Conference Agenda

The FAO has developed the WAPOR database, a publicly accessible, near real time database containing evaporation, transpiration, interception and net primary production maps at different scales, using data gathered by a range of satellites including, among others, Sentinel-2 and meteorological data from ERA5. This database is the backbone of the WaPOR project that, now in its second phase, works with ten partner countries to build their capacity in the use of WaPOR data for its different applications, and to generate solutions to local challenges linked to water and land productivity as well as water management.

WaPOR Accounter bridges the gap between this database and end-users, by enabling them to quickly and easily convert its data into actionable information at two different scales.

At field-scale, users can select agricultural fields and instantly see WaPOR data linked to the selected fields. The app helps users to identify crop seasons for the selected fields and aggregates data accordingly, allowing comparisons of water consumption, biomass production and water productivity between neighbouring farms and/or between seasons. Additionally, users are able to compare a fields water consumption to its optimal consumption, helping them identify water-stressed or waterlogged fields.

At the basin-scale, the app lets users see a basins water-balance and related statistics and indicators, such as the basins exploitable water, utilized flows and water consumption per landuse class.

ARIES is exploring the potential of spaceborne hyperspectral data (PRISMA and EnMap) to address water management and food security in Africa. Within the project prototype EO products are being created and important recommendations for the design of future Copernicus missions (CHIME) are being made. The product design process takes place in close collaboration with several African Partners from Southern, Western and Eastern Africa in order to ensure usefulness and applicability as well as knowledge and capacity transfer.

Getting accurate and timely information on plant water content has been identified as one of the main challenges across the continent, especially for farmers. Plant Water, Leaf Area Index and Canopy Water have thus been chosen as key information products to be produced within the project. We are deriving these, using the agricultural applications available within the free and open source QGIS plugin EnMap Toolbox, which is also implemented on the FS-TEP platform, thus allowing access to the chosen approaches for anyone at any time even after the project lifetime.

The parameters are retrieved from EnMap and PRISMA data collected from March 2023 onwards over six different test sites located in Senegal, Mali, Niger and Zambia, using a dual approach that relies on radiative transfer modelling (PROSAIL) for leaf area retrieval and on canopy water retrieval based on a water absorption feature between 930 and 1060 nm of the electromagnetic spectrum (Wocher et al. 2018). Combining the two products, plant water content can be calculated.

During the workshop we would like to present our approach and results from our test site in Zambia, where knowledge about plant water is essential for decision making with regards to irrigation and harvest.

The project runs from September 2022 until September 2024 and is funded by ESA (ESA Contract No: 4000139191/22/I-DT).

Accurate early-season crop type classification is crucial for the crop production estimation and monitoring of agricultural parcels. However, the complexity of the plant growth patterns and their spatio-temporal variability present significant challenges.
While current deep learning-based methods show promise in crop type classification from single- and multi-modal time series, most existing methods rely on a single modality, such as satellite optical remote sensing data or crop rotation patterns. We propose a novel approach to fuse multimodal information into a model for improved accuracy and robustness across multiple crop seasons and countries.
The approach relies on three modalities used: remote sensing time series from Sentinel-2 and Landsat 8 observations, parcel crop rotation and local crop distribution.
To evaluate our approach, we release a new annotated dataset of 7.4 million agricultural parcels in France (FR) and the Netherlands (NL). We associate each parcel with time-series of surface reflectance (Red and NIR) and biophysical variables (LAI, FAPAR). Additionally, we propose a new approach to automatically aggregate crop types into a hierarchical class structure for meaningful model evaluation and a novel data-augmentation technique for early-season classification.
Performance of the multimodal approach was assessed at different aggregation levels in the semantic domain, yielding to various ranges of the number of classes spanning from 151 to 8 crop types or groups. It resulted in accuracy ranging from 91% to 95% for the NL dataset and from 85% to 89% for the FR dataset.
Pre-training on a dataset improves transferability between countries, allowing for cross- domain and label prediction, and robustness of the performances in a few-shot setting from FR to NL, i.e., when the domain changes as per with significantly new labels.
Our proposed approach outperforms comparable methods by enabling deep learning methods to use the often overlooked spatio-temporal context of parcels, resulting in increased precision and generalization capacity.

