Quantifying the Effects of Land Cover Change on Carbon Budgets in Romania

Remote sensing observations provide a cost-effective means of monitoring changes on Earth's surface over large areas. We used multitemporal observations derived from the Landsat global orthorectified datasets to map land cover changes in a wall-to-wall fashion in Romania over the period circa-1990 and circa-2000. A combination of the multispectral transforms of brightness, greenness, wetness and change in brightness greenness and wetness derived from atmospherically corrected data served as input in a supervised neural network classifier to map land cover changes. The resulting estimates of rates of forest clearing were then used in conjunction with data on forest biomass, and the fates of harvested forest biomass, in a terrestrial carbon bookkeeping model to estimate the emissions of carbon dioxide from land cover change and its effect on the terrestrial carbon cycle of the region. Results show a slow rate of forest harvest of 2.6% over the circa-10 year period. A preliminary independent validation of the remote sensing results shows an overall accuracy of 88.7 %. The terrestrial carbon model estimates the past, the present and the future terrestrial carbon budget of the country. Over the last 200 years Romania has been a large carbon sink. However, if forest harvest rates continue at current rates, about the year 2100 the country will change from a carbon sink to a carbon source. The change is mainly a result of the accumulation and decomposition over time of carbon stored in the long-term carbon pools like timber products.