Accurately identifying building function is essential for urban management, urban renewal, and promoting sustainable city development. Previous studies on building function classification have primarily focused on extracting external physical characteristics from remote sensing imagery and socio-economic attributes from Points of Interest (POI) data. However, these studies often overlook the patterns of human mobility within buildings, making it challenging to identify building functions accurately. To address these issues, this study mines the latent features embedded in POI and human trajectory data, and constructs a deep learning model, which integrates POI category semantics and human mobility patterns to identify urban building functions. We evaluated the proposed model in the 23 districts of Tokyo, Japan. The results indicate that the proposed model is able to extract features obtained from diverse data sources to identify building functions, achieving a test accuracy of 90.27 % and a Kappa coefficient of 0.8858. The building function mapping results in Tokyo demonstrate that the proposed model can accurately classify building functions in megacities. This study finds that human mobility patterns within buildings significantly improve the identifying accuracy of residential and commercial buildings. The building function mapping results of this study can provide effective data support for urban planning in Tokyo.

Global climate change has led to more frequent and severe droughts in the middle and lower reaches of the Yangtze River, intensifying the spatiotemporal variability of crop yields in this region. Winter rapeseed, a major oilseed crop in China, is particularly vulnerable to these drought conditions, which now pose greater risks to local food security. Accurate and timely regional yield predictions are increasingly important for effective agricultural management and disaster response. However, predicting rapeseed yield at the city level is challenging due to complex climate patterns and the strengthened impact of drought. Addressing these challenges requires the integration of multi-source data, including both remote sensing and weather data, to capture the full range of environmental influences on crop growth. Traditional statistical and machine learning methods have often proven inadequate for robust, transferable yield prediction across different regions and years.This study presents a deep learning–based yield prediction framework that integrates multi-temporal remote sensing indicators and meteorological variables to estimate winter rapeseed yield under both normal and drought conditions. Using data from 2014 to 2023 for the middle and lower reaches of the Yangtze River, an Attention–Long Short-Term Memory (Attention-LSTM) model was developed by jointly incorporating time-series remote sensing indices, meteorological factors, and statistical yield records. Key phenological periods for yield estimation were identified through multi-temporal and multi-variable combinations, and input configurations were systematically optimized. The proposed framework outperformed LSTM, Random Forest, and Support Vector Regression models, achieving an R2 of 0.81 and RMSE of 306.73 kg/ha on the validation dataset. Spatiotemporal yield dynamics and regional applicability were further analyzed, and the model’s robustness and adaptability were assessed under drought conditions. Under drought scenarios, the model maintained high accuracy, with an R2 of 0.76 and RMSE of 358.32 kg/ha. These results indicate the framework’s potential for drought-resilient yield prediction and its value for agricultural management and drought assessment under future climate change.

Ground filtering (GF) from Airborne Laser Scanning (ALS) data is a fundamental and critical task in forestry applications, which has remained an extensively researched but not yet fully resolved issue, due to the abrupt topographic changes and intricate forest configurations. This study presents a topography-attentional GF network (TAGF-Net) consisting of three modules. First, the topography aware representation introduces topography distribution to drive abstraction of higher-order landform representations, resulting in discriminative topography awareness. Second, the captured topography-aware features are further processed using the elevation attentive aggregation to stress the significance of elevation information in GF. Finally, the global feature homogeneity is introduced to analyze the long-range interactions and contextual information in the network. The performance of the TAGF-Net is evaluated on a GF benchmark, which was collected in California, USA and encompassed varied topographic settings and forest ecosystems. Compared with representative traditional and deep-learning methods, the TAGF-Net attains cutting-edge performance with total error of 1.16%. Additionally, TAGF-Net trained on the above California dataset can also achieve a better result on the NEWFOR dataset and OpenGF dataset than the representative methods without retraining, which verified its generalization ability and robustness

