Fusing Earth observation embeddings and spaceborne terrain with meteorological forcing for flood peak depth prediction from high-water marks

Wen Zhou; Shaowen Wang; Upmanu Lall; Venkatesh Merwade; Dingqi Ye

cs.LGNov 3, 2026

Fusing Earth observation embeddings and spaceborne terrain with meteorological forcing for flood peak depth prediction from high-water marks

Authors: Wen Zhou, Shaowen Wang, Upmanu Lall, Venkatesh Merwade, Dingqi Ye

Organizations: Department of Geography and Geographic Information Science · CyberGIS Center for Advanced Digital and Spatial Studies · Siebel School of Computing and Data Science · School of Complex Adaptive Systems, College of Global Futures, Arizona State University, Tempe, AZ, USA ·

Abstract

Flood peak depth on normally dry land governs damage and risk to life, but is observed only sparsely, by post-event high-water mark (HWM) surveys, and depth derived from satellite flood extent is produced after the event and inherits the vertical error of the digital elevation model (DEM) it is computed from. Deep-learning flood models have in turn been trained on hydrodynamic simulations rather than on observed depths. We ask whether flood peak depth can be learned from field observations and predicted before the peak from spaceborne terrain, geospatial foundation-model embeddings and antecedent meteorological forcing. We assembled 5689 USGS HWM surveys from 39 events (2011-2025) in 24 U.S. states and Puerto Rico, assigned each a peak time from nearby sensors, paired each with a rain-free window at the same site, and linked them to Copernicus GLO-30 terrain, AlphaEarth embeddings and 144 h of hourly forcing masked to the lead time. A cross-modal fusion network trained on these observations predicts depth at the surveyed marks with a mean absolute error of 0.39 m 6 h ahead of the peak. Supervision by surveyed depth corrects the DEM error implicitly: differencing the measured peak water level with the same DEM gives 0.93 m at the same marks, 2.4 times larger, although it uses the measured water level and the model does not. Forcing accounts for 77% of the attributed reduction in depth error; terrain and the embeddings are statistically equivalent to each other and contribute the remaining 23%. On eleven held-out events the model still separates flooded from rain-free conditions but no longer resolves depth at the marks (R2 = 0.03), so the skill reported here is confined to the events and regions sampled. Field-surveyed depths can thus supervise flood depth prediction and correct terrain error that extent-based estimation cannot, while transfer to unseen storms remains the open problem.

Figures & tables

Flood degree	Level 1	Level 2	Level 3
Depth $d$	$d\leq 0.3$ m	$0.3~\text{m}<d\leq 1.0$ m	$d>1.0$ m

Table 1: Flood severity levels used by the auxiliary classification task

Input	Product and source	Spatial resolution and footprint	Temporal coverage	Role
Labels and reference data
High-water marks	USGS Short-Term Network (STN) Flood Event Viewer surveys	Point	39 events, 2011–2025	Labels: peak elevation and depth; sample locations
Peak timing	USGS peak summaries, water-level sensors and rapid-deployment gauges within 10 km of the HWM	Point	Per event	Assigns the flood peak time ( $t=0$ )
Ground elevation at the HWM	USGS 3DEP lidar-based elevation	Point	Static	Anchor of the residual elevation target (Section 3.3)
Rain-free windows	NOAA Analysis of Record for Calibration (AORC) hourly precipitation	Gridded, nearest cell	$\pm$ 8–60 days around the peak	Selects the rain-free windows
Model inputs

Table 2: Inputs of the benchmark, their sources and their role in TCM-Former.

