GIS and Remote Sensing

Volume XVIII |

How much open water do waterbirds have in the Banat Plain? The first permanent inland water bodies inventory at 10-m resolution using Sentinel-2 imagery at regional – scale

Abstract: In the Banat Plain were recorded more than half of the number of bird species in Romania. There are eleven Nature 2000 Bird Protection Sites (SPAs) and in all of them have been registered aquatic bird species. The water surface areas were greatly reduced since the starting of the hydro-technical works 300 years ago. Nowadays, the anthropic pressure associated with the more frequent drought periods continues to threaten the water bodies, on whose existence the numerous aquatic birds depend. For the mapping and evaluation of open water surfaces in the Banat Plain, in the context of a lack of precise, consistent maps and data, the first water bodies inventory at 10-m resolution was made based on Sentinel-2 multispectral satellite images. The heterogeneous and fragmented landscape, with a great diversity of aquatic units, conducted us to develop an object-oriented approach, which allowed a multi-scale classification process. Both spectral bands and normalized differentiation water indices were used in the rule-based classification algorithm for water detection. The accuracy assessment indicated a very good overall accuracy of 96%, with a Kappa coefficient of 0.91. Also in our dataset, the small water bodies were mapped more accurately than in the other six water bodies datasets at global or European scale we compared with. The results drawn from both qualitative and quantitative assessments indicated that the water dataset developed in this study could be used as an inventory and a reference map for the permanent open water areas from the Banat Plain. It may represent the starting point for better aquatic management and elaboration of ecological strategies to support bird necessities and to counteract the increasing vulnerability and environmental threatening of waterbird species.

Volume XVIII |

New application of fuzzy logic algorithm in GIS for land classification

Abstract: Population growth and disorganization of urban planning have led to unsuitable city development in the center of Iran. Suitable region recognition for urban land develop-ment is an important step towards future planning. In the present study, fuzzy logic algorithms (OR, And, Sum, Prod-uct and Gamma) were used within GIS in order to identify valuable land for appropriate residential development. Moreover, effective factors of urban land development (elevation, slope, aspect, geology, land use, drainage net-work, main and bypass roads, distribution of urban and rural areas, and fault line layer) were examined on fuzzy analyses to find the most effective ones. The results showed that, by considering the regional priorities and constraints, the best operator was Gamma, with a power of 0.9. According to this, 74 percent of the total regions are located between less and the least valuable lands and the remaining surfaces (i.e. 26% of the region) were classified from valuable to the most valuable lands. The sensitivity measurement of the layers used in the study showed that fault and distribution of urban and rural layers were the most and the least effective layers on region recognition (i.e. by 25.83% and 3.29%, respectively).