Timely crop type mapping is crucial to inform actionable interventions for food security, serving as a key input for estimating crop area and deriving information relevant for yield forecasting. While traditional approaches rely on in-season Ground-Truth (GT) data collection, recent methods utilize deep-learning models solely based on data from previous years. This can contribute to reducing costs related to data collection. To perform well, such models require a substantial amount of data for training. In Europe, a source of data is the annual GeoSpatial Application (GSA) datasets, i.e., the farmers declarations supporting the aid application under the Common Agricultural Policy, including the agricultural parcels boundaries with cultivated crop types. When made public, these datasets become valuable sources of extensive GT data for mapping crop types but lack harmonization across national datasets. Leveraging this GT data source, we have established the CHEAP (Common Harmonized European Agricultural Parcels) database. It consists of a multi-annual parcel dataset with harmonized crop types across years and countries, spanning from 2008 to 2022, covering 13 countries (more than 22M parcels). Here, we directly integrate the CHEAP dataset with multi-annual Sentinel-2 time series to evaluate the capacity for in-season crop type prediction at parcel level. Using a stratified random sampling, we obtained raw and smoothed multi-annual time series (TS) of Sentinel-2 (S2) spectral data for 5 million parcels per year. This Satellite Image Time Series, which stands out for its distinctive spatial and temporal coverage, was employed to assess the capability of a recently published multi-modal deep-learning model (Barriere, Claverie et al. 2024, RSE) to predict crop type at parcel level during the early stages of the season for ten countries and four seasons (2019-2022). This unique and vast evaluation highlights the potential of the algorithms to perform over a large variety of cropping systems.

Timely information on nutrient concentrations (micro-nutrients, macro-nutrients, protein) in staple crops over large areas is lacking which limits our understanding of how nutrients vary across various geographic areas. In the absence of this information, we cannot efficiently guide research activities dedicated to alleviating potential nutrient deficiencies through genetic biofortification or agronomic biofortification by applying fertilizers. Conventional methods for measuring the grain nutrient levels typically consist of collecting grains at harvest and performing wet chemical analysis, near-infrared spectroscopy, or hyperspectral imaging of the crop grains in the laboratory. These methods are time-consuming and cost-prohibitive and, consequently, unsuitable for consistent quantification of nutrients across large spatial extents. In addition, as the nutrients are only measured after harvest, this approach precludes effective intervention with fertilizers while the crop is still growing.

To overcome the scale and cost limitations of laboratory analysis, the EO4Nutri team is investigating the potential of various Earth Observation data including hyperspectral, multispectral, and thermal data to estimate wheat grain protein content, as well as Calcium (Ca), Iron (Fe), Magnesium (Mg), Nitrogen (N), Phosphorus (P), Potassium (K), Selenium (Se), Sulphur (S), and Zinc (Zn) in wheat, maize and rice grains and soil. The team collected soil, plant, and grain samples together with field spectroscopy data and satellite EnMAP and PRISMA satellite images in Jolanda di Savoia and Munich-North-Isar (MNI) test sites and assessed the lifecycle of nutrients of wheat, maize, and rice from the soil to crop canopy to crop grains with state-of-the-art ML analytical techniques. Preliminary results show the potential of hyperspectral data to predict the nutrients and protein in the final agriculture production.

The Russian forces invaded Ukraine on 24th February 2022 leading to widespread disruption of Ukraine's agricultural system. Ukraine is a major exporter of crops , the invasion therefore poses a significant risk to global food security. Total production is one of the critical indicators in this regard which in turn is directly proportional to the total harvested area. The majority of remote sensing based harvest detection studies require a complete satellite phenological time series and use the assumption that senescence leads to harvest. Both these conditions cannot be applied in this case as all planted fields need not be harvested.

Given these constraints and challenges, we developed a method to monitor crop harvest near-real time using high resolution Planet satellite imagery. Our method includes training a model to cluster change patterns on historic window based spatio-temporally sampled data and then identify harvest patterns in the current season. The sampling approach ensures we capture a complete representation of change patterns that exist. Clusters are assigned as ‘harvested’ or ‘non-harvested’ by visually inspecting imagery at a higher temporal resolution, using which harvest can be seen as a change event. Our method works in the absence of training labels.

In free Ukraine we found 94% of planted winter crops to be harvested and in occupied Ukraine it was 88% as of 19th September 2022. Strong visual patterns of non-harvested crops were observed along the occupation borders.We visually interpreted satellite imagery at a higher temporal frequency to generate statistically significant validation data for model accuracy calculation. We obtained an overall accuracy of 85% with an f1-score of 90% for the harvested class and 73% for the non-harvested class. Our assessments and analysis were directed to different organizations and agencies dealing with the Ukraine crisis and led to several key insights and derived interpretations.

.

83% of greenhouse gas emissions from cereal and oilseed crops originate from fertiliser when both fertiliser-induced field emissions and emissions from the manufacture of synthetic fertiliser are included [HGCA/AHDB 2012]. Nitrogen Use Efficiency (NUE) is the nitrogen in the crop at harvest as a percentage of the nitrogen supply including both Soil nitrogen supply and applied fertiliser. UK farmers achieve, on average, around 60% nitrogen fertiliser use efficiency, but research suggests 80% efficiency could be attainable. Excess applied nitrogen is lost through leaching into water courses, and as Nitrous oxide emissions, which impacts the environment and farmers’ costs.

