Since large language models (LLMs) generally have a large parameter scale and acquire knowledge across multidisciplinary corpora, their parameters tend to exhibit varying levels of importance for GIS knowledge alignment. Specifically, certain model layers of LLMs may already align well with GIS knowledge, while others may require significant adaptation and thus play a more critical role during the fine-tuning process. However, representative fine-tuning methods, such as low-rank adaptation (LoRA), apply a uniform treatment to all parameters, overlooking their varying importance for GIS downstream tasks. This strategy may disrupt well-adapted layers or limit less-adapted ones from effectively encoding GIS knowledge. Therefore, we develop a novel GIS knowledge-aware LoRA allocation (GLoRA) scheme that adaptively allocates trainable parameters to layers based on their importance. Specifically, GLoRA first investigates a GIS knowledge-aware strategy to identify the importance of model layers for a given GIS task in a data-dependent way. Second, it dynamically adjusts the size of LoRA modules, allocating more parameters to more important layers to enhance their representation capability. This adaptive approach ensures efficient parameter utilization while preserving the strengths of well-adapted layers. We evaluated GLoRA on three GIS-related tasks, and the results show its improved performance compared to recent baselines with comparable parameter budgets.

Large Language Models (LLMs) excel in natural language-relevant tasks like text generation and question answering Q&A. To further expand their application, efforts focus on enabling LLMs to utilize real-world tools. However, their tool-use ability in professional GIS remains under explored due to two main challenges. Firstly, LLMs are usually trained on general-domain corpora, lacking sufficient and comprehensive GIS-specific data to align with professional knowledge, including understanding the functions of GIS tools. Secondly, researchers often need to combine multiple GIS tools to solve geospatial tasks. To address these challenges, we propose a trainable method to enable LLMs to master GIS tools. We curated a comprehensive set of resources: instruction-response data (GeoTool, 1950 instructions) to enhance the understanding of LLMs for GIS tools, instruction-solution data (GeoSolution, 3645 instructions) to improve their ability to generate tool-use solutions for geospatial tasks, and annotated instruction-solution evaluation data (GeoTask, 300 instructions) for evaluating LLMs’ GIS tool-use proficiency. Using the collected training data (GeoTool and GeoSolution), we fine-tuned a professional-domain LLM called GeoTool-GPT based on an open-source general-domain LLM, the LLaMA-2-7b model. The experiment based on evaluation data validates our method’s effectiveness in enhancing the tool-use ability of general-domain LLMs in the professional GIS domain, with the performance of our model closely approaching that of GPT-4.

Intelligent map analysis is an important yet challenging topic. Recently, the development of large models, especially Visual Language Models (VLMs), has shown potential for intelligent image analysis. However, these models are primarily trained on natural images, which have intrinsic differences from maps. Consequently, there remains a gap in applying existing general-domain VLMs to map analysis. To address this issue, we propose a framework for developing a specialized VLM, called MapReader. To achieve this goal, a comprehensive data resource is collected using a strategy that combines self-instruct with expert refinement, including training data (MapTrain: 2,000 pairs of maps and descriptions) and evaluation data (MapEval: 250 maps and 500 map-related questions). Based on the training data, MapReader is fine-tuned on top of a general-domain VLM to learn to understand and describe map contents. The evaluation results on MapEval suggest that: (1) MapReader can accept map inputs and generate detailed descriptions of core geographic information, and it also possesses visual question-answering capabilities, showing potential for application in various map analysis scenarios, such as accessible map reading and robotic map usage; (2) The proposed data collection strategy is effective, and the collected dataset can serve as a benchmark to promote further map analysis research.

Accurately predicting the next track point of vehicle travel is crucial for various Intelligent Transportation System (ITS) applications, such as travel behavior studies, traffic control, and traffic congestion monitoring. Recent works on trajectory prediction follow a paradigm that first represents the raw trajectory and subsequently makes predictions based on that representation. Currently, trajectory representation methods tend to project trajectory points to road networks by map matching and represent trajectories based on the representation of matched roads. However, precisely matching trajectories to roads is a challenge in ITS, as the matching precision is greatly affected by the quality of the trajectory. Meanwhile, since it is difficult to discern whether trajectory matching results are accurate or confounded, how to effectively utilize this type of uncertain geographic context information is also a challenge, which is defined as the Uncertain Geographic Context Problem (UGCoP) in geographic information science. Therefore, we propose a flexible strategy of subgraph learning, referred to as SLM, for predicting the next track point of vehicles. Specifically, a subgraph generation module is first proposed to extract topology contextual information of the roads around historical trajectory points. Secondly, a subgraph learning module is designed to learn rich spatial and temporal features from generated subgraphs. Finally, the extracted spatiotemporal features will be fed into a prediction module to predict the next track points of vehicles on road networks. Our model enables the effective utilization of uncertain geographic context information of trajectories on road networks while avoiding the error brought by map matching. Extensive experiments based on trajectory datasets in two different cities confirm the effectiveness of our approach.

Previous research has tended to use a global threshold of proximity to determine neighbors, neglecting spatial heterogeneity. Flexible thresholds implemented by adaptive search radii methods account for either the spatial structures or the non-spatial similarities of objects, but few consider both. By combining the spatial and non-spatial information of objects, we propose a novel approach that can automatically determine the neighbors that are strongly related to the object of interest. We introduce the sparse reconstruction technique from the signal processing domain, which aims to remove trivial relationships in a dataset. We extend the sparse reconstruction model by assuring three principles in spatial data, including retention of the correlation of data in the non-spatial attribute domain, preservation of local dependencies in the spatial domain, and removal of trivial relationships. Extensive experiments, based on road network missing value imputation and building clustering, show that our approach can make better use of both spatial and non-spatial information than a simple addition of them.

As a key focus of cartography and terrain analysis, the simplification of a digital elevation model (DEM) is used to preserve the pattern features of the terrain surface while suppressing its details over multiple scales. Statistical filtering and structural analysis methods are commonly used for this process. The structural analysis method performs well in identifying terrain structural edges, while it tends to discard the smooth morphology of a terrain surface. In addition, the filter that aims to reduce noise on a surface may over-smooth the terrain structural edges. Therefore, to preserve both the terrain structural edges and smooth morphology, we propose to combine the techniques of statistical filtering and structural analysis. Specifically, all the critical elevation points and structural edges are first detected from the DEM surface by using the structural analysis method. Then, the iterative guided normal filter is used to smooth the generalized DEM with the guidance of the structure of the original surface. After this process, the terrain structure is retained in the smooth surface of the DEM. The experimental results with a real-world dataset show that our method can inherit the merits of both structural analysis and statistical filter in preserving terrain features for multi-scale DEM representations.

As one of the major concerns in cartographic generalization, road network generalization aims at maintaining the patterns of road networks across map scales. Previous methods define the pattern of road networks mainly from the perspectives of geometry and topology. However, for navigation purposes, traffic flow information is also important to generalize road networks. More specifically, road segments that have a proximity relationship in the traffic flow system should be retained together on small-scale maps to preserve the completeness of the driving route. In this regard, this study proposes an improved method for road network generalization that considers network geometry, topology, and traffic flow patterns. First, strokes are constructed from the road network data based on the ‘every best fit’ geometric principle. Then, the relationships among strokes are developed on the basis of traffic flow patterns, which are extracted from taxi trajectory data. The strokes are then selected in sequence based on the indicators of geometry, topology, and traffic flow. Our experimental results demonstrate that the proposed method can preserve both the ‘Good Continuity’ principle and the transport function relationship of roads after generalization.