Development of country mosaic using IRS-WiFS data

Abstract

The remote sensing satellites revolve around the earth in the polar orbits and they acquire the data over a predefined swath which is defined by its field of view (FOV) and orbit. Each of these satellites is equipped with imaging sensors. Which are broadly categorized based on their spatial resolution, spectral resolution, radiometric resolution and temporal resolution. Indian Remote Sensing Satellites-1C/1D (IRS - 1C/1D) are equipped with a Wide Field Sensor (WiFS), which has a moderate spatial resolution of 188 meters and temporal resolution of 5 days. The remote sensing operators provide the data by defining a scene in terms of Path-Row, which cuts across the swath over a particular path. The IRS-WiFS scene covers about 810km × 810km on the ground. If the area of interest is not covered in one scene with a particular spatial resolution, then the multiple scenes are needed for specific studies over an area larger than a scene. In order to study the vast countries, like India, spreading largely from west to east by about 3000km and north to south by about 2900km, several scenes spreading over different paths are required. Since they are acquired over different paths on different dates the radiome-try of these scenes will not be the same. Sometimes the same scene acquired temporally may also need to be processed if it has partial cloud cover. Stitching (joining) of large number of such scenes alone will not be sufficient to create a seamless coverage of entire geographical area of interest due to the geometric and radiometric distortions that are present inherently in each of these scenes. Hence the entire scenes required have to be essentially corrected for geometrical distortions and radiometric balance across the area of interest by the process called mosaicing. In this case study a country mosaic like India is taken up for creating a seamless Mosaic using IRS - 1C/1D WiFS sensor data. Since the country is spreading largely from west to east, all the images required for mosaic need to be projected. The LCC projection is best suitable for making proper measurements of such geographical extents. All the adjacent images need to be geometrically corrected to the sub pixel accuracy to avoid the discontinuities or blurring in the features at the overlapping regions. By identifying the matched Ground Control Points (GCPs) across the pairs of images, a polynomial transformation model is established to rectify the geometrical distortions. Radiometric normalization is obtained by achieving normalization of the adjacent scenes over the required geographical extent. Two adjacent images are normalized by histogram matching, either by considering the histograms of the entire image or by considering the histograms of the overlap areas only. Generally these automatic techniques used for this do not result in the best normalization, particularly when the large area has to be mosaiced. Hence, suitable interactive techniques are adopted in this case study. The geographical extent of India is spread between the paths 90 and 115 covering 50 scenes. Firstly, path wise mosaics are created by mosaicing all the scenes in a path. Secondly, the adjacent paths are mosaiced to cover the entire extent of Indian region. In order to achieve the desired scales of photographic product, the source image spatial resolution of 188m is modified to 200 m pixels by resampling the image data. To obtain the precise statistics, authenticated boundary information has been overlaid and used for clipping the final mosaic of India. This paper describes various techniques adopted in creating the mosaic of India using IRS - 1C/1D WiFS sensor data.