Effective modeling of spatio-temporal contexts to support geographic reasoning is essential for advancing Geospatial Artificial Intelligence. Inspired by masked language models, this paper introduces the Masked Geographical Information Model (MGIM), a novel self-supervised framework for learning context-aware representations from multi-source spatio-temporal data. The framework's core innovations include a parcel-scale method for multi-source data fusion and a custom self-supervised masking strategy for diverse geographic elements. This integrated modeling approach enables the model to capture complex spatio-temporal relationships and achieve consistently strong performance across diverse geographic reasoning tasks, such as trajectory inference, people flow inference, event identification, and land parcel function analysis. MGIM accurately reasons from spatio-temporal contexts and dynamically adjusts inferences according to contextual changes. The visualization of attention mechanisms further illustrates MGIM's capacity to construct contextually-aware representations and task-specific attention patterns analogous to natural language processing models. This study presents a new paradigm for general-purpose spatio-temporal modeling in real-world geographic scenarios, offering significant theoretical and practical value, and promising an effective solution for building a geographic foundation model.

As rapid urbanization and climate warming intensifies, disparities in people’s heat exposure are emerging as a critical climate-adaptation and human health concern. Local Climate Zones (LCZs) bridge urban morphology and climate, yet current data-driven mapping is limited in cross-modal integration and poor spatial-structural consistency at street scale, reducing mechanistic insight and cross-climate comparability for heat analysis. This study proposes a unified geographic semantic-driven multimodal fusion framework (GS-Fusion) that embeds geographic priors into adaptive feature fusion and applies spatially aware hierarchical aggregation to produce structure-consistent street-block LCZ maps. Extensive experiments on the constructed Chinese multi-source LCZ dataset (MSLCZ-12) covering 12 representative cities demonstrated that the proposed framework yielded optimal outcomes, achieving an overall accuracy exceeding 96%, with a cross-regional transfer accuracy of 69%. Benefiting from the proposed framework’s capabilities, a comprehensive LCZ map for 46 cities was generated, including 34 Chinese provincial capitals, with low-cost. By integrating LCZ maps, Landsat-derived LST and WorldPop data, we assessed heat exposure levels and their heterogeneity for vulnerable populations (children and the elderly). The results demonstrated that: (1) Heat exposure exhibits a significant dependence on LCZs, with high exposure consistently concentrated in compact built-up, especially LCZ 1-2; (2) Pearson correlation analyses indicate predominantly negative correlations between LCZs and heat exposure with a clear heat exposure gradient across LCZ types within cities; (3) Inequity in heat exposure is systematically higher for the elderly than for children at both city and LCZ levels as indicated by Gini coefficients; (4) LCZ-level inequality is not uniform across urban forms, peaking in specific LCZ categories, particularly LCZ 2, 8, and 10, revealing pronounced intra-class heterogeneity. These findings provide a replicable pathway for precise and localized urban heat exposure governance among vulnerable populations, enhancing the scientific rigor and practical effectiveness of heat mitigation strategies.

Terrorist attacks significantly threaten a nation’s stability, prosperity, and social cohesion. Therefore, predicting terrorist attacks and identifying their underlying drivers are crucial for formulating effective counterterrorism strategies. Existing studies often prioritize either temporal or spatial dimensions, while their interplay and specific socioeconomic drivers are less explored. In this study, global news data are leveraged to construct a novel global conflict index (GCI), which integrates multisource datasets to comprehensively characterize the key drivers of terrorist attacks. TerrorXG is proposed to predict terrorist attacks, and SHAP analysis is applied to quantitatively interpret the importance and contributions of the driving factors. TerrorXG demonstrated superior performance (RMSE: 0.319; PCC: 0.777) and high computational efficiency. Compared with the second most influential factor (population size), the proposed GCI has a 42.4% greater impact on terrorist attacks. The interpretability analysis of the model highlights socioeconomic inequality as a primary determinant: the impacts of child malnutrition and infant mortality are 38.4% to 108.5% greater than the effect of urbanization. The influence of ethnicity represents only 9.7% of the impact of the GCI, providing empirical evidence that challenges traditional theoretical perspectives on ethnic conflict in terrorism research. This study provides valuable insights for optimizing the allocation of counterterrorism resources.