NUE-Profits (funded by Innovate-UK/DEFRA) is a farmer-led collaboration with industry specialists and academic experts to combine measurement, monitoring, modelling, and forecasting of crop and soil to enable farmers to improve NUE without compromising yield or grain protein thus improving farmer profit margins and reducing crop environmental impact.

Earth observation data compliments in-field measurements, reducing the number and frequency of measurements a farmer needs to make to monitor crop growth and plant nitrogen uptake and to observe spatial and temporal variation. Leaf Area Index (LAI) can accurately characterise the plant phenology, particularly the timing and rate of crop growth, which provides information about plant nitrogen uptake. We generate gap-filled LAI time-series from Sentinel 2 observations with cloud and cloud shadow removed using a cloud mask we optimised for monitoring cereal crops. We also assimilate LAI, alongside in-situ measurements and other datasets (for example soil moisture), into a physiological crop growth model to localise the model to the field and sub-field level. This allows us to model and forecast crop yield and development based on historical and forecast weather data. The data assimilation framework is adaptable to incorporating additional data sets when they are available and, importantly, allows quantification of uncertainty.

Climate change, environmental degradation, and the rising demand for sustainable food systems urgently demand a transformation of our agricultural practices and supply chains. From ensuring food safety and reducing waste to promoting fair trade, the complexity of our food systems underscores the need for greater transparency and traceability in how we source our food. Earth Observation (EO) has proven a powerful tool for understanding land use, deforestation, and agriculture's environmental impacts. However, emerging technologies like blockchain have the potential to disrupt the sector further, promising unprecedented visibility and accountability. This could empower consumers, farmers, and regulators alike to make informed choices that support sustainable farming practices and facilitate compliance with regulations like the EU Deforestation Regulation (EUDR).

The RUDEO project, a 1-year ESA project carried out by GMV and GMV-SES, proposes use cases to carry out a critical examination of the transformative potential of blockchain within agriculture. The project plans to evaluate existing solutions and propose new workflows that will assess the actual implications of using blockchain technology to streamline certification processes, ensure fair compensation for farmers who adopt sustainable methods, support carbon markets linked to cropland carbon stocks, create transparent platforms connecting producers and consumers, and trace the journey of commodities for ethical sourcing.

The RUDEO project's investigations will shed light on whether blockchain, alongside established EO technologies, can truly revolutionize agricultural practices and even go beyond aligning with the goals of the European Green Deal's Farm to Fork Strategy. Success in this area could offer solutions to the urgent challenges facing agriculture under pressure, facilitating more robust Monitoring, Reporting, and Verification (MRV) systems. In turn, this would support initiatives like the CAP Eco-Schemes by streamlining certification processes and incentivizing innovative agricultural practices with low environmental impact. Additionally, this integrated approach has the potential to reduce GHG emissions, combat biodiversity loss, and promote deforestation-free products. If achievable, the combination of EO data with blockchain could offer significant added value by driving collaboration, fostering trust, and accelerating the adoption of sustainable practices that benefit both farmers and the planet.

The Common Agricultural Policy implementers are on their roadmap to switch to a full Area Monitoring System (AMS). AMS is not just to integrate IT and EO technological novelties into the Integrated Administration and Control System (IACS), it is a real paradigm shift towards a performance-based CAP. The full-scale and time-wise continuous cloud processing has opened a wide range of possibilities to analyse land management and land cover systems with comprehensive territorial and seasonal coverage. The administration must be capable of addressing via the direct payments the challenges of climate mitigation, soil protection, biodiversity and different agro-ecosystem services. Beyond the parcel-level validation duties of CAP Paying Agencies for planning and monitoring the payment schemes, AMS requires further multi-annual EO data analysing capabilities. Extended data sharing and data interoperability are also key conditions of new AI-based modelling techniques.

The proposed Agri-Dashboard implements Data as a Service and visualisation by widget-based apps. It enables Paying Agencies to get fast access to reliable information on agricultural parcel level required for customised monitoring, expert judgements and policy reporting obligations.

We highlight solutions for understanding large scale outputs of Machine Learning products, detections taken along the threshold-driven marker logic, analysing class-based decision models or integrating Geotagged Photos. Agri-Dashboard supports data slicing and data analysis using misclassification matrices, class-level accuracy measures and statistics or analysing the quality of training and test data. These data insight methods are fundamental for AMS Quality Assurance.

