Monitoring morphological changes in river deltas exploiting GEE and the full Landsat archive

River deltas are considered among the most diverse ecosystems with significant environmental and agricultural importance. These landscapes are vulnerable to any human activity or natural process that can disturb the fragile balance between water and land, thus causing morphological changes in the delta fronts. Earth observation (EO), with its capability to provide systematic, inter-temporal and cost-effective data provides a promising potential in monitoring the dynamic changes of river deltas. Recent advances in geoinformation technologies have allowed the development of cloud-based platforms for EO processing such as Google Earth Engine (GEE). It offers unique advantages such as rapid processing of a large amount of data in a cost and time-efficient manner. This study aims to assess the added value of GEE in monitoring the coastal surface area of river deltas based on the full Landsat archive (TM, ETM+, OLI, L9) and a machine learning (ML) technique. As a case study two river deltas, Axios & Aliakmonas, were selected located in northern Greece. Those are two of the largest rivers of the country, with Axios being also the second largest in the Balkans. Their joint river deltas create a fertile valley with great environmental and agricultural importance, which has also exhibited very strong dynamics in terms of its morphological characteristics over the last decades. In order to gain a better insight into the coastal dynamics of the studied region, Landsat multi-spectral data covering the period 1984 - present time was integrated into GEE and a Machine Learning (ML) classification approach was developed in the cloud-based environment. The two rivers delta dynamics were also mapped independently using photo interpretation serving as our reference dataset to map the river delta dynamics, in accordance to other studies. All the geospatial data analysis of the extracted morphological features of the river deltas was conducted in a geographical information system (GIS) environment. Our results evidenced the unique advantages of cloud platforms such as GEE, towards the operationalization of the investigated approaches for coastal morphological changes such as those found in the studied river deltas. Unique characteristics of the proposed herein methodology consist of the exploitation of the cloud-based platform GEE together with the advanced ML image processing algorithm and the full utilization of the Landsat images available today. The proposed approach also can be fully-automated and is transferable to other similar areas and can prove valuable help in understanding the spatiotemporal changes in coastal surface area over large areas.

KEYWORDS: Google Earth Engine, Landsat, Machine Learning, Earth Observation, river delta