Land use/cover change (LUCC) modeling is essential for understanding human-environment interactions and informing sustainable land governance. Existing LUCC approaches are predominantly based on the integration of machine learning and cellular automata (CA). However, they often fall short in capturing complex temporal dynamics within high-dimensional geospatial data. With the emergence of large language models (LLMs), new opportunities arise for flexible and generalized modeling of spatial-temporal patterns. To explore the applicability of LLMs in LUCC simulation, this study proposes a novel LLM-CA framework for LUCC simulation, which integrates fine-tuned LLMs with CA-inspired prompts. It encodes historical land use data and driving factors into structured prompts to guide simulation. The framework was validated in the central urban area of Wuhan, China. Among the evaluated models, LLaMA-3.2-3B achieved the highest simulation accuracy, with an overall accuracy (OA) of 85.80% and a figure of merit (FoM) of 26.20% Experiments on spatial neighborhood sizes, temporal intervals, and multi-scenario LUCC simulation further demonstrated the framework's robustness and adaptability. This study pioneers the integration of LLMs into LUCC modeling, offering a novel and scalable pathway for future land system simulations.

Fine-grained land-cover mapping is crucial for accurately assessing environmental degradation and monitoring socioeconomic dynamics. Few-shot learning of hyperspectral images offers a promising solution in cases where sample collection is limited. However, previous studies, such as tree species mapping, typically use 1% or 0.5% of samples per class, yielding thousands of samples for common species but struggling to identify unseen or rare species (only one sample/shot) in real-world scenarios. Furthermore, inevitable cross-sensor, cross-category, and cross-scene variations significantly increase the occurrence of unseen or rare classes and spectral heterogeneity within common land-cover types. To this being, we propose Knowing-Net, a knowledge-data-model-driven multimodal few-shot learning network, to bridge the application gap for fine-grained mapping of unseen or rare classes. In Knowing-Net, prior knowledge of sensor, i.e., spectral parameters, is leveraged to reconstruct cross-sensor hyperspectral images, mitigating heterogeneity in spectral responses across datasets and enabling cross-domain transfer across different sensors, scenes, and land cover types. To breakthrough the gap in recognizing unseen classes, multimodal data, including textual descriptions and natural images of unseen classes, is embedded into network to construct shared side information through modality-specific feature learning. By designing a cross-alignment mechanism for hyperspectral and multimodal information in a shared semantic space, distinct encoders are guided to produce consistent distribution for the same class across different modalities, reducing sample dependency and facilitating the identification of unseen or rare classes. Finally, inspired by the first law of geography, a sliding discriminant window is designed to incorporate spatial context, enhancing geography interpretability and robustness to noise. We evaluate Knowing-Net on five challenging airborne hyperspectral datasets with a fine-grained classification system, covering crop type, tree species, and similar urban land covers with varying materials. Extensive experiments on five datasets consistently demonstrate Knowing-net’s superiority over state-of-the-art methods in both mapping performance and cross-domain generalization. Notably, the unified framework achieves state-of-the-art results in one-shot learning and establishes a new paradigm in zero-shot classification for fine-grained land cover tasks. To the best of our knowledge, this is the first comprehensive generalization of FSL across sensor, category, and scene for hyperspectral image-based fine mapping.

The optimal placement of sensors is crucial for the maximization of resource utilization in environmental monitoring. The location optimization methods for maximal coverage are not suitable for applications that aim to predict the values of environmental variables at locations where no sensor is placed. For such a task, spatial sampling strategies have been developed to enhance the prediction accuracy by optimal placement of a limited number of sensors. These sensor placement strategies, however, primarily focus on the spatial variation of an environmental variable when selecting sampling locations, neglecting the variable's spatiotemporal heterogeneity, leading to inadequate spatiotemporal representativeness and limited prediction accuracies at different times. This study proposes a novel adaptive spatiotemporal sampling (ASTS) framework for sensor placement by combining the spatiotemporal variation features of the environmental variable with an adaptive sampling principle. The framework employs the integrated nested Laplace approximation for stochastic partial differential equations (INLA-SPDE) to capture the spatiotemporal variation and uses the adaptive sampling design principle to optimally place a predetermined number of sensors to maximize the prediction accuracy over space and time. The efficacy of the ASTS was validated using a PM2.5 data set from Wuhan, China. The ASTS framework reduced prediction root mean square error by 26.11 percent and computational time by 47.83 percent compared to methods focusing only on spatial variation and genetic algorithm combined with INLA-SPDE, respectively. The ASTS framework can be used to establish a new sensor placement plan or enhance existing networks by optimally adding sensors, ensuring high prediction accuracy across geographic locations and time periods.

