skip to main content


The NSF Public Access Repository (NSF-PAR) system and access will be unavailable from 11:00 PM ET on Thursday, May 23 until 2:00 AM ET on Friday, May 24 due to maintenance. We apologize for the inconvenience.

Title: Anomalous human activity fluctuations from digital trace data signal flood inundation status
The emergence of mobile platforms equipped with Global Positioning System technology enables real-time data collection affording opportunities for mining data applicable to rapid flood inundation assessment. The collected data can be employed to complement existing methods for rapid flood inundation assessment, such as remote sensing, to enhance situational awareness. In particular, telemetry-based digital trace data related to human activity have intrinsic advantages to be used for inundation assessment. In this study, we investigate the use of Mapbox telemetry data, which provides human activity indices with high spatial and temporal resolutions, for application in rapid flood inundation assessment. Using data from Hurricane Harvey in 2017 in Harris County, Texas, we (1) study anomalous fluctuations in human activities and analyze the differences in activity level between inundated and non-inundated areas and (2) investigate changes in the concentration of human activity, to explore the disruption of human activity as an indicator of flood inundation. Results show that both analyses can provide valuable rapid insights regarding flood inundation status. Anomalous activities can be significantly higher/lower in flooded areas compared with non-flooded areas. Also, the concentration of human activity during the flood propagation period across affected watersheds can be observed. This study contributes to the state of knowledge in smart flood resilience by investigating the application of ubiquitous telemetry-based digital trace data to enhance rapid flood inundation assessment. Accordingly, the use of such digital trace data could provide emergency managers and public officials with valuable insights to inform impact evaluation and response actions.  more » « less
Award ID(s):
Author(s) / Creator(s):
; ; ;
Date Published:
Journal Name:
Environment and Planning B: Urban Analytics and City Science
Page Range / eLocation ID:
1893 to 1911
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Research in different agricultural sectors, including in crop loss estimation during flood and yield estimation, substantially rely on inundation information. Spaceborne remote sensing has widely been used in the mapping and monitoring of floods. However, the inability of optical remote sensing to cloud penetration and the scarcity of fine temporal resolution SAR data hinder the application of flood mapping in many cases. Soil Moisture Active Passive (SMAP) level 4 products, which are model-driven soil moisture data derived from SMAP observations and are available at 3-h intervals, can offer an intermediate but effective solution. This study maps flood progress in croplands by incorporating SMAP surface soil moisture, soil physical properties, and national floodplain information. Soil moisture above the effective soil porosity is a direct indication of soil saturation. Soil moisture also increases considerably during a flood event. Therefore, this approach took into account three conditions to map the flooded pixels: a minimum of 0.05 m3m−3 increment in soil moisture from pre-flood to post-flood condition, soil moisture above the effective soil porosity, and the holding of saturation condition for the 72 consecutive hours. Results indicated that the SMAP-derived maps were able to successfully map most of the flooded areas in the reference maps in the majority of the cases, though with some degree of overestimation (due to the coarse spatial resolution of SMAP). Finally, the inundated croplands are extracted from saturated areas by Spatial Hazard Zone areas (SHFA) of Federal Emergency Management Agency (FEMA) and cropland data layer (CDL). The flood maps extracted from SMAP data are validated with FEMA-declared affected counties as well as with flood maps from other sources. 
    more » « less
  2. Abstract

    Flood nowcasting refers to near-future prediction of flood status as an extreme weather event unfolds to enhance situational awareness. The objective of this study was to adopt and test a novel structured deep-learning model for urban flood nowcasting by integrating physics-based and human-sensed features. We present a new computational modeling framework including an attention-based spatial–temporal graph convolution network (ASTGCN) model and different streams of data that are collected in real-time, preprocessed, and fed into the model to consider spatial and temporal information and dependencies that improve flood nowcasting. The novelty of the computational modeling framework is threefold: first, the model is capable of considering spatial and temporal dependencies in inundation propagation thanks to the spatial and temporal graph convolutional modules; second, it enables capturing the influence of heterogeneous temporal data streams that can signal flooding status, including physics-based features (e.g., rainfall intensity and water elevation) and human-sensed data (e.g., residents’ flood reports and fluctuations of human activity) on flood nowcasting. Third, its attention mechanism enables the model to direct its focus to the most influential features that vary dynamically and influence the flood nowcasting. We show the application of the modeling framework in the context of Harris County, Texas, as the study area and 2017 Hurricane Harvey as the flood event. Three categories of features are used for nowcasting the extent of flood inundation in different census tracts: (i) static features that capture spatial characteristics of various locations and influence their flood status similarity, (ii) physics-based dynamic features that capture changes in hydrodynamic variables, and (iii) heterogeneous human-sensed dynamic features that capture various aspects of residents’ activities that can provide information regarding flood status. Results indicate that the ASTGCN model provides superior performance for nowcasting of urban flood inundation at the census-tract level, with precision 0.808 and recall 0.891, which shows the model performs better compared with other state-of-the-art models. Moreover, ASTGCN model performance improves when heterogeneous dynamic features are added into the model that solely relies on physics-based features, which demonstrates the promise of using heterogenous human-sensed data for flood nowcasting. Given the results of the comparisons of the models, the proposed modeling framework has the potential to be more investigated when more data of historical events are available in order to develop a predictive tool to provide community responders with an enhanced prediction of the flood inundation during urban flood.

