Search USGS⌕ Search

Geology topics

Wenwen Li

Publications and source records attributed to Wenwen Li.

17 recordsLinked to original sources
USGS2026

Geospatial Agentic Services: A framework for interoperable geospatial intelligence

Autonomous GIS uses large language models as its reasoning core to automate spatial data collection, analysis, modeling, and visualization for solving spatial problems with minimal human intervention. Often implemented through autonomous geospatial agents that work independently or collaboratively, it represents a shift from GIS as a passive software environment requiring expert operation towards systems capable of task-oriented interaction, workflow generation and increasingly autonomous spatial analysis and decision support. However, current GIS agents are often isolated and tightly coupled with specific platforms, tools, or applications, which limits their reuse and interoperability across systems and domains. In this paper, we introduce Geospatial Agentic Services, or GAS, as a service-oriented framework for publishing, discovering, invoking, and composing GIS agents as interoperable geospatial services. GAS extends traditional geospatial interoperability beyond access to data and predefined operations towards geospatial intelligence interoperability, where GIS agents function as discoverable, reusable, and collaborative service entities that encapsulate spatial knowledge and reasoning. Building on service-oriented architecture, GAS supports structured geospatial skill declarations, agentic service registry and discovery, and distributed agent orchestration. It also integrates validation, provenance, reproducibility, and governance to support transparent and trustworthy geospatial analyses. To demonstrate the framework, we present a GAS implementation and illustrate its use through common invocation modes and representative AI-orchestrated workflows. By connecting traditional geospatial service architecture with emerging agentic interoperability, GAS provides a conceptual and architectural foundation for interoperable, reusable, and scalable agentic geospatial services and workflows in the era of autonomous GIS.

Annals of GIS↗
USGS2025

GIScience in the era of Artificial Intelligence: A research agenda towards Autonomous GIS

The advent of generative AI exemplified by large language models (LLMs) opens new ways to represent and compute geographic information and transcends the process of geographic knowledge production, driving geographic information systems (GIS) towards autonomous GIS. Leveraging LLMs as the decision core, autonomous GIS can independently generate and execute geoprocessing workflows to perform spatial analysis. In this vision paper, we further elaborate on the concept of autonomous GIS and present a conceptual framework that defines its five autonomous goals, five levels of autonomy, five core functions, and three operational scales. We demonstrate how autonomous GIS could perform geospatial data retrieval, spatial analysis, and map making with four proof-of-concept GIS agents. We conclude by identifying critical challenges and future research directions, including fine-tuning and self-growing decision-cores, autonomous modelling, and examining the societal and practical implications of autonomous GIS. By establishing the groundwork for a paradigm shift in GIScience, this paper envisions a future where GIS moves beyond traditional workflows to autonomously reason, derive, innovate, and advance geospatial solutions to pressing global challenges. Meanwhile, we emphasize that as we design and deploy increasingly intelligent geospatial systems, we carry a responsibility to ensure they are developed in socially responsible ways, serve the public good, and support the continued value of human geographic insight in an AI-augmented future.

Annals of GIS↗
USGS2025

A review of abrupt permafrost thaw: Definitions, usage, and a proposed conceptual framework

Purpose of Review We review how ‘abrupt thaw’ has been used in published studies, compare these definitions to abrupt processes in other Earth science disciplines, and provide a definitive framework for how abrupt thaw should be used in the context of permafrost science. Recent Findings We address several aspects of permafrost systems necessary for abrupt thaw to occur and propose a framework for classifying permafrost processes as abrupt thaw in the future. Based on a literature review and our collective expertise, we propose that abrupt thaw refers to thaw processes that lead to a substantial persistent environmental change within a few decades. Abrupt thaw typically occurs in ice-rich permafrost but may be initiated in ice-poor permafrost by external factors such as hydrologic change (i.e., increased streamflow, soil moisture fluctuations, altered groundwater recharge) or wildfire. Summary Permafrost thaw alters greenhouse gas emissions, soil and vegetation properties, and hydrologic flow, threatening infrastructure and the cultures and livelihoods of northern communities. The term ‘abrupt thaw’ has emerged in scientific discourse over the past two decades to differentiate processes that rapidly impact large depths of permafrost, such as thermokarst, from more gradual, top-down thaw processes that impact centimeters of near-surface permafrost over years to decades. However, there has been no formal definition for abrupt thaw and its use in the scientific literature has varied considerably. Our standardized definition of abrupt thaw offers a path forward to better understand drivers and patterns of abrupt thaw and its consequences for global greenhouse gas budgets, impacts to infrastructure and land-use, and Arctic policy- and decision-making.

