ADB is co-financing the construction of Song Bung 4, a hydropower plant in the Nam Giang District, Quang Nam Province, in compliance with ADB’s safeguard requirements and the restoration and improvement of livelihoods of the affected population.
The objective is to contribute to meet Viet Nam's increasing power demand in an environmentally sustainable and socially inclusive manner. The construction of the dam and the removal of trees in the reservoir area requires new transmission lines, new access roads into forests and improvement of the existing regional and local road network. This will increase access to the forested areas in the vicinity of the reservoir and potentially increase illegal logging. Special concern is expressed regarding the Song Thanh Nature Reserve (STNR), of which the reservoir inundates an area of about 37.5 ha. Access to it will increase significantly through the reservoir and the realigned highway 14D.
ADB is involved in proper management and mitigation of environmental issues, and Earth Observation (EO) is a likely tool for independent and effective monitoring. In this respect, the EO support project contributed with baseline and change maps to be used as reference after plant operation start, allowing measuring environmental changes due to encroachment of human activity around the dam and the reservoir. Additionally, precise ground and dam displacements in the order of a few mm/year during and after dam construction were delivered, to characterise the stress that the filled reservoir exerts on the dam and the ground in its vicinity. Information on illegal logging and other undesired and desired effects near the reservoir, the relocation sites and the nearby SNTR was also provided.
Land cover / land use classification and its changes over time
In order to monitor land use changes related to the development of the Song Bung 4 dam, a land use map for each of the following three periods was created: 2001–2003, 2005–2007 and 2012–2015. These three periods are key points in time for the development in the area.
Historic satellite imagery proved valuable to get insight in what happened over a period of almost 15 years. It was challenging to generate comparable output from the wide variety of used sensors. In total 16 satellite images from SPOT 5, SPOT 6, Pléiades, WorldView-2, Landsat 7 and Landsat 8 were used for the land use mapping, all with different spatial resolutions and spectral bands. The challenge was thus to translate this imagery into understandable and easy-to-use maps that can help monitor the impact of the development of the dam. An object-based classification approach was chosen, where similar groups of pixels are combined and classified as one land use type. For the end user this provided an easy tool to distinguish patterns and calculate for example the size of a village or agricultural field.
The first study period represents the initial situation before development started. Some of the small villages along the Song Bung river were inundated after the reservoir filled up in the last period. The road network is rather limited in the entire study area.
During the second period saw mostly infrastructure development. The main construction of the dam started September 2010. The land use map of this period reflects this, with the appearance of roads to the dam site and the locations of the new settlement relocation areas. A clear decrease in natural vegetation was observed, mostly due to the development of infrastructure and agriculture.
The most recent period saw the dam being completed and the reservoir filled. Several villages were relocated to other areas as part of a complex resettlement and livelihood restoration process for indigenous and ethnic minorities.
The three land use maps of the different periods were combined into change maps. By combining certain land use classes detailed and meaningful information was provided about the location and type of occurresd changes. Good examples are the development of the relocation areas, or changes in natural vegetation, agriculture and the road network/infrastructure in the area. Based on the figures, the amount of natural vegetation has decreased, and the man-made areas have increased.
As change maps contain a large number of class combinations, these are still rather technical products, and not optimal for communicating change information. An additional number of products have therefore been produced aiming to present changes in such way that they provide direct meaning and value instead of just data. An example is the thematic map of deforestation, where the user gets direct insight where and into what natural vegetation has changed.
Digital Elevation Model (DEM)
A DEM is a useful source of information for various applications. In this particular project it was used to orthorectify satellite imagery, derive locations of rivers and streams and generate slope maps that can be used for future erosion risk analyses.
The significant changes in topography, mainly due to the construction of the dam and associated reservoir, required the use of multiple DEMs. A 30 m horizontal resolution DEM provided commercially by Airbus DS was used for the entire study area, representing the situation from before the construction of the dam.
Furthermore, a recent stereo pair of SPOT 6 images (February 2015) was acquired and used to generate a DEM covering the completed dam and fully inundated reservoir. The two DEMs were finally merged to be able to perform analysis for the entire study area based on the latest situation including the reservoir.
Terrain deformation mapping
Deformations of the dam and its immediate vicinity have been mapped using interferometric techniques (so-called PS-InSAR) to infer precise ground or dam displacements in the order of a few mm/year. The period of interest for monitoring was from April 2014 to January 2015, using data from the (commercial) TerraSAR-X satellite. This offers the hydropower project stakeholders the possibility to study changes in the stress that the filled reservoir exerts on the dam and the ground in its vicinity, with monitoring extending beyond the timeframe of the present project.
The results showed indications of which sides of the reservoir are susceptible to deformation, with the westward slope of the eastern reservoir wall deforming at rates of up to 10 mm/year. Considering the size of SAR images, this clearly demonstrates the technique’s possibility to regularly monitor large numbers of hydropower dams for reasons of safety, operational efficiency and/or quality of engineering execution after completion of construction.
Regarding the overall feasibilty of deformation mapping in these highly vegetated, mountainous and humid areas, a series of at least 25–30 images are necessary for optimal PS-InSAR analysis. The current results should be considered as a minimum result. Extensive analysis was also performed using ALOS PALSAR satellite data to extract deformation information for the period before dam construction, but the lower sensor resolution and the heavily forested nature of the area with only few rocky outcrops, lead to too much temporal decorrelation.
The current project demonstrates that EO is an effective tool to monitor land use changes over the period of the development.
EO products can however also be valuable during earlier phases of a project such as the development of a future hydropower dam. In the planning phase, satellite imagery derived land use maps could provide information about potentially affected settlements and natural environment. Combined with a DEM it could provide information about potential erosion risks caused by the development of a hydropower dam at proposed locations. It then becomes a decision support tool. Involvement from the very beginning of a project will furthermore deepen the understanding of the project.
EO products are especially valuable for remote areas where no or very limited geospatial information is available. Most of the potentially interesting areas for still to be developed hydropower dams are in remote areas.
Continuation of deformation mapping of the dam area and its vicinity would further improve the insights in ongoing deformations. The current stack of imagery was just sufficient for a reliable result, but more would optimise results significantly.
The outlook for the ADB is to use these maps as part of their primary process in monitoring & evaluation of a bank project. With the case of Hydropower in Viet Nam several projects are implemented and more are to come. Land use maps are a valuable tool to monitor the land use changes. Deformation monitoring can be done continuously and reported with alert mode and as a twice yearly status update.