Climate change amplifies extreme weather events, posing critical risks to South African agriculture, where rainfed farming and climatic variability heighten vulnerability. Maize, the country’s staple crop, is sensitive to compounded hazards. However, current risk assessments largely consider hazards in isolation and often fail to capture their interactions across time, space, and crop development stages, limiting their usefulness for decision-making. Therefore, this paper presents a novel risk-scoring system within a multi-hazard assessment framework applied to South African maize production. This system integrates irrigation-defined clusters, climatic indices, and crop-stage-specific dynamics. Climatic hazards—including droughts, heatwaves, cold spells, heavy rainfall, diseases, and frost—were profiled, and risk scores calculated from their correlations with yield. Statistical models, linking spatio-temporal yield variability to multi-hazard risk scores, captured how hazard interactions compound across regions, irrigation levels, and phenological stages. Results show that yield losses are not driven by the number of hazards occurring each year, but by specific hazard interactions across phenological stages. Yield gaps between rainfed and moderately irrigated systems narrow from 41
Vision–language models (VLMs) have shown significant promise in remote sensing applications, particularly for land-use and land-cover (LULC) mapping via zero-shot classification and retrieval. However, current approaches face several key challenges, such as the dependence on caption-based supervision, which is often not available or very limited in terms of the covered semantics, and the fact of being adapted from generic VLM architectures that are suitable for very high resolution images. Consequently, these models tend to prioritize spatial context over spectral and temporal information, limiting their effectiveness for medium-resolution remote sensing imagery.In this work, we present TimeSenCLIP, a lightweight VLM for remote sensing time series, using a cross-view temporal contrastive framework to align multispectral Sentinel-2 time series with geo-tagged ground-level imagery, without requiring textual annotations. Unlike prior VLMs, TimeSenCLIP emphasizes temporal and spectral signals over spatial context, investigating whether single-pixel time series contain sufficient information for solving a variety of tasks.Our approach is trained on the LUCAS and Sen4Map datasets and evaluated across four main mapping tasks: land cover, land use, habitat mapping and crop type classification. The CLIP text encoder can be used to probe the learned representations using semantically meaningful categories, enabling effective zero-shot generalization without task-specific text supervision. We further extend our evaluation to bioregions mapping and country-level image retrieval. Although coarse, these tasks are valuable for probing whether the model captures geographically meaningful representations, such as regional climate regimes, vegetation patterns, and land-use structures. TimeSenCLIP achieves consistently better performance than existing CLIP-based remote sensing models in both zero-shot classification and cross-modal retrieval. Notably, single-pixel multispectral time series variants remain highly competitive, particularly with extended temporal coverage, demonstrating that temporal–spectral dynamics can compensate to a substantial degree for the reduced spatial footprint.While larger spatial patches still offer advantages for tasks where spatial patterns are inherently informative, such as ecosystem type classification, the results suggest that single-pixel multispectral time series can provide effective remote sensing vision–language pipelines, supporting scalable and efficient modeling in scenarios where large spatial tiles or extensive textual annotations are impractical. Code is available at https://github.com/pallavijain-pj/TimeSenCLIP.
Pre-trained vision-language models (VLMs), such as CLIP, demonstrate impressive zero-shot classification capabilities with free-form prompts and even show some generalization in specialized domains. However, their performance on satellite imagery is limited due to the underrepresentation of such data in their training sets, which predominantly consist of ground-level images. Existing prompting techniques for satellite imagery are often restricted to generic phrases like a satellite image of ..., limiting their effectiveness for zero-shot land-use and land-cover (LULC) mapping. To address these challenges, we introduce SenCLIP, which transfers CLIPs representation to Sentinel-2 imagery by leveraging a large dataset of Sentinel-2 images paired with geotagged ground-level photos from across Europe. We evaluate SenCLIP alongside other SOTA remote sensing VLMs on zero-shot LULC mapping tasks using the EuroSAT and BigEarthNet datasets with both aerial and ground-level prompting styles. Our approach, which aligns ground-level representations with satellite imagery, demonstrates significant improvements in classification accuracy across both prompt styles, opening new possibilities for applying free-form textual descriptions in zero-shot LULC mapping.
The expectations of digital technologies in sustainable agricultural development are considerable. However, applying these technologies in agri-food value chains can have downsides, which are still barely studied. The main objectives of this systematic literature review were to discover the state of the art of the research in the use of digital technologies in business models contributing to sustainability in the agri-food sector, and to make recommendations for future research and management practice. In order to bring concepts together and develop a theoretical framework and advance knowledge, performing a literature review is conducive. Here, the commonly-used PRISMA-method was used to develop a systematic literature review. From this review, an overview of business model innovations, and drivers, benefits and drawbacks of digitalisation in agri-food value chains were distinguished. Key themes found in the literature were the effects of COVID-19 on digitalisation and business resilience, the economic sustainability of business models, and the importance of communication technologies in agri-food value chains. This article recommends for future research and management practice to use a framework that looks through a value co-creation and open innovation perspective to the individual business model level and the interaction between (sustainable) business models in local and global food systems.
The French wine industry is spread across the country and represents 789,000 ha (2023). Over 20% of the plant protection products (PPPs) sold in France are used in viticulture on less than 4% of the French UAA (Utilized Agricultural Area). The share of wine estates with organic farming certification has risen sharply, reaching 9% of French vineyards in 2016. The position occupied by the wine sector on both the national and international scale confirms the need to examine the impacts of different management practices in viticulture on human health and the environment. This study presents an approach to the assessment of plant protection practices in vineyards based on indicators of plant protection pressure and risk. It was carried out on wine-growing farms in the southwest of France, surveyed according to the two farming systems: conventional/integrated and organic. The main objective of this study was to compare the health and environmental impact of the PPPs used in these two farming systems. The impact assessment result of wine-growing plant protection practices shows that some pesticides and molecules used in organic farming, especially those based on copper and sulfur, are more harmful than products used in conventional/integrated farming, in particular to the environment. For this reason, all stakeholders involved in pesticide management should recognize the health and environmental impact of PPPs in order to reduce and to control their toxicity risks to public health and the natural environment.