Current Climate Change Reports↗
USGS2025

Harnessing geospatial artificial intelligence and deep learning for landslide inventory mapping: Advances, challenges, and emerging directions

Recent advancements in artificial intelligence (AI) and deep learning enable more accurate, scalable, and automated mapping. This paper provides a comprehensive review of the applications of AI, particularly deep learning, in landslide inventory mapping. In addition to examining commonly used data sources and model architectures, we explore innovative strategies such as feature enhancement and fusion, attention-boosted techniques, and advanced learning approaches, including active learning and transfer learning, to enhance model adaptability and predictability. We also highlight the remaining challenges and potential research directions, including the estimation of more diverse variables in landslide mapping, multimodal data alignment, modeling regional variability and replicability, as well as issues related to data misinterpretation and model explainability. This review aims to serve as a useful resource for researchers and practitioners, promoting the integration of deep learning into landslide research and disaster management.

Remote Sensing↗
USGS2024

GeoAI for science and the science of GeoAI

This paper reviews trends in GeoAI research and discusses cutting-edge ad- vances in GeoAI and its roles in accelerating environmental and social sciences. It ad- dresses ongoing attempts to improve the predictability of GeoAI models and recent re- search aimed at increasing model explainability and reproducibility to ensure trustworthy geospatial findings. The paper also provides reflections on the importance of defining the "science" of GeoAI in terms of its fundamental principles, theories, and methods to ensure scientific rigor, social responsibility, and lasting impacts.

Journal of Spatial Information Science↗
USGS2024

Segment anything model can not segment anything: Assessing AI foundation model's generalizability in permafrost mapping

This paper assesses trending AI foundation models, especially emerging computer vision foundation models and their performance in natural landscape feature segmentation. While the term foundation model has quickly garnered interest from the geospatial domain, its definition remains vague. Hence, this paper will first introduce AI foundation models and their defining characteristics. Built upon the tremendous success achieved by Large Language Models (LLMs) as the foundation models for language tasks, this paper discusses the challenges of building foundation models for geospatial artificial intelligence (GeoAI) vision tasks. To evaluate the performance of large AI vision models, especially Meta’s Segment Anything Model (SAM), we implemented different instance segmentation pipelines that minimize the changes to SAM to leverage its power as a foundation model. A series of prompt strategies were developed to test SAM’s performance regarding its theoretical upper bound of predictive accuracy, zero-shot performance, and domain adaptability through fine-tuning. The analysis used two permafrost feature datasets, ice-wedge polygons and retrogressive thaw slumps because (1) these landform features are more challenging to segment than man-made features due to their complicated formation mechanisms, diverse forms, and vague boundaries; (2) their presence and changes are important indicators for Arctic warming and climate change. The results show that although promising, SAM still has room for improvement to support AI-augmented terrain mapping. The spatial and domain generalizability of this finding is further validated using a more general dataset EuroCrops for agricultural field mapping. Finally, we discuss future research directions that strengthen SAM’s applicability in challenging geospatial domains.

Remote Sensing↗
USGS2023

Reimagining standardization and geospatial interoperability in today’s GeoAI culture

Integrating Geospatial Artificial Intelligence (GeoAI) into our technological landscape has revolutionized our capacity to understand and engage with the world. However, the burgeoning adoption of GeoAI applications has underscored the imperative of data, format, and conveyance standardization and enhancing geospatial interoperability. This vision paper delves into the intricacies of the evolving GeoAI environment, emphasizing the indispensable role of standardized practices and elevated interoperability. By synthesizing insights from geography, computer science, and data ethics, this contribution envisions a future characterized by seamless synergy between AI systems and geospatial data, driving impactful decision-making and transformative innovation.

