Author : Apoorva Ramesh Shastry
Publisher :
ISBN 13 :
Total Pages : pages
Book Rating : 4.:/5 (115 download)
Book Synopsis Improving Topography Data for Flood Modeling by : Apoorva Ramesh Shastry
Download or read book Improving Topography Data for Flood Modeling written by Apoorva Ramesh Shastry and published by . This book was released on 2019 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: Flood models predict inundation extents, and can be an important source of information for flood risk studies. Accurate flood models require high resolution and high accuracy digital elevation models (DEM); current global DEMs do not capture the topographic details in floodplains, and this often leads to inaccurate prediction of flood extents by flood models. Flood extents obtained from remotely sensed data provide indirect information about topography. Here, we attempt to use this information along with model predictions to produce better floodplain topography. To illustrate the importance of accurate DEMs, we build a hydraulic model of the Logone Floodplain in Cameroon. We use LISFLOOD-FP, a two-dimensional hydrodynamic model, to build our floodplain model, and incorporate the effects of small-scale, local features like man-made fish canals, fishnets and depressions. Fish canals and depressions are represented as sub-grid elements connected to the river channel, and the fishnet structure as a combination of weir and mesh screens. The Logone Floodplain model is calibrated manually by adjusting the DEM. Flood inundation predictions from the model are compared with classified Landsat images, and the DEM is adjusted accordingly. Manual calibration was performed during one year (2006) and validation over five years (2001 2005, 2007). The Logone model accurately predicts the measured discharge downstream of the floodplain, with NashSutcliffe Efficiency (NSE) of 0.95, indicating that the channel to floodplain flow was modeled accurately. The spatial pattern of inundation is captured with a mean hit rate of 54% and mean critical success (CSI) index of 36% before manual adjustment of the DEM, and 63% and 43% after DEM adjustment. We then develop an algorithm to systematically update DEMs, and eliminate the manual calibration step. We use a data assimilation-style scheme where predictions from an ensemble of particle DEMs are merged with satellite derived observations of flood extents. The algorithm we describe is a two-step process: first, we reduce the noise along the observed flood boundaries for all particle DEMs. Then, the model predictions from these modified DEMs are assimilated with observations using a particle batch smoother. We first implemented the algorithm for a synthetic test case, and explored the sensitivity of the algorithm to errors of various magnitude. We observed a significant improvement in accuracy in terms of RMSE, bias and standard deviation. Flood inundation maps produced from the final estimate DEMs also improved on its prior. We then adapt this algorithm and implement it to a real-world case in the Logone Floodplain. The main difference in the algorithm is that in addition to constraining the ground elevations along the flood boundary, we also constrain them within the inundated regions for the ensemble of particle DEMs. We find that the updated DEM from the algorithm produced flood maps that had higher values for hit rate and CSI, with a mean hit rate of 72% and CSI of 44%. When evaluated during a validation period not used in data assimilation, we find a mean hit rate of 66% and a mean CSI of 45%. We find that modifying the DEM significantly improved the inundation prediction capability of the model. The algorithm produces promising results, and this type of analysis can be performed in data-poor floodplains to obtain better DEMs where high resolution DEMs do not exist.