    more » « less
  3. Sankey, Temuulen ; Van Den Broeke, Matthew (Ed.)
    Rapid impact assessment of cyclones on coastal ecosystems is critical for timely rescue and rehabilitation operations in highly human-dominated landscapes. Such assessments should also include damage assessments of vegetation for restoration planning in impacted natural landscapes. Our objective is to develop a remote sensing-based approach combining satellite data derived from optical (Sentinel-2), radar (Sentinel-1), and LiDAR (Global Ecosystem Dynamics Investigation) platforms for rapid assessment of post-cyclone inundation in nonforested areas and vegetation damage in a primarily forested ecosystem. We apply this multi-scalar approach for assessing damages caused by the cyclone Amphan that hit coastal India and Bangladesh in May 2020, severely flooding several districts in the two countries, and causing destruction to the Sundarban mangrove forests. Our analysis shows that at least 6821 sq. km. land across the 39 study districts was inundated even after 10 days after the cyclone. We further calculated the change in forest greenness as the difference in normalized difference vegetation index (NDVI) pre- and post-cyclone. Our findings indicate a <0.2 unit decline in NDVI in 3.45 sq. km. of the forest. Rapid assessment of post-cyclone damage in mangroves is challenging due to limited navigability of waterways, but critical for planning of mitigation and recovery measures. We demonstrate the utility of Otsu method, an automated statistical approach of the Google Earth Engine platform to identify inundated areas within days after a cyclone. Our radar-based inundation analysis advances current practices because it requires minimal user inputs, and is effective in the presence of high cloud cover. Such rapid assessment, when complemented with detailed information on species and vegetation composition, can inform appropriate restoration efforts in severely impacted regions and help decision makers efficiently manage resources for recovery and aid relief. We provide the datasets from this study on an open platform to aid in future research and planning endeavors. 
    more » « less
  4. The Amazon River basin harbors some of the world’s largest wetland complexes, which are of major importance for biodiversity, the water cycle and climate, and human activities. Accurate estimates of inundation extent and its variations across spatial and temporal scales are therefore fundamental to understand and manage the basin’s resources. More than fifty inundation estimates have been generated for this region, yet major differences exist among the datasets, and a comprehensive assessment of them is lacking. Here we present an intercomparison of 29 inundation datasets for the Amazon basin, based on remote sensing only, hydrological modeling, or multi-source datasets, with 18 covering the lowland Amazon basin (elevation < 500 m, which includes most Amazon wetlands), and 11 covering individual wetland complexes (subregional datasets). Spatial resolutions range from 12.5 m to 25 km, and temporal resolution from static to monthly, spanning up to a few decades. Overall, 31% of the lowland basin is estimated as subject to inundation by at least one dataset. The long-term maximum inundated area across the lowland basin is estimated at 599,700 ± 81,800 km² if considering the three higher quality SAR-based datasets, and 490,300 ± 204,800 km² if considering all 18 datasets. However, even the highest resolution SAR-based dataset underestimates the maximum values for individual wetland complexes, suggesting a basin-scale underestimation of ~10%. The minimum inundation extent shows greater disagreements among datasets than the maximum extent: 139,300 ± 127,800 km² for SAR-based ones and 112,392 ± 79,300 km² for all datasets. Discrepancies arise from differences among sensors, time periods, dates of acquisition, spatial resolution, and data processing algorithms. The median total area subject to inundation in medium to large river floodplains (drainage area > 1,000 km²) is 323,700 km². The highest spatial agreement is observed for floodplains dominated by open water such as along the lower Amazon River, whereas intermediate agreement is found along major vegetated floodplains fringing larger rivers (e.g., Amazon mainstem floodplain). Especially large disagreements exist among estimates for interfluvial wetlands (Llanos de Moxos, Pacaya-Samiria, Negro, Roraima), where inundation tends to be shallower and more variable in time. Our data intercomparison helps identify the current major knowledge gaps regarding inundation mapping in the Amazon and their implications for multiple applications. In the context of forthcoming hydrology-oriented satellite missions, we make recommendations for future developments of inundation estimates in the Amazon and present a WebGIS application ( we developed to provide user-friendly visualization and data acquisition of current Amazon inundation datasets. 
    more » « less
  5. null (Ed.)
    As sea level rises, urban traffic networks in low-lying coastal areas face increasing risks of flood disruptions. Closure of flooded roads causes employee absences and delays, creating cascading impacts to communities. We integrate a traffic model with flood maps that represent potential combinations of storm surges, tides, seasonal cycles, interannual anomalies driven by large-scale climate variability such as the El Niño Southern Oscillation, and sea level rise. When identifying inundated roads, we propose corrections for potential biases arising from model integration. Our results for the San Francisco Bay Area show that employee absences are limited to the homes and workplaces within the areas of inundation, while delays propagate far inland. Communities with limited availability of alternate roads experience long delays irrespective of their proximity to the areas of inundation. We show that metric reach, a measure of road network density, is a better proxy for delays than flood exposure. 
    more » « less