Abstract. Hazard mapping is carried out in Italy according to the AINEVA guidelines, which require (i) data driven avalanche dynamic modelling to assess end mark and pressure, and (ii) assessment of maximum yearly three-day snow depth increase h₇₂ for 30 to 300 years return period. When no historical avalanche data are present, model tuning and data based assessment of avalanche return periods are hardly feasible. Also when (very) short series of h₇₂ are available, station based quantile estimation for such high return periods is very uncertain, and regionally based approaches can be used. We apply an index value approach for the case study avalanche of Rigopiano, where a 105 m³ snow mass hit the Rigopiano Hotel killing 29 persons on January 18^th 2017. This area is poorly monitored avalanche wise, and displays short series (max 14 years) of snow depth measurements, no historical avalanche maps are available on the avalanche track, and no hazard maps have been developed hitherto. First, we tune the recently developed Poly-Aval dynamic avalanche model (1D/q2D) against the 18^th January event data (release zone, release depth, end mark) from different sources. We then use snow data from 7 snow stations in Abruzzo (75 equivalent years of data) to tune a regionally valid distribution of h₇₂. We then calculate the 30-years, 100-years, and 300-years runout zone and flow pressures, including confidence limits. We demonstrate that (i) properly tuned 1D/quasi2D models can be used for avalanche modeling even within poorly monitored area as here, and (ii) the use of regional analysis allows hazard mapping for large return periods, reducing greatly the uncertainty against canonical, single site analysis. Our approach is usable in poorly monitored regions like Abruzzo here, and we suggest that (i) avalanche hazard mapping needs to be pursued with regional approaches for h₇₂, and (ii) confidence limits need to be provided for the proposed zoning.