Interannual landsat-MSS reflectance variation in an urbanized temperate zone

Abstract

A chronological time series of 40 Landsat-MSS images over southeastern Canada for the period of 1972–1984 has been analyzed to monitor short-term (seasonal) and long-term variations in land-surface parameters which are either of anthropogenic origin or due to climate-related variability. The data base, generated on the Canadian Landsat Production System, was systematically processed for radiometric correction and atmospheric correction based on clear lakes as standard reflectors and accounting for the background contamination effect. The reflectance values of selected gravel pit sites, which we demonstrate to be stable reflectors, are analyzed as a function of time. The significant and regular variations observed permit the determination of correction coefficients for inter-satellite calibration between the sensors of Landsats 1–5. The resulting coherent time series of corrected reflectances shows a mean seasonal variation for the vegetation phenology (between May and October) which can be modeled by a regression function of the form ( t - t 0)sin[( t - t 0)2 π/ T] for the two near-IR MSS bands. The ground reflectances of the four bands are summed over and combined to compute the long-term “MSS albedo” and normalized vegetation index (NVI) variations. These parameters are analyzed with regard to meteorological variations. The results show that, for the Montreal study area, the yearly mean albedo underwent a significant decreasing trend (− 0.05) since 1972 due to processes of urbanization, deforestation, and decreasing farmland. The effects of these respective land-surface alterations are discussed. For the whole data base, the year-to-year NVI fluctuations correlate with the cumulative amount of summer precipitation which in turn is related to the mean summer evapotranspiration and global solar radiation.