Conference Paper↗
USGS2023

Assessment of a new GeoAI foundation model for floodinundation mapping

Vision foundation models are a new frontier in GeoAI research because of their potential to enable powerful image analysis by analyzing and extracting important image features from vast amounts of geospatial data. This paper evaluates the performance of the first-of-its-kind geospatial foundation model, IBM-NASA’s Prithvi, to support a crucial geospatial analysis task: flood inundation mapping. This model is compared with popular convolutional neural networks and vision transformer-based architectures regarding mapping accuracy for flooded areas. A benchmark dataset, Sen1Floods11, is used in the experiments, and the models' predictability, generalizability, and transferability are evaluated based on both validation datasets and datasets completely unseen by the model. Results show the impressive transferability of the Prithvi model, highlighting its performance advantages in segmenting flooded areas in previously unseen regions. The findings also suggest areas for improvement for the Prithvi model in adopting multi-scale representation learning, developing more end-to-end pipelines for high-level image analysis tasks, and offering more flexibility in allowable input data bands.

Conference Paper↗
USGS2023

GeoAI for spatial image processing

The development of digital image processing, as a subset of digital signal processing, depended upon the maturity of photography and image science, introduction of computers, discovery and advancement of digital recording devices, and the capture of digital images. In addition, government and industry applications in the Earth and medical sciences were paramount to the growth of the technology. From the early days when photography was first introduced to science to today, artificial intelligence and deep learning technologies have been intensively used to analyze imagery. Spatial image processing has experienced breakthroughs and evolutions. This chapter presents an overview of the history of image processing, GeoAI-based image processing applications, and the role of GeoAI in advancing image processing methods and research. We also discussed the remaining challenges to using GeoAI for image processing regarding training data annotation, the issues of scale, resolution, and change in space over time.

Book chapter↗
USGS2022

GeoImageNet: A multi-source natural feature benchmark dataset for GeoAI and supervised machine learning

The field of GeoAI or Geospatial Artificial Intelligence has undergone rapid development since 2017. It has been widely applied to address environmental and social science problems, from understanding climate change to tracking the spread of infectious disease. A foundational task in advancing GeoAI research is the creation of open, benchmark datasets to train and evaluate the performance of GeoAI models. While a number of datasets have been published, very few have centered on the natural terrain and its landforms. To bridge this gulf, this paper introduces a first-of-its-kind benchmark dataset, GeoImageNet, which supports natural feature detection in a supervised machine-learning paradigm. A distinctive feature of this dataset is the fusion of multi-source data, including both remote sensing imagery and DEM in depicting spatial objects of interest. This multi-source dataset allows a GeoAI model to extract rich spatio-contextual information to gain stronger confidence in high-precision object detection and recognition. The image dataset is tested with a multi-source GeoAI extension against two well-known object detection models, Faster-RCNN and RetinaNet. The results demonstrate the robustness of the dataset in aiding GeoAI models to achieve convergence and the superiority of multi-source data in yielding much higher prediction accuracy than the commonly used single data source.

GeoInformatica↗
USGS2022

GeoAI and the future of spatial analytics

This chapter discusses the challenges of traditional spatial analytical methods in their limited capacity to handle big and messy data, as well as mining unknown or latent patterns. It then introduces a new form of spatial analytics—geospatial artificial intelligence (GeoAI)—and describes the advantages of this new strategy in big data analytics and data-driven discovery. Finally, a convergent spatial analytical framework is suggested as a potential future pathway for spatial analysis.

Book chapter↗
USGS2021

The evolution of geospatial reasoning, analytics, and modeling

The field of geospatial analytics and modeling has a long history coinciding with the physical and cultural evolution of humans. This history is analyzed relative to the four scientific paradigms: (1) empirical analysis through description, (2) theoretical explorations using models and generalizations, (3) simulating complex phenomena and (4) data exploration. Correlations among developments in general science and those of the geospatial sciences are explored. Trends identify areas ripe for growth and improvement in the fourth and current paradigm that has been spawned by the big data explosion, such as exposing the ‘black box’ of GeoAI training and generating big geospatial training datasets. Future research should focus on integrating both theory- and data-driven knowledge discovery.

