Assessment of land-use and land-cover changes from 1965 to 2014 in Tam Giang-Cau Hai Lagoon, central Vietnam

This study integrates the use of multi-source and multi-resolution remote sensing, topographic and field-based datasets to quantify land-use and land-cover (LULC) changes along a coastal stretch of Thua Thien Hue Province (central Vietnam). The LULC change analysis involves the Tam Giang-Cau Hai lagoon, the largest lagoon system in Southeast Asia, which is running nearly 70km along the coast and having about 22,000ha of water surface. The LULC change analysis was performed by computer-aided visual interpretation for 5 years (1965, 1989, 2000, 2006 and 2014) using satellite imagery from LANDSAT MSS, TM, ETM+ and 8, ASTER and SPOT5. National topographic maps were also used for the 1965 and 2000 years. To adequately represent the LULC features and peculiarities of central Vietnam coastal areas, an adapted CORINE Land Cover nomenclature was used where new 3rd and 4th level classes were adopted. Due to their intrinsic relative high spatial and radiometric resolution, SPOT5 images from 2006 were assumed as a reference for interpretation keys and first delineation. Changes were mapped by editing those vectors representing features which underwent LULC change prior or after 2006. Spatial and temporal changes were analyzed by post-classification approach and validated by ground truth information. High detail object-based classification was finally performed to infer the capability of medium spatial resolution imagery for extracting cadastral scale pond maps. The accuracy of classification was checked by a polygon by polygon comparison with an existing aquaculture facility inventory.Five LULC maps were obtained by applying a legend of 21 classes including two newly defined: "Aquaculture ponds" and "Mangrove forest". The overall classification accuracy of the LULC map is 85% while the KHAT statistics 0.81 for the year 2006. Accuracy of the object-based aquaculture facilities classification is 84% or better for the SPOT5 imagery and 47.9% for the ASTER imagery. The study provides a synoptic LULC representation for the largest lagoon system of Southeast Asia and delivers quantitative estimates of main changes occurred during the last 50 years. Moreover, it reveals the adaptability of the CORINE Land Cover method outside European environment. Finally, SPOT5 provides good results to map aquaculture features at cadastral scale, even if in some circumstances (e.g. tidal areas), the integration with higher spatial resolution multispectral sensors should be envisaged