Deep learning models have shown encouraging capabilities for mapping accurately forests at medium resolution with TanDEM-X interferometric SAR data. Such models, as most of current state-of-the-art deep learning techniques in remote sensing, are trained in a fully-supervised way, which requires a large amount of labeled data for training and validation. In this work, our aim is to exploit the high-resolution capabilities of the TanDEM-X mission to map forests at 6 m. The goal is to overcome the intrinsic limitations posed by midresolution products, which affect, e.g., the detection of narrow roads within vegetated areas and the precise delineation of forested regions contours. To cope with the lack of extended reliable reference datasets at such a high resolution, we investigate self-supervised learning techniques for extracting highly informative representations from the input features, followed by a supervised training step with a significantly smaller number of reliable labels. A 1 m resolution forest/non-forest reference map over Pennsylvania, USA, allows for comparing different training approaches for the development of an effective forest mapping framework with limited labeled samples. We select the best-performing approach over this test region and apply it in a real-case forest mapping scenario over the Amazon rainforest, where only very few labeled data at high resolution are available. In this challenging scenario, the proposed self-supervised framework significantly enhances the classification accuracy with respect to fully-supervised methods, trained using the same amount of labeled data, representing an extremely promising starting point for large-scale, very high-resolution forest mapping with TanDEM-X data.

Trust and interpretability are crucial for the use of Artificial Intelligence (AI) in scientific research, but current models often operate as black boxes offering limited transparency and justifications for their outputs. We introduce AiSciVision, a framework that specializes Large Multimodal Models (LMMs) into interactive research partners and classification models for image classification tasks in niche scientific domains. Our framework uses two key components: (1) Visual Retrieval-Augmented Generation (VisRAG) and (2) domain-specific tools utilized in an agentic workflow. To classify a target image, AiSciVision first retrieves the most similar positive and negative labeled images as context for the LMM. Then the LMM agent actively selects and applies tools to manipulate and inspect the target image over multiple rounds, refining its analysis before making a final prediction. These VisRAG and tooling components are designed to mirror the processes of domain experts, as humans often compare new data to similar examples and use specialized tools to manipulate and inspect images before arriving at a conclusion. Each inference produces both a prediction and a natural language transcript detailing the reasoning and tool usage that led to the prediction. We evaluate AiSciVision on three real-world scientific image classification datasets: detecting the presence of aquaculture ponds, diseased eelgrass, and solar panels. Across these datasets, our method outperforms fully supervised models in low and full-labeled data settings. AiSciVision is actively deployed in real-world use, specifically for aquaculture research, through a dedicated web application that displays and allows the expert users to converse with the transcripts. This work represents a crucial step toward AI systems that are both interpretable and effective, advancing their use in scientific research and scientific discovery.

The advancements of language language models (LLMs) have piqued growing interest in developing LLM-based language agents to automate scientific discovery end-to-end, which has sparked both excitement and skepticism about their true capabilities. In this work, we call for rigorous assessment of agents on individual tasks in a scientific workflow before making bold claims on end-to-end automation. To ensure the scientific authenticity and real-world relevance of our benchmark, we extract 102 tasks from 44 peer-reviewed publications in four disciplines and engage nine subject matter experts to validate them. We unify the target output for every task to a self-contained Python program file and employ an array of evaluation metrics to examine the generated programs, execution results, and costs. Each task goes through multiple rounds of manual validation by annotators and subject matter experts to ensure its annotation quality and scientific plausibility. We also propose two effective strategies to mitigate data contamination concerns. Using our benchmark, we evaluate five open-weight and proprietary LLMs, each with three frameworks: direct prompting, OpenHands CodeAct, and self-debug. Given three attempts for each task, the best-performing agent can only solve 32.4% of the tasks independently and 34.3% with expert-provided knowledge. In addition, we evaluate OpenAI o1 with direct prompting and self-debug, which demonstrates the effectiveness of increasing inference-time compute. Still, our results underscore the limitations of current language agents in generating code for data-driven discovery, let alone end-to-end automation for scientific research.