Extractor	Lead time	Depth $H$ (m)		Residual $E_{r}$ (m)		Severity level
Extractor	Lead time	MAE	RMSE	MAE	RMSE	MIF1	MAF1
Three-tier descriptor	6 h	0.257	0.468	0.530	0.818	0.727	0.642
	12 h	0.280	0.486	0.528	0.805	0.718	0.643
	18 h	0.273	0.487	0.536	0.820	0.709	0.624
	24 h	0.271	0.489	0.543	0.816	0.717	0.639
TabPFN-TS	6 h	0.258	0.457	0.541	0.812	0.728	0.647

Table 3: Flood depth (primary task) and auxiliary-task skill of TCM-Former at different lead times with the three-tier hydrometeorological descriptor, and with the two learned temporal encoders at the 6-h lead time (mean over three seeds; s.d.

\leq

0.04 m for errors and

\leq

0.018 for F1 scores). H: flood depth (height above ground); Er: residual flood peak elevation; MIF1 and MAF1: micro- and macro-F1 of flood degree classification.

Modalities	Depth $H$ (m)		Residual $E_{r}$ (m)		Severity level
Modalities	MAE	RMSE	MAE	RMSE	MIF1	MAF1
DEM, Meteo, AE	0.257	0.468	0.530	0.818	0.727	0.642
DEM	0.431	0.643	0.828	1.136	0.429	0.378
AE	0.410	0.625	0.801	1.174	0.474	0.426
Meteo	0.310	0.538	0.727	1.148	0.629	0.559
DEM, Meteo	0.269	0.485	0.550	0.848	0.724	0.644

Table 4: Flood depth (primary task) and auxiliary-task skill for different modality combinations. MAE-H and RMSE-H: MAE and RMSE of flood depth (height above ground); MAE-Er and RMSE-Er: MAE and RMSE of the residual flood peak elevation; MIF1 and MAF1: micro- and macro-F1 of flood degree classification. 6-h lead time; mean over three seeds; errors in m.

Dataset	Depth $H$ (m)		Residual $E_{r}$ (m)		Severity level
Dataset	MAE	RMSE	MAE	RMSE	MIF1	MAF1
Random split	0.257	0.468	0.530	0.818	0.727	0.642
Event split	0.386	0.648	0.997	1.489	0.614	0.537

Table 5: Generalization of TCM-Former to unseen flood events (6-h lead time; mean over three seeds; errors in m). The random split is the test set of Table 3 (885 samples); the event split holds out 11 whole events (897 samples).

Sampling	Depth $H$ (m)		Residual $E_{r}$ (m)		Severity level
Sampling	MAE	RMSE	MAE	RMSE	MIF1	MAF1
4-intra	0.374	0.649	0.982	1.423	0.655	0.550
8-intra	0.359	0.582	0.775	1.090	0.650	0.577
16-intra	0.361	0.628	0.903	1.273	0.646	0.557
32-intra	0.356	0.620	0.787	1.128	0.623	0.562

Table 6: Generalization of TCM-Former to spatially disjoint sub-regions within HUC-4 watersheds (6-h lead time; mean over three seeds; errors in m).

Methods	Lead time	Depth $H$ (m)		Residual $E_{r}$ (m)		Severity level
	Lead time	MAE	RMSE	MAE	RMSE	MIF1	MAF1
Ours (TCM-Former)	6 h	0.257	0.468	0.530	0.818	0.727	0.642
TFT	6 h	0.261	0.456	0.759	1.152	0.694	0.632
LSTM+CNN-ViT	6 h	0.267	0.461	0.545	0.814	0.722	0.653
Majority class + constant depth (training median)	—	0.416	0.744	—	—	0.599	0.250
Class-prior random classifier + training mean depth	—	0.487	0.617	—	—	0.440	0.332

Table 7: Comparison of TCM-Former and baseline models at the 6-h lead time (random split; mean over three seeds; TFT depth predictions clipped at zero, as in its loss).

