Major disasters typically impact large areas, cause considerable damages, and result in significant human and economic losses. The timely and accurate estimation of impacts and damages is essential to better understand disaster conditions and to support emergency response operations. Geo-information drawn from various sources at multi spatial-temporal scales can be used for disaster assessments through a synthesis of hazard, exposure, and post disaster information based on pertinent approaches. Along with the increased availability of open sourced data and cooperation initiatives, more global scale geo-information, including global land cover datasets, has been produced and can be integrated with other information for disaster dynamic damage assessment (e.g., impact estimation immediately after a disaster occurs, physical damage assessment during the emergency response stage, and comprehensive assessment following an emergency response). Residential areas and arable lands affected by the flood disaster occurring from July to August 2015 in Myanmar were assessed based on satellite images, GlobeLand30 data, and other global open sourced information as a study case. The results show that integrating global open geo-information could serve as a practical and efficient means of assessing damage resulting from major disasters worldwide, especially at the early emergency response stage.

Semantic segmentation and height estimation in remote sensing imagery are two pivotal tasks for scene understanding, and they are highly interrelated. Although deep learning methods have achieved remarkable progress in these tasks in recent years, several challenges remain. Recent studies have shown that multi-task learning methods can enhance the complementarity of task-related features, thus maximizing the prediction accuracy of multiple tasks at a low computational cost. However, due to factors such as complex semantic categories and the inconsistent spatial scales of remotely sensed images, existing multi-task learning methods often fail to achieve better results on these two tasks. To address this issue, we propose CTME-Net, a novel architecture termed the Cross-Task Mutual Enhancement Network, designed to jointly perform height estimation and semantic segmentation tasks on remote sensing imagery. Firstly, to generate discriminative initial features for each task branch and activate dedicated pathways for cross-task feature disentanglement, we design a universal initial feature embedding module for each downstream task. Secondly, to address the impact of redundancy in general features during global–local fusion, we develop an Adaptive Task-specific Feature Distillation Module that enhances the model’s ability to acquire task-specific features. Finally, we propose a task feature interaction module to optimize features across tasks through mutual optimization, maximizing task-specific feature expression. We conduct extensive experiments on the ISPRS Vaihingen and Potsdam datasets to validate the effectiveness of our approach. The results demonstrate that our proposed method outperforms existing methods in both height estimation and semantic segmentation.