Situated within the geodynamically active Alpine-Himalayan orogenic belt, Lorestan Province in Iran hosts a remarkable diversity of geomorphosites, presenting significant potential for geotourism. This study employs an innovative artificial intelligence approach to model the relationship between the multidimensional ASPECTS assessment model—evaluating intrinsic, functional, effectiveness, and use values—and the morphological coding of these sites. Data from 47 selected geomorphosites were analyzed using the C&R Tree decision-making algorithm in SPSS Modeler. Results indicate that the intrinsic dimension is the most influential predictor of a site's morphological classification. Sites with active and rare point features scored highest in intrinsic and functional values, whereas those shaped by Quaternary alluvial processes demonstrated greater resilience and higher conservation potential. The research underscores the necessity for developing integrated management strategies tailored to the morphological and evaluative characteristics of each site to ensure the sustainable development of geotourism in Lorestan.

2. Methodology

This research employed a descriptive-analytical approach to model the relationship between the dimensions of the ASPECTS assessment model (intrinsic, functional, effectiveness, use) and the morphological coding of geomorphosites in Lorestan Province, Iran. Data were collected through structured questionnaires and systematic field observations from a purposive sample of 47 selected geomorphosites, chosen based on criteria of representativeness and rarity. The ASPECTS model was operationalized to evaluate each site across its core dimensions, translating qualitative attributes into quantifiable scores. Morphological coding classified sites based on form, dominant process, and scale, enabling systematic comparison.

The analysis was conducted using the C&R Tree decision tree algorithm within SPSS Modeler software, selected for its capability to handle both qualitative and quantitative data and identify complex, non-linear relationships. Prior to modeling, data underwent preprocessing, including cleaning, handling of missing values, and partitioning into training (80%) and testing (20%) subsets to ensure model robustness and generalizability. The model's performance was validated based on its accuracy and ensemble learning score. This artificial intelligence-driven methodology provided a powerful, interpretable framework for discerning key predictive variables and formulating data-informed management strategies tailored to specific morphological classes.

3. Results

This study demonstrates a significant and quantifiable relationship between the morphological characteristics of geomorphosites in Lorestan Province and their multidimensional valuation as per the ASPECTS model. The finding that the intrinsic dimension exerts the strongest influence (with a significance coefficient of 0.41) on a site's morphological code is pivotal. It underscores that inherent values—such as a site’s scientific representativeness, uniqueness, and geomorphological integrity—are the primary determinants of its fundamental classification. This aligns with the core principles of geoheritage assessment, which posit that inherent characteristics form the bedrock for any subsequent evaluation of functional or protective potential. Furthermore, the research delineates a clear functional dichotomy based on morphological coding. Sites classified as active and rare point features, such as certain waterfalls or caves, consistently achieved higher scores in both intrinsic and functional dimensions. Their distinctiveness and dynamism render them exceptionally valuable for scientific interpretation and geotourism appeal. Conversely, landforms originating from Quaternary alluvial processes, while often less distinctive intrinsically, demonstrated greater resilience and higher conservation scores. This inverse relationship highlights a critical management trade-off: sites of highest intrinsic value are often the most vulnerable, necessitating stringent protective measures and controlled visitation.

The application of an artificial intelligence algorithm (C&R Tree) proved instrumental in deciphering these complex, non-linear relationships between the ASPECTS model's dimensions. This methodological innovation moves beyond traditional statistical analysis, offering a powerful predictive tool for site prioritization. By accurately modeling how a site’s physical form influences its potential use and susceptibility to damage, the AI-driven approach facilitates evidence-based decision-making.

Consequently, this research provides a robust scientific foundation for developing differentiated management strategies. It advocates for a shift from generic planning to tailored interventions, where conservation protocols and tourism development plans are meticulously aligned with the specific morphological code and assessed potential of each geomorphosite. This ensures that the profound geoheritage of Lorestan can be leveraged for sustainable development while preserving its scientific and aesthetic value for future generations.

5. Conclusion

Based on the integration of the ASPECTS evaluation model and artificial intelligence through decision tree analysis, this study demonstrates that the intrinsic dimension—encompassing scientific representativeness, rarity, and geomorphological integrity—is the most influential factor (with a significance coefficient of 0/41) in determining the morphological coding of geomorphosites in Lorestan Province. Sites characterized by active and rare point features exhibit higher intrinsic and functional values, whereas those formed by Quaternary alluvial processes show greater resilience and higher conservation potential due to lower sensitivity. These findings underscore the necessity of developing tailored management strategies that align conservation efforts with the specific morphological and evaluative attributes of each site. The application of AI not only enhances the precision of geosite assessment but also supports sustainable geotourism planning by enabling data-driven prioritization. This approach offers a scalable framework for managing geoheritage in diverse mountainous regions, balancing preservation with responsible tourism development .