Subset of test HWMs	$n$	Median	Depth MAE (m)
(random split)		$\|$ GLO-30 $-$ 3DEP $\|$ (m)	WSE $-$ GLO-30	WSE $-$ 3DEP	TCM-Former
All HWMs	567	1.66	0.93	0.51	0.39
$\|$ GLO-30 $-$ 3DEP $\|$ $<$ 1 m	183	0.55	0.65	0.39	0.37
$\|$ GLO-30 $-$ 3DEP $\|$ 1–3 m	207	1.68	1.10	0.54	0.37
$\|$ GLO-30 $-$ 3DEP $\|$ $>$ 3 m	177	6.47	1.03	0.61	0.44
Tree canopy	99	5.30	1.08	0.69	0.45

Table 8: Depth error at the 567 surveyed HWMs of the random test set, by local GLO-30 error and land cover. Differenced depths subtract GLO-30 (converted to NAVD88, the datum of the marks) or the 3DEP lidar ground at the mark from the surveyed water-surface elevation (WSE), with negative depths set to zero, i.e. the depth an extent-based method would return given a perfect water surface; TCM-Former predicts depth at the 6-h lead time (mean over three seeds). Land cover from NLCD 2021 (tree canopy: forest and woody wetland; built-up: developed, low to high intensity).

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Variable	Description
time	Hourly flood peak timestamp string
temperature_2m	Air temperature at 2 meters above ground
relative_humidity_2m	Relative humidity at 2 meters above ground
dewpoint_2m	Dew point temperature at 2 meters above ground
apparent_temperature	“Feels like” temperature (derived from temperature, wind, humidity, radiation, etc.)
precipitation	Total precipitation (commonly rain + snow water equivalent)

Appendix

Table A1: The used time-series factors obtained from Open-Meteo API

Flood Forecasting

Flood forecasting predicts river, coastal, and urban water levels or inundation to support earlier warnings and flood response.

See all 20 papers

Jul 27, 2026

Physics-Informed CNN-LSTM for Street-Scale Urban Flood Prediction: Reconciling Aggregate Accuracy and Street-Level Plausibility

Deep learning surrogate models trained with mean-squared-error loss produce statistically accurate but physically unconstrained flood predictions: water may flow uphill, appear spontaneously, or smooth over street-level corridors. We develop a physics-informed training framework for CNN-LSTM models that predict urban flood depths at 15 min intervals over a 128x128 spatial grid. Three differentiable penalty terms are embedded into the loss: (i) a gravity loss penalizing depth increases against the water-surface-elevation gradient, (ii) a continuity loss enforcing local mass conservation with rainfall-adaptive thresholds, and (iii) a topography-aware false-alarm penalty modulated by the topographic wetness index (TWI). We evaluate on the Norfolk, Virginia flood dataset spanning two storm events (August 2017 and September 2022, 300 samples), with all variants trained on identical splits and robustness assessed over repeated random splits and leave-one-storm-out tests. A road-proximal evaluation restricted to a TWI-derived street mask quantifies street-level skill. The physics-constrained model achieves near-zero gravity violations (order 1e-6) and the highest street-channel recall (0.77 +/- 0.09 vs 0.44 +/- 0.10 for the unconstrained baseline), the capability most relevant to traffic routing, and its advantage more than doubles on a held-out storm; a uniform false-alarm variant attains 16% lower mean absolute error but suppresses street recall to 0.25. The TWI-modulated penalty reconciles this trade-off: it improves on the uniform variant on every metric, recovering 60% higher street recall at the lowest MAE among constrained variants and the best street-level F1. These results expose a fundamental tension between aggregate pixel-level error and application-specific physical plausibility, and show that terrain-aware loss modulation offers a principled resolution.

