Keywords
flood mapping, water volumes, remote sensing, Google Earth Engine
Start Date
27-6-2018 10:40 AM
End Date
27-6-2018 12:00 PM
Abstract
Episodic flood events have a significant impact on ecosystems and human settlements, and can be important drivers for the water availability. This study maps the flood extension at a catchment scale, and determines the volumes associated with inundation events for selected return periods by coupling a water detection algorithm and three methods for water depth estimation. The study was carried out in the Namoi catchment of Australia by using the Google Earth Engine platform. The extension of inundated areas was obtained by applying the open water likelihood (OWL) algorithm on MODIS surface reflectance imagery. For the estimation of the associated water volumes, three different data driven methodologies were compared, all of which use digital elevation models (DEM) to obtain water depths. These involve the obtaining of the maximum elevation in the flooded polygons, and the use of the Cohen and Doble algorithms. Two DEM products were used, a 5 m resolution LiDAR dataset of the floodplain in the catchment and the 1 second SRTM derived elevation model. Flood volumes were compared with rainfall volumes and the discharge at several gauge stations located at different reaches of the river. Return periods were obtained from the probabilities of pixels being inundated in a year. The relation between flood volume estimations and the stream discharge varied depending on the gauge position in the catchment. Flood volume estimation was improved using methods that took into account the flood pattern connection with the channels. A single flood frequency curve was developed for the entire catchment.
Spatio temporal analysis of floods at a catchment scale
Episodic flood events have a significant impact on ecosystems and human settlements, and can be important drivers for the water availability. This study maps the flood extension at a catchment scale, and determines the volumes associated with inundation events for selected return periods by coupling a water detection algorithm and three methods for water depth estimation. The study was carried out in the Namoi catchment of Australia by using the Google Earth Engine platform. The extension of inundated areas was obtained by applying the open water likelihood (OWL) algorithm on MODIS surface reflectance imagery. For the estimation of the associated water volumes, three different data driven methodologies were compared, all of which use digital elevation models (DEM) to obtain water depths. These involve the obtaining of the maximum elevation in the flooded polygons, and the use of the Cohen and Doble algorithms. Two DEM products were used, a 5 m resolution LiDAR dataset of the floodplain in the catchment and the 1 second SRTM derived elevation model. Flood volumes were compared with rainfall volumes and the discharge at several gauge stations located at different reaches of the river. Return periods were obtained from the probabilities of pixels being inundated in a year. The relation between flood volume estimations and the stream discharge varied depending on the gauge position in the catchment. Flood volume estimation was improved using methods that took into account the flood pattern connection with the channels. A single flood frequency curve was developed for the entire catchment.
Stream and Session
Stream B: (Big) Data Solutions for Planning, Management, and Operation and Environmental Systems
B3: Sixth Session on Data Mining as a Tool for Environmental Scientists (S-DMTES-2018)