and 30% testing. The reliability, accuracy and completeness of the predictive model capability were computed through performance indices such as the root mean square error (RMSE) and mean absolute error (MAE). The best network architecture was determined along with the corresponding test set RMSE, and Correlation coefficient. Statistical and graphical error analysis of the results showed that the random forest model performed better than existing models with 0.98 correlation coefficients for bubblepoint pressure. Better accuracy of reservoir properties prediction could be achieved using this random forest reservoir fluid properties prediction model.

Random forest (RF) modeling has emerged as an important statistical learning method in ecology due to its exceptional predictive performance. However, for large and complex ecological data sets, there is limited guidance on variable selection methods for RF modeling. Typically, either a preselected set of predictor variables are used or stepwise procedures are employed which iteratively remove variables according to their importance measures. This paper investigates the application of variable selection methods to RF models for predicting probable biological stream condition. Our motivating data set consists of the good/poor condition of n = 1365 stream survey sites from the 2008/2009 National Rivers and Stream Assessment, and a large set (p = 212) of landscape features from the StreamCat data set as potential predictors. We compare two types of RF models: a full variable set model with all 212 predictors and a reduced variable set model selected using a backward elimination approach. We assess model accuracy using RF's internal out-of-bag estimate, and a cross-validation procedure with validation folds external to the variable selection process. We also assess the stability of the spatial predictions generated by the RF models to changes in the number of predictors and argue that model selection needs to consider both accuracy and stability. The results suggest that RF modeling is robust to the inclusion of many variables of moderate to low importance. We found no substantial improvement in cross-validated accuracy as a result of variable reduction. Moreover, the backward elimination procedure tended to select too few variables and exhibited numerous issues such as upwardly biased out-of-bag accuracy estimates and instabilities in the spatial predictions. We use simulations to further support and generalize results from the analysis of real data. A main purpose of this work is to elucidate issues of model selection bias and instability to ecologists interested in using RF to develop predictive models with large environmental data sets.

Agriculture constitutes a dominant land cover worldwide, and rural landscapes under extensive farming practices acknowledged due to high biodiversity levels. The High Nature Value farmland (HNVf) concept has been highlighted in the EU environmental and rural policies due to their inherent potential to help characterize and direct financial support to European landscapes where high nature and/or conservation value is dependent on the continuation of specific low-intensity farming systems.Assessing the extent of HNV farmland by necessity relies on the availability of both ecological and farming systems' data, and difficulties associated with making such assessments have been widely described across Europe. A spatially explicit framework of data collection, building out from local administrative units, has recently been suggested as a means of addressing such difficulties.This manuscript tests the relevance of the proposed approach, describes the spatially explicit framework in a case study area in northern Portugal, and discusses the potential of the approach to help better inform the implementation of conservation and rural development policies.Synthesis and applications: The potential of a novel approach (combining land use/cover, farming and environmental data) to provide more accurate and efficient mapping and monitoring of HNV farmlands is tested at the local level in northern Portugal. The approach is considered to constitute a step forward toward a more precise targeting of landscapes for agri-environment schemes, as it allowed a more accurate discrimination of areas within the case study landscape that have a higher value for nature conservation.

High nature value (HNV) grasslands in mountain areas are important ecosystems for biodiversity maintenance and offer a multitude of ecosystem services, but they are constantly threatened by abandonment or intensive fertilization. The aim of this study was to assess the effects of organic and mineral fertilization, under mulching and abandonment scenarios, on the floristic composition and diversity of Nardus stricta-dominated grasslands located in the North-Eastern Carpathians (Romania). The field experiment included 11 variants (control, low, moderate, and high inputs), analyzed as communities with cluster, ordinations, indicator species, and α indices. The results showed a clear separation of communities along the input gradient, from the oligotrophic grassland dominated by Nardus stricta (control variant) to mesotrophic/eutrophic communities dominated by Dactylis glomerata, Festuca pratensis, and Trifolium pratense at moderate and high inputs. Moderate fertilization (10–20 t ha−1 manure; N50P50K50–N100P100K100) maximized species richness (37–38 species), Shannon diversity (H′ = 2.5–2.6), and evenness (E = 0.70–0.75). High inputs reduced diversity and favored competitive grasses. Indicator species analysis highlighted a multitude of species that show the plant communities’ response to adaptive management. Moderate fertilization provides a viable trade-off between productivity and biodiversity, while abandonment or overfertilization accelerates biodiversity loss.

