Algorithm for Automatic Extraction of Waterbodies using AWiFS data of Resourcesat-1

要旨

Temporal fluctuations in water resources occur in different seasons of the year, with great variations in water spread area of water bodies during monsoon to summer. Since due to natural phenomena like rainfall, floods, drought brings changes in availability water resources, it requires continuous monitoring of waterbodies for the assessment of water resources. The traditional classification algorithms viz. supervised, unsupervised, band thresholding, use of NDWI index require decision based approach for each scene, enough subjectivity is involved in applying the classification parameters and it would also be dependant on image processing softwares. Country like India which is based on the agriculture requires temporal (weekly/ fortnight) information on the availability of water resources for effective water resources planning and management at various levels. Hence use of above methods would be difficult to derive quick information. Satellite data acquired through Advanced Wide Field Sensor (AWiFS) on board Resourcesat 1 (IRS P6) which provide information at 56m spatial resolution with 5-day repetivity was used for automatic extraction of waterbodies. In the present paper, an algorithm developed for automatic extraction of waterbodies was presented which has been derived from the characterization of spectral properties of water bodies through pixel wise analysis of several sample water bodies (various types of water bodies viz. medium, major reservoirs, tanks of various turbidity levels, shapes and varying depths) from AWIFS data of India. This algorithm enabled the quick automatic extraction of waterbodies from the satellite data acquired by AWiFS sensor. Results are validated by comparing the water spread area estimated from on screen visual interpretation and area derived from the algorithm.