Citizen Science (CS) leverages the collective efforts of non-specialist/ordinary volunteers in different research tasks, such as collecting, analyzing, and classifying data to solve technical and scientific challenges. CS applications have attracted the attention of academic researchers due to the abundance of data created with high quality at low cost. According to an article in CERN Courier Magazine,3 CS is beneficial for the scientific community, the volunteers involved in the projects, and society as a whole. On the researcher's side, CS helps to achieve scientific data/metadata quickly, obtaining large amounts of valuable information that can contribute to advancing research.3 On the other hand, volunteers become aware of a scientific methodology, are recognized for their contributions, and feel satisfied for being part of a project with scientific and social relevance.2

Remotely sensed data are dominated by mixed Land Use and Land Cover (LULC) types. Spectral unmixing (SU) is a key technique that disentangles mixed pixels into constituent LULC types and their abundance fractions. While existing studies on Deep Learning (DL) for SU typically focus on single time-step hyperspectral (HS) or multispectral (MS) data, our work pioneers SU using MODIS MS time series, addressing missing data with end-to-end DL models. Our approach enhances a Long-Short Term Memory (LSTM)-based model by incorporating geographic, topographic (geo-topographic), and climatic ancillary information. Notably, our method eliminates the need for explicit endmember extraction, instead learning the input-output relationship between mixed spectra and LULC abundances through supervised learning. Experimental results demonstrate that integrating spectral-temporal input data with geo-topographic and climatic information significantly improves the estimation of LULC abundances in mixed pixels. To facilitate this study, we curated a novel labeled dataset for Andalusia (Spain) with monthly MODIS multispectral time series at 460m resolution for 2013. Named Andalusia MultiSpectral MultiTemporal Unmixing (Andalusia-MSMTU), this dataset provides pixel-level annotations of LULC abundances along with ancillary information. The dataset (https://zenodo.org/records/7752348) and code (https://github.com/jrodriguezortega/MSMTU) are available to the public.

In the initial stage of human life, communication, seen as a process of social interaction, was always the best way to reach consensus between the parties. Understanding and credibility in this process are essential for the mutual agreement to be validated. But, how to do it so that this communication reaches the great mass? This is the main challenge when what is sought is the dissemination of information and its approval. In this context, this study presents the ALT software, developed from original readability metrics adapted to the Portuguese language, available on the web, to reduce communication difficulties. The development of the software was motivated by the theory of communicative action of Habermas, which uses a multidisciplinary style to measure the credibility of the discourse in the communication channels used to build and maintain a safe and healthy relationship with the public.

Fishermen gaffing catfish at the Madeira River rapids in the state of Rondônia, Brazil. These rapids are now completely submerged, and a nearby …

Microsoft: A new Artificial Intelligence (AI) platform was launched this Wednesday (4) to facilitate actions to prevent and combat deforestation in the Amazon rainforest. The PrevisIA tool, developed by Microsoft, the Amazon Institute of Man and Environment (Imazon) and the Vale Fund, anticipates information on regions susceptible to deforestation and fires. The algorithm analyzes data on topography, land cover and legal and illegal roads in satellite images, to find risks of felling trees or fire and inform public agencies to carry out prevention and combat actions. Alerts generated by the platform are also open to the public on an initiative's dashboard. Microsoft Azure cloud capabilities and Imazon's AI algorithm to detect roads helped improve the deforestation risk model to identify territories most threatened by deforestation in the Amazon, such as Indigenous Lands and Conservation Units.

Nearly 10,000 square kilometers of the Brazilian Amazon, an area the size of Lebanon, is at high risk of being cleared, according to a new tool using artificial intelligence technology to help forecast deforestation before it actually happens. Named PrevisIA (from the Portuguese previsão for "forecast" and IA for "artificial intelligence"), the tool analyzes images provided by European Space Agency satellites, and through an algorithm created by the Brazilian conservation nonprofit Imazon, finds areas prone to deforestation. Imazon studies published in scientific journals show that 95% of accumulated deforestation in the Amazon is located within a 5.5-kilometer (3.4-mile) radius of roads; 90% of annual fires occur 4 km (2.5 mi) from illegal roads built in the middle of the forest for logging, mining and land grabbing. In 2006, the non-profit started to monitor satellite images manually to find these roads before the area around them was cleared of trees, but the laborious and time-consuming work prevented it from scaling up -- a problem that the new technology aims to solve. The tool mapped so far the Brazilian Amazon, but could potentially be expanded to any forested area on Earth, the developers say.

Chinese artist Ai Weiwei announced a new documentary at Art Basel Miami Beach, where he has several pieces on display, the Art Newspaper reports. With one film already in the works on animals and the environment, Ai sent a camera team to the Brazilian states of Rondônia, Mato Grosso, and Amazonas to capture footage of the ongoing fires in the Amazon Rainforest, along with another team which went to Pará to shoot cattle farms. This footage will be used for a separate documentary on the fires, as well as in next year's production of Turandot at the Teatro dell'Opera di Roma, which Ai is directing. Agribusiness and the deforestation of the Amazon are inextricably linked issues. Ai said in his announcement: "We can clearly see that the fires are a part of a wide-ranging and premeditated plan to cause deforestation to increase land use for agriculture and cattle farming."