ABSTRACT Urban economic forecasting is crucial for the formulation of a macroeconomic development strategy for a region toward harmonious socio‐economic and ecological progress. Cellular Automata (CA) enables fine‐grained urban economic modeling, but its conventional discrete‐state structure performs poorly when handling continuous economic variables like GDP density, and furthermore often lacks the ability to integrate with general economic growth patterns. These shortcomings ultimately compromise the simulation accuracy of urban economic dynamics. This paper innovatively proposes a continuous‐state Density Cellular Automata (DCA) framework integrated with S‐shaped economic growth responses, aiming to enhance GDP density simulation accuracy. Taking the Wuhan metropolitan area as a case, we simulated the 2010–2020 GDP density evolution and compared the performance with two baseline models. The results indicate that all cities in the metropolitan area follow S‐shaped growth but with significant inter‐city differences. The DCA demonstrates a 58.21% reduction in MAE compared to the discrete‐state PLUS model and a 29.78% decrease in RMSE relative to the conventional continuous‐state Gray‐Cells CA model. The scale effect analysis shows that smaller zoning scales improve the accuracy of economic simulations. DCA can provide references of economic trends and support urban planners and managers in balancing urbanization with the ecological environment.

The increasing frequency of extreme heat events poses serious challenges to public health and urban sustain-ability. Urban expansion is a key driver of extreme heat, yet the distinct mechanisms behind daytime and nighttime heat remain underexplored. This study proposes a multi-scale analytical framework to examine how 2D and 3D urban landscape changes influence extreme heat intensity (EHI), using both macro-scale (Spatial Difference-in-Differences) and finer-scale (Causal Forest) approaches. Two key findings emerge: 1) at the macro scale, urbanization significantly intensifies EHI, demonstrating its detrimental impact on thermal environments; 2) at the finer scale, heterogeneity analysis reveals that the landscape changes of building, impervious surface, cropland, and water bodies affect EHI in varied and localized ways. The results indicate the need for differentiated daytime and nighttime heat mitigation strategies, including enhancing blue-green infrastructure, optimizing urban landscape, and preserving cropland-water spatial balance to improve urban thermal resilience.

Accurate and fine-scale land use/land cover (LULC) mapping across rapidly evolving urban-rural landscapes under a unified framework remains a frontier challenge for Earth observation and spatial planning. Existing global and national products still suffer from three major bottlenecks: (1) inability to delineate fine-scale functional units across both urban and rural areas using road networks alone, leading to coarse and fragmented spatial boundary representation; (2) scarcity and imbalance of training samples across urban-rural contexts, restricting integrated city-level mapping and hindering nationwide scalability; and (3) substantial disparities in LULC categories and land functions between urban and rural land blocks, which hinder unified classification across heterogeneous landscapes. Overcoming these issues is crucial for advancing urban-rural sustainability studies and providing actionable information for planning. Here we propose RURBAN-Map, a knowledge-guided urban-rural mapping framework that transforms LULC mapping from data-driven to geographic cognition-based recognition. The framework integrates a coherent knowledge system including spatial structure, temporal semantics, and zoning strategy, which operationalizes block morphology, planning standards, and urban-rural data characteristics to guide block construction, feature extraction, and unified urban-rural LULC mapping. Using this framework, we constructed the first nationwide fine-scale block and sample dataset in China (UR-Blocks), consisting of 7,107,806 blocks and 3,624,452 labeled samples across 17 LULC categories. Automated delineation achieved a block boundary accuracy of 72.9%, while large-scale mapping across 20 representative cities and three major urban agglomerations yielded an overall accuracy of 78.8%. Compared with existing products that are restricted to urban built-up areas, RURBAN-Map reduces functional mixing and improves boundary fidelity by over 10%, while increasing mapping coverage by 80%, thereby achieving unified mapping across the urban-rural areas. Mapping results further show that rural regions contain meaningful functional land-use types, which have become increasingly relevant to rural policies. Beyond methodological advancement, RURBAN-Map establishes a benchmark dataset and a scalable knowledge-guided paradigm for fine-grained LULC mapping, offering both scientific value for land system modeling and practical support for urban-rural planning and policy implementation (https://zenodo.org/records/17205996).

