About me:
I am a Lecturer/Researcher at the faculty of Architecture and Engineering at Epoka University, in Tirana- Albania. At the same time, I am a visiting lecturer at the Agricultural University of Tirana.
I am an architect by education, earning the PhD degree in Landscape Architecture and Planning from Istanbul Technical University in 2018.
My research interest on wildfires initiated during my engagement as a researcher in the ERASMUS+ project “Knowledge FOr Resilient society, K-FORCE” (2017).
During this period, I have worked on understanding the wildfire ignition and spreading dynamics and the driving factors at landscape scale. Then, I have been exploring to utilize open access geospatial data in GIS-based modelling to classify the vegetated surfaces by their wildfire ignition and spreading capacities. The developing regions which are lacking local level data, are the main study area of my research.
I have been working on GIS-based on the Wildfire Ignition Probability/ Wildfire Spreading Capacity Index (WIPI/ WSCI). Initially, the method defines criteria that have proven relation with either the wildfire occurrence or propagation. The number of criteria varies according to the available data and the specifics of the regional context. Currently, the model integrates 16 criteria regarding the anthropogenic, hydrometeorological, fuel, and geophysical features of the study area.
Our current study includes the updates that we push forward as improvements of WIPI/ WSCI, applied in Romania’s case. At this stage, the workflow consists of seven sequential stages. Besides the inventory procedure, the first stage includes defining the vegetated surfaces within the study area and the reference points that spatially represent the vegetation surfaces. The reference points serve as data collecting pivots loaded with all 16 criteria’ unique values, as shown in Fig. 1. The unequal range of inventory values necessitates a normalizing procedure before indexing calculations. This stage equalizes the range of inventory values of each criterion into a gradient between 0 and 1.
The third stage consists of subgrouping the criteria into two sets according to their relationship with either wildfire ignition or spreading (see Fig.1, third stage). Moreover, a relevancy indicator is given to each criterion according to their direct or indirect relationship with wild-fire regimes. The first three methodical stages are borrowed from our previous studies.
In the fourth stage, we propose ROC/AUC analysis (via SPSS software) as a weighting method among criteria, besides the analytic hierarchy processing (AHP) pairwise comparison method. This relies on the specific characteristics of the study area and remotely sensed historical data on fire regimes. All steps of the model are automated via Model Designer in QGIS to provide a customizable tool that is easily reproducible to other study areas.
The model is still under improvement, and welcomes any critique, suggestion, or collaboration.
For further details, please refer to our current preprint article.
