The paper describes the development of an Early Warning (EW) prototypal application, working within the GIS environment, able to support the integrity management of oil & gas pipelines. The system is able of forecasting possible hydrogeological hazards related to natural events (flooding and slope instability), with focus on possible landslides induced or reactivated by local precipitation occurring in the area.
The Early Warning system has been applied on a case study area (Val d'Agri region in the South Italy). It defines, using all available geological, climatic, geomorphological and geotechnical data, a hazard classification of the pipeline routing (baseline), as well as critical thresholds of rainfall that can activate a slope instability. The hourly precipitation forecasts issued by a dedicated proprietary model (e-kmf®: Kassandra Meteo Forecast, KMF), with a horizontal resolution of 1.25 km on the downscaled area, have been integrated in the EW system, together with the precipitation records of the previous days. These predicted amounts of rain are compared to the thresholds, the exceeding of which generates a warning alarm in any segment of the pipeline, useful for the asset integrity management of the infrastructure.
The implemented EW methodology is based on highlighting: i) a multi-criteria analysis (MCA) for the classification of the areas in accordance with objective and repeatable criteria; ii) a multiscale approach, adopted for the early warning requirements needed to enable its activation; iii) machine learning methods applied to performance evaluation for calibrating and optimizing the system, updating thresholds and baseline. After that, a fully prototype system is working on the case study by including the local data available in the area, a feasibility assessment and the definition of the operative plan. The EW prototype has been developed and tested including the automation of the hourly weather forecast data (covering the next 60 hours) and of the warning function with periodic/alert output. Data analysis for temperature, wind and precipitation have been assessed over a period of two years and a good agreement between forecasts and several measurement points in situ has been achieved. The engineered EW system is planned in the future deployment, generalizing the models for the application to other assets.