The human-land system reflects complex links among population growth, urban expansion, and resource-environment dynamics, where rising food demand and urbanization directly occupy cropland and ecological land while indirectly affecting distant ecosystems through cropland compensation. The cascading effect triggered by urban expansion not only directly encroaches on surrounding cropland and ecological land but also indirectly impacts ecological areas in remote regions through the cropland compensation mechanism. Based on this understanding, this study proposes a multivariate coupled simulation framework comprising three modules: population, ecology, and food, which are independent yet interact with one another, and exhibit both flexibility and scalability to simulate dynamic feedback and hysteresis effects among these factors. Taking Hubei Province of China as an example, this framework explores how urban expansion triggers cascading impacts on regional ecological patterns, under the guiding principle of the cropland balance policy that seeks to align land occupation with equivalent compensation. Results show that the indirect ecological land take triggered by urban expansion is up to 87 times greater than direct land take, indicating far more extensive cascading effects. Although cropland protection policies have increased the total cropland area, both cropland and ecological land exhibit outward spatial displacement, implying that spatial mismatches and hidden ecological costs remain unresolved. This study enhances the understanding of complex human–land interactions and provides a scientific basis for optimizing regional resource allocation and promoting sustainable land use.

Non-motorized travelers, such as pedestrians and cyclists, are highly sensitive to infrastructures like stairs, tunnels, and overpasses that impede movement. Dynamically avoiding such barriers to find optimal paths remains a key challenge in route planning, as existing models often inadequately capture the interplay of road types, traveler behavior, and environmental dynamics. To address this, we propose UED-Q, a novel path planning method based on Q-learning, enhanced by Upper Confidence Bound (UCB)-guided exploitation and a dynamic learning rate tuning mechanism. UED-Q introduces a reward-shaping function that integrates road characteristics, user behavior, and real-time environmental changes. Evaluated on Wuhan’s real-world road network, UED-Q outperforms several advanced methods, including Q-learning with the Artificial Potential Field (QAPF), Combining Manhattan Distance and Q-Learning (CMDQL), Deep Q-Network (DQN), Dueling DQN, and A* in both static and dynamic scenarios. It reduces the average path length by 7.12% (vs. DQN), 4.26% (vs. A*), and 9.47% (vs. Dueling DQN), and cuts the estimated travel time by up to 8.36%. Notably, the convergence speed improves by 92.58% over DQN and 96.71% over Dueling DQN. This work delivers an efficient, empirically validated framework for intelligent non-motorized navigation, supporting sustainable urban mobility.

Evaluating residents' subjective perceptions of the rural environment is crucial for formulating effective rural planning. Due to the difficulty in obtaining rural data and the limitations of traditional questionnaire survey methods, existing research mostly focuses on small-scale perception evaluations in specific areas, making it difficult to reveal rural perception characteristics at the regional scale. To address this issue, our study proposes a rural living environment perception evaluation model based on street view images and deep neural networks, achieving a quantitative evaluation of rural perception in 118 cities nationwide. We collected a large data set of rural street view images nationwide through crowdsourcing and established an index system comprising five subjective perception dimensions: wealthy, tidy, lively, habitable, and terroir. By training a multidimensional quantitative evaluation model, we comprehensively evaluated residents' subjective perceptions of China's rural environment. Furthermore, we explored the relationship between these subjective perceptions and objective socioeconomic indicators. The model achieves an average evaluation accuracy of 75 percent across five dimensions, with the wealthy dimension exceeding 80 percent. Rural environment perception is comprehensively influenced by various factors such as economic base and traditional feature protection, showing significant differences between different regions. The perception of rural environment in the eastern region is closely related to economic levels, whereas perception in the western region is more affected by infrastructure improvement and social development. Overall, this study provides scientific evidence for formulating more targeted and effective rural planning.

