ABSTRACT
Previous studies have been able to successfully detect changes in gently-sloping forested areas with low-diversity and homogeneous vegetation cover, using medium-resolution satellite data such as landsat. The aim of the present study is to examine the capacity of multi-temporal landsat data to identify changes in forested areas with mixed vegetation and generally located on steep slopes or non-uniform topography. landsat thematic mapper (TM)and landsat enhanced thematic mapper plus (ETM+) data for the years 1987-2000 was used to detect changes within a 19,500 ha forested area in the Western Black sea region of Turkey. The data comply with the forest cover type maps previously created for forest management plans of the research area. The methods used to detect changes were: postclassification comparison, image differencing, image rationing and NDVI (Normalized Difference Vegetation Index) differencing methods. Following the supervised classification process, error matrices were used to evaluate the accuracy of classified images obtained. The overall accuracy has been calculated as 87.59% for 1987 image and as 91.81% for 2000 image. General kappa statistics have been calculated as 0.8543 and 0.9038 for 1987 and 2000, respectively. The changes identified via the post-classification comparison method were compared with other change detetion methods. Maximum coherence was found to be 74.95% at 4/3 band rate. The NDVI difference and 3rd band difference methods achieved the same coherence with slight variations. The results suggest that landsat satellite data accurately conveys the temporal changes which occur on steeply-sloping forested areas with a mixed structure, providing a limited amount of detail but with a high level of accuracy. Moreover, it has been decided that the post-classification comparison method can meet the needs of forestry activities better than other methods as it provides information about the direction of these changes.
ABSTRACT
Pollen, charcoal, and C14 analyses were performed on a sediment core obtained from Lake Tapera (Amapá) to provide the palaeoenvironmental history of this part of Amazonia. A multivariate analysis technique, Detrended Correspondence Analysis, was applied to the pollen data to improve visualization of sample distribution and similarity. A sedimentary hiatus lasting 5,500 years was identified in the Lake Tapera. Because the timing of the hiatus overlapped with the highest Holocene sea-level, which would have increased the local water table preventing the lake from drying out, it is clear that sea-level was not important in maintaining the lake level. Lake Tapera probably depended on riverine flood waters, and the sedimentary gap was caused by reduced Amazon River discharge, due to an extremely dry period in the Andes (8,000-5,000 years BP), when precipitation levels markedly decreased. The lack of Andean pollen (river transported) in the record after this event supports this interpretation. The pollen analysis shows that when sedimentation resumed in 1,620 cal. years BP, the vegetation around the lake was changed from forest into savanna. This record demonstrates the need to improve our understanding of climate changes and their associated impacts on vegetation dynamics.
Análises de pólen, carvões e datações C foram conduzidas em um testemunho coletado no lago Tapera (Amapá) com o objetivo de interpretar a história paleoambiental desta parte da Amazônia. Uma das técnicas de análises multivariadas, Análise de Correspondência Destendenciada (DCA), foi utilizada a fim de melhor visualizar a distribuição e similaridade das amostras. Foi identificado um hiato sedimentar com duração de 5.500 anos no lago Tapera. Como o hiato ocorreu simultaneamente ao nível do mar mais alto do Holoceno, o que deveria ter aumentado o lençol freático, impedindo assim o lago de secar, é evidente que variações do nível do mar não foram importantes na manutenção do nível do lago. O lago Tapera provavelmente dependia de água das cheias fluviais,e o hiato sedimentar foi causado por uma redução da descarga do Amazonas, devido a um período extremamente seco nos Andes (8.000-5.000 anos cal. AP), quando os níveis de precipitação diminuíram drasticamente. A ausência de pólen transportado dos Andes pelo rio Amazonas no registro sedimentar, após este evento, apóia esta interpretação. A análise palinológica demonstra que quando a sedimentação reiniciou, em 1.620 anos cal. AP, a vegetação em torno do lago havia mudado de floresta para savana. Este registro demonstra a necessidade de melhorar nosso entendimento sobre a extensão e os impactos das mudanças climáticas sobre a evolução da vegetação.