Detection of pollution-induced forest decline in the Kola Peninsula using remote sensing and mathematical modelling

Forests on the Kola Peninsula in Northern Russia grow close to the northern tree line. They are subjected to both natural and anthropogenic stress factors. The Cu-Ni smelter 'Severonikel' (Lat. 67 deg 55'N; Long. 32 deg 57'E) near Monchegorsk is one of the two major sources of sulphur dioxide and heavy metals emissions on the Kola Peninsula. These emissions have caused significant deterioration of the surrounding vegetation. The thesis demonstrates how methods of Remote sensing, ground survey and mathematical modelling can be integrated for monitoring of the smelter's environmental impact on the surrounding vegetation: ground truth data are used for calibration of remote-sensed data, which further serve to verify mathematical models. The study aims were: * to estimate the scale of airborne sulphur pollution from the smelting industry on the Kola Peninsula and its effect on vegetation; * to assess spatial extent of the forest decline in the 'Severonikel' smelter impact zone; * to display dynamics of the forest damage area in spatial and temporal perspective; * to validate the Gaussian plume sector model and the IIASA forest impact model as components of the analysis of forest damage. The Regional Acidification Information and Simulation model (RAINS) was applied to calculate sulphur deposition and loads in Fennoscandia in order to assess the contribution of the Kola sources to the deposition pattern in the region. The percentage of the ecosystems where the critical load had been exceeded was calculated. For more detailed assessments, calculations based on local and meso-scale models were made. Landsat-MSS summer images from 1978, 1986 and 1992 and a Landsat -TM image from 1996 were used for change-detection analyses. The methods applied were bandwise histogram matching and subsequent differencing. An unsupervised classification of land-cover was made using the 1996 Landsat-TM image. In situ observations of vegetation type and degradation levels on permanent field plots were used for labelling the classes. Multispectral changes observed between 1978 and 1992 were used to evaluate the relevance of the Gaussian plume model as a component for assessment of forest decline. The IIASA model for accumulated impact in forests under long-term exposure to airborne sulphur was applied for the period of 1960-1996. The input data used were plant sensitivity parameters and SO2 ambient concentrations predicted at a previous stage. The model was validated by ground truth data and the results of the 1996 classification. Effects of topography and episodes of high concentrations were additionally investigated by a 3D numerical modelling Results: * Comparative analysis by the RAINS model showed a significant, but local impact of the Kola sources on adjacent parts of the Nordic countries. * Remote sensing has revealed vegetation decline in large areas around the 'Severonikel'. The damaged area expands more north than south of the smelter due to a dominance of southern winds during the vegetation period, and a sheltering role of topography as revealed by 3D modelling. * Multispectral changes detected between 1978 and 1992 were found to have a statistically significant correspondence to the modelled long-term SO2 concentration levels in ambient air. This validates the Gaussian plume model. * The impact model was validated by in situ field data and by the classified 1996 scene. * Both Remote sensing analysis and the impact model show a rapid expansion of the degraded area during 1978-1992 and its stabilisation since that time which is consistent with the reduced emissions. * An exceptional expansion of forest damage south of the smelter in 1992 is likely to be linked to an episode of high concentration prior to the image acquisition or extreme climatic conditions.