Conference Agenda

With ongoing global warming, hydro-climatic extremes such as heatwaves and droughts are becoming more frequent and intense. In particular, compound hot-dry extremes have increased in recent decades and pose a growing risk to agricultural production and water resources by intensifying the mismatch between water supply and demand. The AgriWATER project focuses on detecting these compound events and quantifying their impacts across Europe and China using Earth observation (EO) data and modelling approaches. So far, progress has been made in three main aspects.

First, we developed a framework to detect compound hot-dry extremes by combining satellite observations, reanalysis, and in situ measurements. This allows a consistent analysis of their spatial and temporal patterns over the past two decades, as well as their main driving factors. Related work has also highlighted persistent uncertainties in land–atmosphere coupling derived from EO products and models, which directly affect the representation of compound extremes. Second, we quantified the impacts of these extremes on agriculture and water resources in major crop regions. By integrating EO-based variables such as soil moisture, evapotranspiration, and vegetation indicators with machine learning and process-based hydrological and crop models, we provide improved estimates of yield anomalies and water deficits under compound stress conditions. Recent results show that soil heat extremes intensify faster than atmospheric extremes in many regions, indicating strengthened land–atmosphere feedback under dry conditions. In addition, analyses of long-term crop data reveal an increasing role of climate factors, particularly precipitation deficits, in shaping yield variability, consistent with increasing exposure to compound hot-dry extremes. Third, we explore potential adaptation strategies to reduce the vulnerability of agricultural water systems. Scenario-based analyses are used to assess possible management options under increasing climate variability.

The collaboration between European and Chinese partners has been active and productive, including joint data analysis, model development, and regular exchanges. Young scientists are closely involved in the project and contribute directly to its core topics. Their work includes improving the understanding of land–atmosphere coupling uncertainties, analysing soil heat extremes using global in situ networks, quantifying shifts in climate-driven crop yield variability, and integrating satellite observations to constrain subsurface heat storage. These efforts have resulted in publications in leading journals such as Science Advances, completed Master’s theses, and joint PhD research stays at European partner institutes. In addition, delegation visits from Chinese partners to Helmholtz Centre for Environmental Research – UFZ have further strengthened collaboration.

Climate change and anthropogenic pressures are increasingly threatening global food security and quality, underscoring the need for advanced monitoring and analytical tools. The activities carried out to date in project ID# 95250 aim to evaluate the effectiveness of synergistically combining European and Chinese Earth Observation (EO) data for retrieving key parameters related to soil and crop conditions.

Regarding soil, our objective is to assess the efficacy of integrating visible and thermal sensor data to retrieve soil texture and properties such as Soil Organic Carbon (SOC), Calcium Carbonate (CaCO₃), and pH. The primary goal is to evaluate the potential of a combined visible-thermal approach in view of future ESA missions (e.g., CHIME and LST) as well as NASA missions (e.g., SBG and SBG-TIR).

While current satellite missions such as PRISMA and EnMAP have demonstrated their capabilities in soil monitoring, the thermal infrared (TIR) spectral range remains underexplored. Moreover, PRISMA and EnMAP acquire images only on demand, whereas HYTES, being an airborne mission, covers limited and localized areas. This makes it difficult to obtain simultaneous VSWIR and TIR images over the same area, hindering the development of a combined approach. However, with the upcoming ESA missions CHIME and LSTM, featuring larger swaths (130 km and 670 km, respectively) and higher revisit frequencies (11 and 2 days), the likelihood of acquiring concurrent thermal and visible images increases significantly.

This study focuses on the Jolanda di Savoia site, located in the agriculturally important Po Valley in northern Italy (centered at 44.87°N, 11.97°E). This highly fertile region is a major producer of high-quality rice. The soil composition exhibits considerable variability, mainly due to the area's historical landscape evolution, notably the presence of buried paleo-channels—remnants of former marshlands—which contribute to soil heterogeneity. A time series of PRISMA and EnMAP imagery from 2020 to 2025 has been used, along with HYTES airborne survey data acquired in 2023 during a joint ESA-ASI campaign.

Ongoing research has integrated in-situ soil wet lab measurements with VSWIR (0.4–2.5 µm) and LWIR (7.5–11.6 µm) data. Various preprocessing techniques and machine learning algorithms, including Gaussian Process Regression (GPR) and Partial Least Squares Regression (PLSR), have been employed to develop robust models for retrieving soil properties. The efficacy of these models is assessed based on their accuracy and ability to effectively leverage combined visible and thermal data.

Results indicate that SOC could be predicted using VSWIR time series with an RMSE of 2.79, R² of 0.54, and RPD of 1.47, while using the LWIR spectral range yielded an RMSE of 1.22, R² of 0.54, and RPD of 1.42. Future research will focus on the combined use of VSWIR and LWIR to emulate the future CHIME-LSTM tandem.

In China, the demand for accurate and up-to-date information on wheat Grain Protein Content (GPC) has become increasingly urgent, driven by rising food consumption market demands and intensifying international competition. To address this challenge and fill the data gap, the first long-term winter wheat GPC dataset (CNWheatGPC-500) at 500 m spatial resolution has been developed, covering major wheat-growing regions in China. It was created by integrating multi-source data from ERA5/MODIS and employing a Hierarchical Linear Model (HLM) that deciphers the spatial hierarchical relationships among temporal phenological, meteorological, and remote sensing data. CNWheatGPC-500 provides valuable insights for improving wheat production and quality control, and for supporting decision-making in the agricultural sector. The product is freely accessible.

Regarding crop threats, the core system for detecting yellow rust outbreaks in maize and wheat crops will be built using PRISMA satellite data. Several vegetation indices (NDVI, SIPI, PRI, PSRI, MSR) derived from hyperspectral images will be used to implement a Disease Infection Index. The impact of spatial resolution on the capability to detect yellow rust in crops will be one of the key outcomes of this activity.

For desert locust presence risk forecasting, we have proposed a dynamic prediction method for the Somalia–Ethiopia–Kenya region. Monthly prediction experiments have been conducted, extracting high, medium, and low risk areas for desert locust occurrence in the study area. Results demonstrated an overall accuracy of 77.46%, and the model enables daily dynamic forecasting of desert locust risk up to 16 days in advance, providing early warning and decision support for preventive ground control measures.

In addition, we have established remote sensing-based monitoring and risk assessment methods for biological disasters in forests and grasslands. For pine wilt disease (PWD) monitoring, we proposed a spatiotemporal change detection method to reduce false detections in tree-scale PWD monitoring under complex landscape conditions. Results showed that the proposed method effectively distinguishes wilted pine trees from other easily confused objects. For grasshopper monitoring, we took two steppe types as research objects and established a remote sensing monitoring model for grasshopper potential habitat. The results demonstrated that the most suitable and moderately suitable areas are mainly distributed in the southern part of the meadow steppe and the eastern and southern parts of the typical steppe. We also found that soil temperature during the egg stage, vegetation type, soil type, and precipitation amount during the nymph stage are significant factors in both meadow and typical steppes.