Organizations: College of Geodesy and Geomatics, Shandong University of Science and Technology, Qingdao 266590, China · Professorship of Big Geospatial Data Management, School of Engineering and Design, Technical University of Munich, Munich, 85521, Bavaria, Germany · State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, China · Tsinghua Shenzhen International Graduate School, Tsinghua University, Shenzhen 518055, China · Helmholtz-Zentrum Dresden-Rossendorf, Responsible AI Group, 09599 Freiberg, Germany · University of Iceland, Faculty of Electrical and Computer Engineering, Reykjavik 102, Iceland · Chair of Data Science in Earth Observation, Technical University of Munich, 80333 Munich, Germany · Munich Center for Machine Learning, 80333 Munich, Germany · Department of Geography, National University of Singapore, Singapore 117568, Singapore
Crowdsourced imagery provides timely, fine-grained, street-level observations for disaster mapping, complementing conventional remote sensing imagery (RSI) during emergency response. However, such imagery is often unstructured, spatially ambiguous, and lacks reliable geographic metadata, making manual geolocalization and interpretation labor-intensive and difficult to scale. This work proposes a multi-task Geospatial Reasoning Disaster mapping framework, namely GRDisaster, to examine the potential of vision-language models (VLMs) in understanding, geolocalizing, and reasoning over crowdsourced disaster imagery. GRDisaster is built on a newly curated benchmark dataset derived from PhotoMappers, comprising 26,340 images organized into human-validated volunteered geographic information (VGI), street-view imagery (SVI), RSI cross-view triplets covering multiple disaster events from 2018 to 2024. The framework combines deterministic and probabilistic cross-view geolocalization with multi-view fusion to associate VGI images with georeferenced SVI and RSI. It introduces two sets of spatial reasoning indicators for cross-view geolocalization validation and disaster damage assessment. These indicators use structural, environmental, and global-scene cues to validate cross-view correspondences and visually observable damage evidence with expert-verified annotations to assess disaster severity, improving the interpretability of VLM outputs. To our knowledge, this study provides the first systematic investigation and unified evaluation framework for examining how VLM-based spatial reasoning can transform crowdsourced disaster imagery into actionable geospatial artificial intelligence (GeoAI) through cross-view geolocalization validation, interpretable spatial reasoning, and damage-aware severity assessment.
Figures & tables
Fig. 1: Examples from the PhotoMappers dataset: (a) crowdsourced VGI and (b) corresponding RSI across different disaster types.
Fig. 2: Overview of the three interconnected tasks in the GRDisaster framework using the VGI–SVI–RSI benchmark: deterministic and probabilistic cross-view geolocalization, and explainable VLM-based reasoning for match verification and damage severity classification.
Fig. 3: Overview of deterministic and probabilistic cross-view geolocalization.
No.
Indicator name
Evaluation focus
Dimension
Appearance
Spatial layout
Landmark cue
1
Building type and scale
Building and structure: Building appearance consistency
✓
2
Overall building appearance
Building and structure: Building appearance consistency
✓
3
Building position relative to nearby roads or open space
Building and structure: Building spatial-position consistency
✓
4
Building position and shape correspondence in SVI–RSI
Building and structure: Building spatial-position consistency
✓
5
Recognizable building outline
Building and structure: Corresponding building landmark
✓
TABLE I: Evaluation indicators for VLM-based interpretive reasoning of cross-view geolocalization results.
Indicator name
Damage severity
Low
Medium
High
Flooding and water impact
Shallow or geolocalized water with limited impact on visible objects.
Clear flooding affects roads, buildings, vehicles, or public spaces.
Deep water, strong flow, storm surge, or widespread inundation.
Building and structural damage
Buildings are mostly intact , with no clear or only minor damage.
Partial damage is visible, such as doors, windows, walls, roofs, or water intrusion .
Severe structural damage, collapse , heavy inundation, or loss of building function.
Road and access disruption
Roads or paths remain mostly usable , with minor debris or shallow water.
Roads are partially blocked or difficult to access due to water, debris, vehicles, or fallen trees.
Roads are impassable or heavily disrupted by deep water, large debris, collapse, or severe surface damage.
Vegetation and natural damage
Vegetation remains intact , with only small branches or geolocalized disturbance.
Fallen trees, broken branches, or damaged vegetation partially affect roads, buildings, or open spaces.
Severe vegetation damage, large fallen trees, or widespread natural disturbance blocks access or damages objects.
Infrastructure and facility damage
Facilities are mostly intact , with only minor or geolocalized disturbance.
Some facilities are damaged or disrupted , such as poles, signs, fences, utilities, or public equipment.
Critical infrastructure or multiple facilities are severely damaged, submerged, collapsed , or nonfunctional .
TABLE II: Image-based indicators for VLM-based disaster damage severity reasoning and classification.