The multi-annual EO-data of AMS seriously has the potential to become a solid knowledge-base of a real performance-oriented payment system opening higher flexibility and spatial sensitivity to design eligibility conditions, impact indicators in zonal logic, meaning that AMS products could form the base of strategic planning.

Agricultural landscape features (LF) are small fragments of non-productive (not directly used for agricultural production) and typically, but not only, natural or semi-natural vegetation (hedges, tree, grass margin, stone walls, ponds, ditches etc). They are located in the agricultural landscape and can provide ecosystem services and support for biodiversity. In the latest LUCAS - a harmonized land cover and land use data collection exercise that extends over the whole EU territory - Landscape Features module done in 2022, the survey is comprised of 3.8 million sampling points, distributed over 93,000 one-hectare quadrats and the methodology combines office-based photo interpretation with field surveys . In a first phase, all points for the field survey were photo-interpreted in the office. Surveyors classified points as non-LF or one of four types of LF (woody, grassy, wet, and stony) based on areal EO ortho-photo of ~ 20 cm resolution. This was followed by field observation to confirm/correct the previous step. Based on the outcomes of the LUCAS LF module, we assess the influence of the density and type of landscape features on the biodiversity of pollinators and farmland birds. We also consider information on agricultural intensity and compare the results with previous landscape surveys within LUCAS. We assess how this information can be used to evaluate the performance of the Common Agricultural Policy in halting and reversing the loss of farmland biodiversity for a more sustainable agriculture.

Feeding a growing world population sustainably is crucial, but sustainable practices have varied, ever-changing impacts. This is because agriculture is a complex system, influenced by various local factors such as soil composition, soil health, land use, and climate variability. It is important to understand the heterogeneity of effects of practices at the field-level to help farmers personalise the decisions.

Analysing complex interactions and estimating effects boils down to answering causal queries. In recent years, observation-based causal inference approaches have emerged as a mature field. This research introduces a causal inference framework utilising Double Machine Learning to estimate Conditional Average Treatment Effects. The primary objective is to uncover the nuanced and spatially diverse impacts of crop rotation (farming practices of interest) on Soil Organic Carbon (SOC) levels in Lithuania's heterogeneous regions.

The study uses the farmers' declarations of the crops they cultivate (LPIS) from 2018-2022 to derive a binary crop rotation treatment, determined by cultivating at least 4 different crops in 5 years. We control for climate variables from ERA5 (e.g. surface net short-wave radiation, soil temperature), and soil properties (e.g. clay) to help us identify and estimate the impact of crop rotation on SOC content.

Beyond estimating local effects, the study explores the diversity of outcomes across space and time. This deeper analysis helps identify the key drivers influencing these variations, such as soil composition, climate, or even specific crop rotation combinations. The results indicate that crop rotation appears to be more effective in the western croplands of the country while exhibiting a negative impact on SOC in the central-northern regions.

Estimating the effects of practices at the field level allows farmers to make informed choices tailored to their specific conditions. This local-specific approach contrasts with traditional one-size-fits-all recommendations, empowering farmers to optimise their practices for both environmental and economic sustainability.

Monitoring Land Use Land Cover Change (LULCC) due to oil palm plantations in near-real time is instrumental for regulating palm oil imports to EU in the context of EU Deforestation-free Regulation (EUDR).

Earth Observation satellite missions providing free global imagery with high revisit frequencies, are critical for monitoring tropical rainforests and their conversion to oil palm plantations. SAR imagery (Descals et al. 2021) and LandTrendr algorithms (Du et al. 2022) have been used in previous studies to identify closed-canopy oil palm plantations and planting years. However, these approaches have limitations in accurately mapping young and sparse plantations.

This study aims to fill these gaps by detecting plantations in a timely manner. Thanks to the Proba-V full mission Collection 2 spanning 2014 to 2020, we developed a sensor-agnostic method for monitoring deforestation and plantation rotation. RED, NIR and SWIR wavelength allows computing the NDWVI, a new index for discriminating vegetation from land cleared for plantations. Normalisation reduces atmospheric and seasonal variability, while statistical boundaries identify pixels that deviate from the normal distribution of forest values. The algorithm detects LULCC with a bi-weekly timestep at global scale.

The annual maps depict the dynamics typical for oil palm plantations, from land preparation to mature plantations, including dates of clearing and planting. Validation involves 1016 points randomly selected using a stratified sampling. Preliminary results in Southeast Asia show F1-score and OA of 81% for plantations detection. 70% of detections were accurate for the exact year, surpassing significantly previous studies.

This method can support the EUDR implementation for imports and provides valuable insights into yield estimation. This work demonstrates the suitability of 100m spatial resolution and 5-day temporal frequency for global mapping, particularly in cloudy regions and for perennial crops. Release of this oil palm plantation map series is scheduled for May 2024, with publication.

.