Management of semi-natural grasslands is essential to retain the characteristic diversity of flora and fauna found in these habitats. To maintain, restore or recreate favourable conditions for grassland species, knowledge regarding how they occur in relation to grazing intensity and soil nutrient availability is crucial. We focused on grassland plant species, i.e., species selected to indicate high natural values in semi-natural grasslands. Environmental monitoring data collected at 366 grassland sites in southern Sweden between 2006 and 2010 were used to relate the occurrence of indicator species to factors describing geographic location, local site conditions related to nutrients and moisture, and management. Site productivity, soil moisture and cover of trees and shrubs were the main structuring factors, while other factors related to management had a lesser effect (grass sward height, amount of litter, type of grazer). Not surprisingly, these patterns were also reflected in species-wise analyses of the 25 most commonly occurring indicator species, with almost all species negatively related to site productivity and most also to soil moisture. Furthermore, many species were negatively affected by increasing sward height and litter. In contrast, species-wise responses varied among species in relation to increasing cover of trees and shrubs. In comparison to cattle grazing, sheep grazing was detrimental to six species and beneficial to none, while horse grazing was detrimental to no species and beneficial to four species. When evaluating species traits, taller plant species were favoured when site productivity, grass sward height and the amount of grass litter were high. There were no strong patterns related to the flowering time, leaf arrangement, or nutrient and light requirements of species. These results highlight the importance of nutrient-poor and dry sites, e.g., when selecting sites for conservation, and the importance of the type of management executed.

This report presents the European Grassland Butterfly Indicator, based on national Butterfly Monitoring Schemes (BMS) in 19 countries across Europe, most of them in the European Union. The indicator shows that since 1990 till 2011 butterfly populations have declined by almost 50 %, indicating a dramatic loss of grassland biodiversity. This also means the situation has not improved since the first version of the indicator published in 2005. Of the 17 species, 8 have declined in Europe, 2 have remained stable and 1 increased. For six species the trend is uncertain. The main driver behind the decline of grassland butterflies is the change in rural land use: agricultural intensification where the land is relatively flat and easy to cultivate, and abandonment in mountains and wet areas, mainly in eastern and southern Europe. Agricultural intensification leads to uniform, almost sterile grasslands for biodiversity. Grassland butterflies thus mainly survive in traditionally farmed low‑input systems (High Nature Value (HNV) Farmland) as well as nature reserves, and on marginal land such as road verges and amenity areas. (Less)

Monitoring of the environmental effects of a harbour extension and the compensation measures is a very complex task. The Voordelta area has high natural values, but is also of high economic importance. To implement a monitoring strategy for this area a multidisciplinary consortium has been formed, consisting of a number of institutes and companies. A central data management facility was set up for data storage and management. This chapter illustrates the data management approach using the Voordelta monitoring programme for the years 2004 to 2013. A central data management facility was set up for data storage and management. A repository gives access to raw data files to all team members. From the analysis of the raw data a number of information products have been developed and disseminated to the authorities and the public through Google Earth. It will be shown, that the presence of a strong multidisciplinary team and good collaboration is the key to success in this complex programme. The way the data have been managed supports this process enormously.

The world’s peatlands are our largest terrestrial carbon store whilst also providing a sustainable source of drinking water, a haven for wildlife and storing a record of our past. The England Peat Map aims to provide baseline maps for the extent, depth, and condition of peaty soils in England by 2024. This will enable targeting of future restoration, support nature recovery, improve greenhouse emissions reporting and natural capital accounting.The maps will be created using a combination of multi-scale Earth observation imagery (satellite and airborne), existing and new ecological field survey data and machine/deep learning. Extent and depth mapping is implemented with random forest models and uses Sentinel satellite imagery and airborne LiDAR in combination with other ancillary datasets (e.g., geology and climate) for prediction. Assessment of peatland condition requires looking at these landscapes in different ways. Land cover mapping is used as a proxy for condition by targeting reflective classes for condition (e.g., Sphagnum, heather, and bare peat). Random forest and convolutional neural network (CNN) models are used in combination with Sentinel satellite imagery, aerial photography, and airborne LiDAR to produce national outputs. Mapping erosion/drainage features (grips, gullies and haggs) across the landscape is essential in understanding the underlying hydrological condition of the peatland and promising results have been achieved using CNNs with LiDAR and aerial photography. The final aspect of assessed condition is the movement of peat, also termed bog breathing, and is measured using Sentinel-1 Interferometric Synthetic Aperture Radar (InSAR). This opportunity is a result of novel in-situ peat movement cameras being installed across pilot sites to provide ground truth data.The final maps will be released free of charge under an open UK government license, allowing wider application and new opportunities for use compared with currently available datasets. For example, these baseline maps have the potential to contribute towards national peatland monitoring to address further decline of peatland habitats and target restoration interventions to achieve cost effective results. Several challenges have occurred during the initial phase of the project such as the difficulty in licensing suitable training data and in defining what we are mapping when features lack a globally agreed definition (e.g., surface features). The talk will discuss these challenges as well as the future direction of the project and how these challenges can be overcome.

