International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Publications Copernicus
Download
Citation
Volume XLII-4/W19
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-4/W19, 229–232, 2019
https://doi.org/10.5194/isprs-archives-XLII-4-W19-229-2019
© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-4/W19, 229–232, 2019
https://doi.org/10.5194/isprs-archives-XLII-4-W19-229-2019
© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

  23 Dec 2019

23 Dec 2019

RADIOMETRIC CALIBRATION OF LANDSAT-8 OLI AND SENTINEL-2 MSI IMAGES FOR WATER QUALITY MODELLING OF LAGUNA LAKE, PHILIPPINES

E. V. Gubatanga Jr1, A. C. Blanco1, C. H. Lin2, and B. Y. Lin2 E. V. Gubatanga Jr et al.
  • 1Training Center for Applied Geodesy and Photogrammetry, University of the Philippines, Diliman, Philippines
  • 2Department of Geomatics, National Cheng Kung University, Tainan City, Taiwan

Keywords: canonical correlation analysis, pseudo-invariant features, multivariate alteration detection

Abstract. Regular monitoring of water quality in Laguna Lake is important for it supports aquaculture and provides water supply for Metro Manila. Remote sensing makes it possible to monitor the spectral conditions of the lake on a regular time interval and with complete coverage except for the areas with cloud and shadow cover. Along with in-situ water quality measurements, bio-optical models can be developed to determine the relationship between spectral and bio-optical properties of the lake water and consequently enables the estimation of water quality through remote sensing. However, radiometric calibration is needed to minimize the effects of the changing atmospheric conditions over time and to account for the difference in sensors (e.g., Landsat-8 OLI, Sentinel-2 MSI) used for water quality assessment. Canonical correlation analysis is used to detect pseudo-invariant features (PIFs), which are ground objects that do not dramatically vary in spectral properties over time. Road surface and other large man-made infrastructures are the commonly detected PIFs. These PIFs are used to compute for the parameters used to normalize reflectance values of remotely-sensed images obtained on different dates and using different sensors. The normalization resulted to a reduction of difference in reflectance values between the reference image and the adjusted image, though not marginal. This is due to the use of a linear equation to adjust the image, which limits the ability of the reflectance values of the image to fit to the values of the reference image.