Book chapter↗
USGS2020

GeoNat v1.0: A dataset for natural feature mapping with artificial intelligence and supervised learning

Machine learning allows “the machine” to deduce the complex and sometimes unrecognized rules governing spatial systems, particularly topographic mapping, by exposing it to the end product. Often, the obstacle to this approach is the acquisition of many good and labeled training examples of the desired result. Such is the case with most types of natural features. To address such limitations, this research introduces GeoNat v1.0, a natural feature dataset, used to support artificial intelligence‐based mapping and automated detection of natural features under a supervised learning paradigm. The dataset was created by randomly selecting points from the U.S. Geological Survey’s Geographic Names Information System and includes approximately 200 examples each of 10 classes of natural features. Resulting data were tested in an object‐detection problem using a region‐based convolutional neural network. The object‐detection tests resulted in a 62% mean average precision as baseline results. Major challenges in developing training data in the geospatial domain, such as scale and geographical representativeness, are addressed in this article. We hope that the resulting dataset will be useful for a variety of applications and shed light on training data collection and labeling in the geospatial artificial intelligence domain.

Transactions in GIS↗
USGS2019

The effect of resolution on terrain feature extraction

Recent increase in the production of high-resolution digital elevation models (DEMs) from lidar data has led to interest in their use for terrain mapping. Although the impact of different resolutions has been studied relative to terrain characteristics like roughness, slope and curvature, its relationship to the extraction of terrain features remains unclear. To address this question, this study tests the impact of four resolutions on the capture of glacial cirques from DEMs. Mean curvature was derived from one arc-second, one-third arc-second, one-ninth arc-second and half meter DEMs representing a cirque-covered mountainous region southwest of Lake Tahoe, California. Using a GEOBIA workflow, ridge objects were identified, and three scales - via the multi-resolution scale parameter (SP) - of objects bordering the ridges were classified as cirque objects. The resulting classifications were compared to reference cirques digitized at a scale of ~1:10,000. Results show that the one-third arc-second DEM produces the set of cirque objects most closely resembling the reference cirques. The one-ninth arc-second DEM afforded the second-best classification. These results emphasize the importance in carefully choosing resolution relative to the features extracted, rather than using the highest resolution data available. In the case of GEOBIA workflows, the choice of scale parameter is equally important.

California↗
USGS2019

Deep convolutional neural networks for map-type classification

Maps are an important medium that enable people to comprehensively understand the configuration of cultural activities and natural elements over different times and places. Although a massive number of maps are available in the digital era, how to effectively and accurately locate and access the desired map on the Internet remains a challenge today. Previous works partially related to map-type classification mainly focused on map comparison and map matching at the local scale. The features derived from local map areas might be insufficient to characterize map content. To facilitate establishing an automatic approach for accessing the needed map, this paper reports our investigation into using deep learning techniques to recognize seven types of map, including topographic, terrain, physical, urban scene, the National Map, 3D, nighttime, orthophoto, and land cover classification. Experimental results show that the state-of-the-art deep convolutional neural networks can support automatic map-type classification. Additionally, the classification accuracy varies according to different map-types. This work can contribute to the implementation of deep learning techniques in the cartographic community and advance the progress of Geographical Artificial Intelligence (GeoAI).

Conference Paper↗
USGS2018

A spatio-contextual probabilistic model for extracting linear features in hilly terrain from high-resolution DEM data

This paper introduces our research in developing a probabilistic model to extract linear terrain features from high resolution DEM (Digital Elevation Model) data. The proposed model takes full advantage of spatio-contextual information to characterize terrain changes. It first derives a quantifiable measure of spatio-contextual patterns of linear terrain feature, such as ridgelines, valley lines and crater boundaries, and then adopts multiple neighborhood analysis and a probability model to address data uncertainty in terrain surface modeling. Different from traditional approaches, the proposed model has the ability to achieve near-automated processing, and to support effective extraction of terrain features in both smooth and rough surfaces. Through a series of experiments, we demonstrate that the proposed approach outperforms existing techniques, including: thresholding, stream/drainage network analysis, visual descriptor, object-based image analysis and edge detection. We hope this work contributes to both the geospatial data science and geomorphology communities with a new way of utilizing high-resolution imagery in terrain analysis.

International Journal of Geographical Information ↗