Cross-domain hyperspectral image (HSI) classification (HSIC) addresses the challenge of real-time labeling of new regions. To mitigate the performance decline caused by unseen classes, a few-shot learning (FSL) method is used. However, these methods fail to fully consider the problem of sample scarcity and classification imbalance due to FSL methods. In addition, the issue of category confusion stemming from localized spectral fluctuations within the same class is commonly overlooked. To solve these problems, a geographical dual-prior guided few-shot network (Gprior-FSN) is proposed. In Gprior-FSN, combining prior knowledge of the first law of geography, a geographical prior guided bicorrelated (G-B) sample enhancement mechanism is proposed which includes geospatially correlated enhancement (GCE) and spectral feature correlated enhancement (SFCE). GCE uses a hierarchical sampling strategy to tackle the inherent imbalance problem for FSL methods. Subsequently, GCE mitigates sample scarcity via neighborhood sample expansion while identifying candidate pseudosamples with geospatial correlation. To make the acquired pseudosamples of the same category bicorrelated in both geospatial and spectral features, G-B combining spectral feature clustering and probabilistic statistics mechanism is designed. Inspired by the second law of geography, Gprior-FSN uses a spatial constraint mechanism to effectively enhance intraclass similarity by reducing spatial local heterogeneity, while improving global interclass discriminability. Finally, to further capture representative spatial-spectral feature, a weighted dual feature fusion network is designed. Experimental results from three distinct HSI datasets show that Gprior-FSN outperforms advanced HSIC methods in both efficiency and accuracy. In addition, the Gprior-FSN demonstrates strong generalization performance on real GF-5 image.

For large-scale mapping applications, cross-domain hyperspectral image classification (HSIC) has emerged as a highly promising research area. However, the classification accuracy decreased significantly when unseen classes emerged. Few shot learning (FSL) methods are adopted in cross-domain HSIC methods to address this problem. Despite this, existing cross-domain HSIC methods still have three key issues that hamper their classification capabilities: 1) previous works struggle to balance incorporating distinctive intradomain knowledge and managing model complexity in the face of significant domain representation differences; 2) previous works inadequately consider the limited capture capacity of interdomain intrinsic mutually invariant structures; and 3) previous works fail to capture the distinct characteristics of both head categories (e.g., urban buildings) and tail categories (e.g., urban corn) simultaneously when applying FSL to deal with unseen classes problem. In this article, we propose a cycle-resemblance few-shot transformation (CF-Trans) network to effectively handle the aforementioned challenges by integrating intradomain distinctiveness with interdomain invariance. To facilitate efficient intradomain feature aggregation for HSI, a novel lightweight intradomain attentive network is introduced. Different from previous works, to reduce the negative impact caused by inaccurate classifier predictions, from the perspective of interdomain knowledge transformation, a cycle-resemblance adversarial network is designed to capture the intrinsic mutually invariant structures. A dynamic label expansion mechanism is designed to capture the distinctive intradomain features of the head and tail classes. Experimental results on six HSI datasets including agricultural, rural-urban and urban datasets show the remarkably performance of our network.

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.

Parallel computing techniques have been adopted in geospatial cellular automata (CA) models to improve computational efficiency, enabling large-scale complex simulations of land use and land cover (LULC) changes at fine scales. However, the spatial distribution of computational intensity often changes along with the spatiotemporal dynamics of LULC during the simulation, leading to an increase in load imbalance among computing units and degradation of the computational performance of a parallel CA. This paper presents a dynamic load balancing method based on hypergraph partitioning for multi-process parallel geospatial CA models. During the simulation, the sub-domains are dynamically reassigned to computing processes through hypergraph partitioning according to the spatial variation in computational workloads to restore load balance. In addition, a novel mechanism called Migrated-SubCellspaces-First (MSCF) is proposed to reduce the cost of workload migration by employing a non-blocking communication technique to further improve computational performance. To demonstrate and evaluate the effectiveness of our method, a parallel geospatial CA model with hypergraph-based dynamic load balancing is developed. Experiments using a dataset from California showed that the proposed dynamic load balancing method achieved a computational performance enhancement of 62.59% by using 16 processes compared with a parallel CA with static load balancing.