Fig. 4: Deterministic cross-view geolocalization performance under different retrieval and fusion settings: (a) VGI–RSI retrieval, (b) VGI–SVI retrieval, (c) SVI–RSI retrieval, and (d) VGI–SVI–RSI fusion with and without Sinkhorn re-ranking.
Method
Type
View Input
Median (km)
Mean (km)
@1km
@25km
@200km
@750km
@2500km
DINOv3-B
×
VGI–RSI
381.47
747.35
19.13
23.29
37.81
68.17
93.96
DINOv3-L
×
VGI–RSI
218.67
525.35
28.25
33.20
48.35
76.25
96.18
DINOv3-B
×
VGI–SVI
340.37
631.42
25.11
28.59
40.43
70.39
95.10
DINOv3-L
×
VGI–SVI
193.99
545.46
31.44
35.71
50.34
75.40
96.58
DINOv3-B
×
VGI–SVI–RSI
206.96
547.83
31.78
35.99
49.43
74.94
95.90
DINOv3-L
×
VGI–SVI–RSI
65.31
390.08
41.40
45.62
58.60
81.44
98.35
TABLE III: Comparison of geolocalization performance between generative and retrieval-based approaches. Median and mean are errors in kilometers, while @1,km–@2500,km are the percentages of predictions within each distance threshold.
Fig. 5: Heatmaps for (a) and (b) correctly and (c) and (d) incorrectly geolocalized imagery pairs generated by our best-performing DINOv3-L model with VGI–SVI–RSI fusion and Sinkhorn re-ranking, with the corresponding raw VGI ground-truth images included for reference. The first column shows the ground-truth images, and the second column shows the predicted results.
No.
Indicator Abbr.
VLM vs Exp. A
VLM vs Exp. B
Exp. A vs Exp. B
Qwen
GPT
InternVL
Agr.
κw
Agr.
κw
Agr.
κw
Agr.
κw
Agr.
κw
Agr.
κw
1
Building type and scale
0.954
0.964
0.991
0.985
0.957
0.975
0.644
0.434
0.708
0.610
0.537
0.376
2
Overall building appearance
0.983
0.990
0.874
0.921
0.862
0.914
0.632
0.429
0.702
0.613
0.504
0.364
3
Building-road-open space position
0.940
0.965
0.991
0.995
0.948
0.970
0.637
0.407
0.694
0.585
0.517
0.355
4
Building shape correspondence
0.989
0.989
0.968
0.977
0.974
0.985
0.649
0.415
0.710
0.578
0.522
0.351
5
Recognizable building outline
0.991
0.990
0.991
0.990
0.994
0.997
0.633
0.426
0.724
0.630
0.556
0.396
TABLE IV: Agreement and quadratic weighted kappa ( κw ) evaluation for cross-view geolocalization reasoning. Bold the best-performing results in each category. Bold the best-performing results in each category.
Fig. 6: Spatial reasoning for cross-view geolocalization across disaster types. Three VLMs are compared across building and structure, road and transport, vegetation and terrain, objects and facilities, global scene (circles), and cross-view geolocalization (diamond), with different colors indicating reasoning results.
Indicator
Level
VLM vs Expert A
VLM vs Expert B
Expert A vs Expert B
Agr.
κw
Agr.
κw
Agr.
κw
Flooding and water impact
Low
0.966
0.954
0.932
0.906
0.932
0.908
Medium
0.940
0.963
0.940
0.948
0.949
0.953
High
0.940
0.964
0.932
0.957
0.974
0.984
Overall
0.949
0.968
0.934
0.953
0.952
0.964
Building and structural damage
Low
0.983
0.916
0.983
0.927
0.991
0.846
TABLE V: Agreement and quadratic weighted kappa ( κw ) evaluation between Gemini and two expert annotations for disaster damage severity assessment with 20% expert verification.
Indicator
Level
Qwen Agr.
Qwen κw
GPT Agr.
GPT κw
InternVL Agr.
InternVL κw
Flooding and water impact
Low
0.661
0.601
0.717
0.667
0.394
0.306
Medium
0.789
0.851
0.772
0.844
0.583
0.462
High
0.802
0.842
0.877
0.915
0.627
0.427
Overall
0.764
0.831
0.777
0.848
0.551
0.465
Building and structural damage
Low
0.829
0.581
0.927
0.591
0.423
0.175
Medium
0.735
0.757
0.764
0.810
0.239
0.161
TABLE VI: Agreement and quadratic weighted kappa evaluation of VLMs for disaster damage severity assessment using Gemini as a human annotation substitute. Bold the best-performing results in each category.
Fig. 7: Examples of VLM-based disaster damage reasoning across six dimensions. Three VLMs are compared in flooding and water impact, building and structural damage, road and access disruption, vegetation and natural damage, infrastructure and facility damage, and debris and scene disruption. The circles in each group follow this order, with different colors indicating accuracy.
