Studentinformatie

MSc thesis subject: Automatic Identification of Pseudo-Invariant Features (PIF) in Landsat time series

Atmospheric correction is an important step to prepare (optical) satellite images for time series analysis. Modern scientific sensors have dedicated spectral bands to identify clouds and estimate atmospheric parameters (e.g. Aerosol Optical Thickness, Water column), so that physical approaches can be used to correct the images. Less advanced sensors mostly lack these bands and atmospheric correction must be achieved in another way. The empirical line method is one such way. It takes advantage of ground targets that are quasi stable in reflectance, so called pseudo-invariant features. However, the identification of high quality features can be laborious. Time series approaches could overcome this problem by identifying targets that have a stable history.

Through the Dutch Satellite Portal (http://www.spaceoffice.nl/nl/Satellietdataportaal/) the GRS group has access to UK-DMC2, Deimos-1 and SPOT 6/7 images regularly acquired over the whole of the Netherlands. The Climate Data Record (CDR) products within the Landsat archive provide long time series of atmospherically corrected images. The principal aim of this study is to correct the imagery from the Dutch Satellite Portal in such a way that they have comparable accuracy in terms of surface reflectance as the Landsat record.

Objectives

  • Conduct literature review on PIF
  • Identify properties required for PIFs and translate them into metrics that can be derived from Landsat time series
  • Apply empirical line calibration on a set of UK-DMC2/Deimos-1 images
    Validate the approach with simultaneous observations of Landsat and UK-DMC2/Deimos-1
  • Optional: apply methodology to SPOT 6/7 images

Literature

  • Smith, G. M., & Milton, E. J. (1999). The use of the empirical line method to calibrate remotely sensed data to reflectance. International Journal of Remote Sensing, 20(13), 2653–2662.

Requirements

  • Scripting skills (e.g. R, Python, MatLab) are a preference

Theme(s): Sensing & measuring