Future Geomorphologic Changes under the Changing Climate

The evaluation of shoreline changes for a few decades into the future can be done by using historical satellite imageries or numerical shoreline modelling. This work investigates the use of neural networks for this purpose as an additional method. Its advantage is that it provides a more flexible method of data fitting than the linear regression involved with the use of the imageries. Further, this work proposes the use of futuristic data of wind and waves in place of historical ones in the numerical modelling so that the effect of changing climate can be accounted for. Three different types of shoreline stretches representing continuous and discontinuous coastlines along the central west coast of India are considered and shoreline erosion or accretion rate for future 36 years is evaluated using satellite imageries, as well as neural networks and numerical modelling. The latter is based on projected wind and wave data derived from a regional climate model. The future predictions from neural networks were more close to those from the numerical modeling than the direct extrapolation from the past satellite imageries and hence it is suggested that they might act as supplementary to the former when future changes are to be predicted. The networks can also be used to cross-check the numerical predictions as well as the predictions based only on satellite imageries as per the common practice.