Crowdsourced disaster imagery provides timely, ground-level observations when conventional disaster mapping is delayed or incomplete, yet its geographic value depends on where observations are produced, how they are interpreted, and whether they can be reliably located as actionable Geospatial Intelligence (GEOINT). Using more than a decade of PhotoMappers, a crowdsourced disaster-mapping initiative, this study examines the changing geography of disaster information and asks how emerging GeoAI models compare with human geographic interpretation. We combine longitudinal spatial analysis with cross-view geolocalization using street view imagery (SVI) and remote sensing imagery (RSI), and evaluate vision-language models (VLMs) reasoning through 30 indicators spanning geographic cues and reasoning evidence. The results reveal a geographically and temporally uneven archive shaped by changing disaster types, Community Lifelines, and operational priorities. Integrating VGI, SVI, and RSI with Sinkhorn re-ranking achieves 40.90% R@1, 61.10% R@5, and 69.45% R@10, while yearly R@1 varies markedly from 25.00% to 65.07%. Geographic reasoning is similarly uneven: Qwen3.6-27B achieves an overall agreement of 67% compared with 51% for InternVL3-8B, with both models showing greater difficulty with localized and place-specific evidence than broad spatial configurations. We argue that the value of GeoAI lies less in replacing volunteered geographic interpretation than in complementing it, scaling candidate retrieval while retaining human judgment for ambiguous and consequential cases. This provides a pathway toward more human-centered and responsible GeoAI for disaster mapping. All data and code are openly available at https://github.com/bobleegogogo/PhotoMappers-VLM.
Hao Li, Wenping Yin, Fabian Deuser +5
Department of Geography, National University of Singapore, Singapore · College of Geodesy and Geomatics, Shandong University of Science and Technology, China · University of the Bundeswehr Munich, Germany +6
Social media imagery (SMI) provides timely and fine-grained ground perspectives that are valuable for situational awareness and emergency response. Unlike satellite or aerial imagery, SMI can capture disaster impacts and ground-level conditions in a timely manner. However, geographic references in SMI are often vague or ambiguous, making accurate geolocalization challenging. To address this issue, we propose DisasterTD, a disaster toponym disambiguation framework that integrates multimodal large language model (MLLMs)-based semantic reasoning with cross-view geolocalization. First, MLLMs extract toponyms and generate candidate geolocations from noisy textual inputs. Then, cross-view matching between SMI, remote sensing imagery (RSI), and optionally street-view imagery (SVI) is used to verify and refine these candidate results. We evaluate DisasterTD on the Hurricane Harvey dataset, where SMI is augmented with collected RSI and SVI to construct a cross-view benchmark for disaster geolocalization. The dataset is divided into four categories based on toponym clarity and ambiguity, allowing a fine-grained performance analysis across scenarios. Results show that DisasterTD consistently outperforms MLLM-only and cross-view-only baselines without disambiguation, achieving geolocalization accuracies of 71.62% within 1000 m, 62.36% within 500 m, 57.99% within 250 m, 52.09% within 100 m, and 47.01% within 50 m, while reducing the mean and median errors to 11.33 km and 0.68 km, respectively. The largest improvements appear in ambiguous toponyms, where semantic reasoning with cross-view evidence reduces candidate dispersion and errors. These findings demonstrate the effectiveness of integrating MLLM-based candidate generation with cross-view verification for fine-grained disaster geolocalization.
Wenping Yin, Ziqi Liu, Naixia Mou +3
College of Geodesy and Geomatics, Shandong University of Science and Technology, Qingdao 266590, China · School of Environmental Science and Spatial Informatics, China University of Mining and Technology, Xuzhou 221116, China · State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, China +3
Remote-sensing vision-language models (RS-VLMs) have advanced Earth-observation analysis toward visual interpretation and instruction-following, yet fall short of operational geo-intelligence, which demands tool-grounded spatial reasoning and structured, evidence-backed decisions. We introduce GeoDisaster, an operational geospatial disaster reasoning benchmark with 2,921 verified instances across 43 question types and five task families: deforestation monitoring, multi-hazard analysis, building-damage assessment, flood-safe routing, and Sentinel-1 SAR flood monitoring. Instances integrate heterogeneous EO/GIS evidence-optical and SAR imagery, raster masks, vector geometries, road networks, and exposure layers-spanning hazard detection, damage assessment, exposure estimation, and diagnostic report generation. Ground-truth answers are grounded in executable geospatial workflows and deterministic consistency checks, removing the need for language-model annotation. We further propose an orchestrated multi-agent framework with 18 disaster-oriented tools, where role-specialized agents coordinate through explicit execution contracts, aligned via Role-Contract Expectation Alignment (RCEA): failure-aware supervised fine-tuning combined with contract-grounded reinforcement learning over dense step-level signals. Experiments show that GeoDisaster challenges existing RS-VLMs and agentic systems, while RCEA improves tool use, evidence grounding, state consistency, and decision generation.
Maram Hasan, Aman Verma, Savitra Roy +5
Indian Institute of Technology Bombay · Mohamed bin Zayed University of Artificial Intelligence