LIS for Port Based Development

Abstract :
A deepwater container transshipment port is being planned at Vizhinjam, which is 20 km south of Thiruvananthapuram, the capital city of Kerala in India. The geographical location of the proposed port project is Latitude 8o 22’ (North) and Longitude 76o 59’(East) on the West coast of India. The project is to establish a state-of-the-art transshipment hub to cater to all types of container vessels including the Super Post Panamax vessels & mega container carriers. Other related infrastructure facilities includes container yard, container freight station, berths / container cranes and yard equipment, rail & road linkages etc.
In connection with the project, development of port based special economic zones, free trade warehousing zones and logistic corridors are also planned. These are proposed within a radius of 20 km from the project location and where thinly populated large areas of land holdings are available. Identification of such large areas is a difficult task, especially in a densely populated state like Kerala. The entire stretch of 580km long coastline of Kerala State is thickly populated, where majority are fisherfolk. For this purpose, the powerful tool Land Information System using a common platform, called Geographical Information System is proposed to be deployed. None other than a powerful tool like LIS can solve this complex problem, economically and with minimum time and constraints.
Land information system is a subsidiary of the GIS and concerns entirely to land data. It includes information such as size, shape, location, legal description, topography, flood plains, water resources, easements, and zoning requirements for each piece of land in the system.
The detailed cadastral maps in a scale of 1:5000 to 1:10000 are not available for the whole area of interest. But detailed topographic maps of the whole area of interest are available with a map scale of 1:50,000 and above. As these maps will not cater to the requirement, it is decided to use satellite images with low resolution to get the required information.
The three major data sources for obtaining the required information are recommended are
1) High resolution satellite images (IKONOS- or QUICKBIRD-images)
2) Low resolution satellite images (IRS – images)
3) Cadastral data from the Revenue Department
As a first step, a rough identification of the areas of interest are demarcated with the help of IRS images(Indian Remote sensing satellite images, which have a low resolution of 6.5m) High resolution satellite images, provided by the satellites IKONOS (1m panchromatic ground resolution) or QUICKBIRD (0.64 m panchromatic ground resolution) are the best source to study the preliminarily identified areas. Nevertheless, it is necessary to process the satellite data in order to derive the necessary information.
The whole process is then planned accordingly. Firstly, the images have to be geo-referenced, which means in brief to place the image on the respective location in the geographical latitude/longitude grid. Other map data could only then be overlaid and compared with the images. The geo-referencing requires specific software, the so-called “Raster GIS” which enable the processing of the ‘raster data’ as provided in satellite images.
Since the image contains various information, based on electromagnetic radiation in a much wider range than the human eye can recognize, the satellite data have to be classified. The purpose of the classification is to emphasize the objects of interest, in the present case, i.e., land pockets, which are appropriate for settlement. The classification of satellite images is based on the principle that every object on earth has a specific electromagnetic signature. This ‘signature’ has to be identified for the class “appropriate settlement area”. The classification is usually done in three working steps: first comes a so-called ‘unsupervised’ classification (the GIS software will automatically classify the different spectral ranges into a number of classes as defined by the user – at this stage, the meaning of the different spectral classes is not known), in a second stage, the meaning (i.e. the actual land cover for every class is to be identified in the field (during this “ground truthing”, sample areas for every class have to be validated in the field), and in a third stage, a ‘supervised’ classification has to be done (since the different spectral ranges can now be assigned to a land cover class, this will be fed in the GIS software for the final classification).
In reality, the supervised classification is a tedious process through trial and error to make an appropriate classification of the land cover. This is due to overlapping of spectral ranges of different types of vegetation (for instance a flooded paddy field has a similar electromagnetic signature like a fishpond or bushland might overlap with banana plantations, etc.). There are technical ways and means on how to optimize the result, in practice however a 60% to 70% percent match of the satellite classification with the reality is considered as ‘good’.
In this case, there are several factors which simplify the classification process: first, there is only one class of interest which is to identify and second, the source would be high resolution images which are providing a clear visibility of the area so that one can actually see what is the condition of the land (this is not possible with images of lower ground resolution).
The paper explains in detail the complexity of the problem and the methodology proposed to achieve the desired the desired objective. With all certainties we are confident about the level of success intended to be achieved through this effective tool.