Deep learning (DL) has huge potential to provide valuable insights into biodiversity changes in species-rich agricultural ecosystems such as semi-natural grasslands, helping to prioritize and plan conservation efforts. However, DL has been underexplored in grassland conservation efforts, hindered by data scarcity, intricate ecosystem interactions, and limited economic incentives. Here, we developed a DL-based object-detection model to identify indicator species, a group of vascular plant species that serve as surrogates for biodiversity assessment in high nature value (HNV) grasslands. We selected indicator species Armeria maritima, Campanula patula, Cirsium oleraceum, and Daucus carota. To overcome the hurdle of limited data, we grew indicator plants under controlled greenhouse conditions, generating a sufficient dataset for DL model training. The model was initially trained on this greenhouse dataset. Then, smaller datasets derived from an experimental grassland plot and natural grasslands were added to the training to facilitate the transition from greenhouse to field conditions. Our optimized model achieved remarkable average precision (AP) on test datasets, with 98.6 AP50 on greenhouse data, 98.2 AP50 on experimental grassland data, and 96.5 AP50 on semi-natural grassland data. Our findings highlight the innovative application of greenhouse-grown specimens for the in-situ identification of plants, bolstering biodiversity monitoring in grassland ecosystems. Furthermore, the study illuminates the promising role of DL techniques in conservation programs, particularly as a monitoring tool to support result-based agri-environment schemes.

Agroforestry, the association between trees/shrubs and crops, a widespread practice in West Africa, is presented as a lever for ecological intensification to optimize cereal yields in the face of strong population growth and the fight against climate change. Within the framework of the EU-DESIRA SustainSAHEL project, we aim to develop techniques to spatially assess the effect of trees on millet yields on an intra-field scale using imagery from an UAV equipped with a multispectral camera combined with geostatistical approaches. Indeed, recent advances in earth observation technologies position the UAV as an effective tool for evaluating the agronomic performance of agroforestry systems and for taking into account the intra-field variability of yields caused by environmental conditions, agricultural practices or the presence of trees (Roupsard and al., 2020 ; Leroux and al., 2022). The objective of this study was to estimate millet yields intra-field variability using UAV and up-to-date geostatistical approaches.The study was carried out over the 2018-2022 cropping seasons in one representative Faidherbia parkland of the groundnut basin of Senegal. To that end, a Random Forest (RF) algorithm was first calibrated to estimate millet yield at sub-plot scale using a thresholding classification to eliminate non-vegetation elements and also to integrate texture data, in order to take into account the spatial relationships between pairs of pixels. Millet yields data and vegetation and textural index from aerial images at a flight height of 25 meters acquired in farmers’ plots were used to calibrate the RF model. The RF model was used to upscale yield at the whole field scale thus allowing to obtain a map of millet yield. Then Voronoï diagram, with Faidherbia as a reference, was applied to each yield map, considering each Voronoï region as a zone of influence of its included Faidherbia. We then applied a transformation and rotation matrix to overlay all the zones of influence of a population of 50 Faidherbia by putting all the trees at the same geographical position. Finally, we build an atlas, which is an average structure representative of a population and which makes possible to detect the patterns and properties of the evolution of the population considered, to evaluate the distance and directional effect of Faidherbia on vegetation index of the population and then on millet yield.The RF model is able to explain between 70 and 90 % of the millet yield variability. Then the analysis has shown that the tree has an influence on the millet stand density with a distance-decay effect from the tree. This stand density is about 60 % around the tree and 30 % at 15m from the tree.Key words : Agroforestry, Uav, Machine learning, Image analysis, Geostatistics, Atlas

In this study, the decision learning methods of regression tree and random forest analysis are investigated as complements to standard statistical methods such as analysis of variance and grouped regression. For this purpose, three diverse data sets were used. The first set is large and multidimensional and describes nitrous oxide emissions from sites across different geo-positions in the UK receiving various fertilisation treatments. The second set is based on Gliricidia tree provenances and has a small number of samples and an imbalanced distribution of factor classes. Random forest modelling was found to be a very viable option in the case of the first data set but failed in the case of second. The third data set, based on count observations recording osprey egg incubation times, lends itself to tree and forest modelling. These decision learning methods therefore appear well suited to handling the diverse, multi-dimensional and complex data sets that often arise in carrying out agricultural and ecological field experiments.