Luc DCosta, Yidi Wang, Jonathan L. Goodall +1

Chandra Robot Autonomy Lab, Department of Computer Science, University of Virginia, Charlottesville, VA 22904, USA · Link Lab, Multidisciplinary Research Center, University of Virginia, Charlottesville, VA 22904, USA · Hydroinformatics Research Group, Department of Civil and Environmental Engineering, University of Virginia, Charlottesville, VA 22904, USA

Convolutional RNNs Flood Inundation Mapping Physics-Informed ML Flood Forecasting Hydrology

Ask PDF

Jun 2, 2026

Advanced Flood Prediction with Physics-Guided Deep Learning: Combining UNet, FNO, and SAR/Optical Imagery

Accurate and scalable flood mapping remains challenging due to limited ground observations, heterogeneous terrain conditions, and the difficulty of enforcing hydrodynamic consistency within data-driven models. This work introduces a physics-guided deep learning framework that integrates multi-modal remote sensing (Sentinel-1 SAR, Sentinel-2 optical imagery, and DEM-derived terrain features) with constraints from the depth-averaged shallow water equations (SWE). The proposed hybrid architecture combines a UNet to capture fine-scale spatial details with a Fourier Neural Operator (FNO) to model basin-scale hydraulic interactions, while physics-informed residual losses ensure mass and momentum consistency. Evaluated across diverse floodplain settings, the hybrid model achieves an Intersection over Union of 0.82 and an F1 score of 0.90 for flood extent prediction, outperforming UNet-only and FNO-only baselines. Using hydrodynamic simulations as reference data, the model achieves an RMSE of 0.21 m for water depth and 0.15 m/s for flow velocity. Physics consistency is maintained, with low residuals and mass imbalance below 2.1%. Ablation studies confirm that removing physicsbased regularization significantly degrades performance, underscoring the value of physical constraints for stability and generalization. These results demonstrate that embedding hydrodynamic principles into deep learning yields more accurate, reliable, and physically coherent flood predictions, offering strong potential for operational monitoring and large-scale deployment.

Tewodros Syum Gebre, Jagrati Talreja, Leila Hashemi-Beni

Built Environment Department, College of Science and Technology, North Carolina A&T State University, Greensboro, NC 27405, USA · United Nations University Institute for Water, Environment and Health, Richmond Hill, ON, Canada

Flood Inundation Mapping Physics-Informed ML

Ask PDF

Jun 3, 2026

Data-efficient flood depth prediction through domain-aware coreset selection and tabular foundation models

Near-real-time flood depth prediction demands surrogate models that are accurate, fast, and transferable across watersheds. Supervised surrogates can match physics-based simulators in accuracy but need millions of training rows per watershed and cannot extrapolate beyond their original mesh. We propose a domain-aware coreset construction pipeline that conditions a tabular foundation model at inference time. The pipeline stratifies storms by return period and most-affected watershed, then samples hexagons with a target-aware spatial selector. With 0.7% of the per-watershed training pool, the model attains a mean

R^2

of 0.663 across nine Houston-area watersheds, within 98.5% of the supervised reference (

R^2

= 0.673). It transfers to held-out watersheds without task-specific retraining, staying ahead of a coreset-trained supervised baseline. On real storms it exceeds the supervised reference on a far out-of-distribution case and trails it on a mostly in-distribution one. Domain-aware coreset construction lets tabular foundation models deliver data-efficient, watershed-transferable flood predictions without per-watershed training.

Lipai Huang, Adithi Srinath, Manas Singh +2

Urban Resilience.AI Lab, Zachry Department of Civil and Environmental Engineering, Texas A&M University, College Station, TX, USA · Department of Computer Science and Engineering, Texas A&M University, College Station, TX, USA · Resilitix Intelligence LLC, Houston, TX, USA +1

Surrogate Modeling OOD Generalization Coreset Selection Spatiotemporal Forecasting Flood Forecasting

Ask PDF

Abstract

Flood Forecasting

Explore similar work

Physics-Informed CNN-LSTM for Street-Scale Urban Flood Prediction: Reconciling Aggregate Accuracy and Street-Level Plausibility

Advanced Flood Prediction with Physics-Guided Deep Learning: Combining UNet, FNO, and SAR/Optical Imagery

Data-efficient flood depth prediction through domain-aware coreset selection and tabular foundation models