Abstract: In recent years, global climate change and human activities have accelerated the changes in terrestrial ecosystem structure and functions. Consequently, ecosystem service assessment has gradually become a research focus in the fields of ecology and resource and environmental management. Advancements in integrated space-air-ground monitoring systems and the development of multisource data fusion technologies have led to an evolution for the methods used to assess ecosystem services. These methods are transitioning from conventional empirical estimation towards process-based simulation, multi-model integration, and comprehensive evaluation. However, most previous studies have largely focused on individual models, lacking systematic comparison and in-depth investigation of the construction principles, applicable conditions, and uncertainties associated with different types of assessment models. We systematically review the development and evolution of terrestrial ecosystem service assessment models. These models are categorized into four types according to their underlying mechanisms: process-based models, remote sensing-based models, empirical models, and interdisciplinary integrated models. The applications, advantages, and limitations of these models are comparatively analyzed in the assessment of key ecosystem services, including carbon storage, water conservation, soil retention, and biodiversity. A comprehensive analysis indicates that ecosystem service assessment models have been widely applied and have reached a relatively mature stage, but they still face three major challenges. (1) Data sources primarily rely on existing datasets derived from non-in-situ observations and proxy indicators for ecosystem services, which diminishes the credibility of assessment outcomes. (2) Models excessively depend on simplified assumptions and empirical parameters with insufficient validation and uncertainty analysis. (3) Spatial scales fail to align with ecological processes, which compromises the interpretability and applicability of model results. In response to these challenges and current development trends, we put forward suggestions for future improvements from the following three perspectives: model parameter localization, the integration of multi-source data and artificial intelligence, and multi-model coupling. The findings provide a basis for selecting, optimizing and integrating terrestrial ecosystem service assessment models, aiming to provide better supports for land use spatial planning and resource management.

Key words: terrestrial ecosystem services, assessment model, carbon storage, soil conservation, biodiversity