Executive Summary – Phase 1 / 2025
Phase 1 (year 2025) had as its main objective the substantiation of the database and the methods necessary for a multi-temporal inventory of landslides.
Achievements from stage 1 / 2025
Multi-sensor data acquisition and pre-processing (A1.1)
The team downloaded, organized and pre-processed a very large volume of satellite images: Landsat 5, 7, 8/9 (1984–2024) for Romania and the Republic of Moldova; Sentinel-2 (optical) and Sentinel-1 (radar) (2014–2024).
Landslide inventory from historical data (A1.2)
Landslides were identified by assisted visual interpretation and data fusion: multi-temporal satellite imagery (Landsat, Sentinel-2); digital terrain models obtained from LiDAR; high-resolution aerial imagery. By combining these sources, several thousand landslides were mapped: approximately 1,700+ landslides in the Republic of Moldova; approximately 1,000 landslides in the Moldovan Plateau in Romania; over 1,200 landslides in the Subcarpathians and Curvature Carpathians.
Adding the temporal dimension – detection of trigger momentsBased on the series of processed satellite images, an algorithm has been developed that analyzes, for each landslide, the evolution over time of a complex set of vegetation indices, textures and landscape metrics. Sudden and significant changes in these values are quantified by a composite sensitivity score. The calendar dates with the maximum score are interpreted as the most likely times of the landslide triggering or reactivating. Thus, the classical (static) inventory is transformed into a multi-temporal inventory, which captures the dynamics in time of the phenomenon.
Nationwide susceptibility analysis for the Republic of MoldovaBased on an extensive landslide inventory and environmental factors (slope, lithology, land use), a national landslide susceptibility map was developed for the Republic of Moldova, using the MaxEnt algorithm (machine learning).
Data Exchange and Management Protocol (A1.3, A1.4)
A Data Sharing Protocol has been developed and agreed between the University of Bucharest (UB) and the State University of Moldova (USM), which regulates: the use of pre-existing data generated in the project; co-ownership of the resulting products; privacy and compliance with GDPR; the use of open formats and standardized web services.
Dissemination of results
- 3 presentations at international conferences (GeoEarth – Bucharest, MedGU – Athens)
- 2 scientific articles submitted for evaluation in top international journals (Landslides; ISPRS Journal of Photogrammetry and Remote Sensing).