The new index for estimating the size of hail

stima dimensione grandine

Hail events cause significant damage to agriculture, buildings and infrastructure, with profound implications for insurance markets. Hypermeteo, therefore, is addressing the following operational and pragmatic issues:

  • Identifying the areas affected by hail;
  • Accurately estimating the size of the hailstone.

Hypermeteo has developed a specific index for estimating hail size and tested it in Italy. It is based on an innovative approach, made possible by the technologies and types of data available today, which integrates observed and modelled variables, allowing the algorithm to recognise the relevance of each piece of data. The model uses as measured data the Constant Altitude Plan Position Indicators (CAPPI) derived from the mosaic produced by the National Civil Protection Department’s weather radar network.

To train the model, hail reports in Italy during the period August 2022 – August 2023, collected by the Italian association PRETEMP, were used.

These reports were validated and processed to reduce any errors in the temporal or geographical location of the phenomenon. The resulting database consists of approximately 3000 hail events, to which an equal number of non-hail events were added in order to obtain a dataset consisting of the greatest possible variety of case histories (approximately 6000 in total) and thus sufficient for training the algorithm. These events were combined with radar and modelling variables and a machine learning method for hailstone diameter recognition. The resulting model was then validated and tested.

Figure 1 shows the comparison between the observed diameters and those predicted during testing.

Figure 1: Evolution of the hail diameter predicted by the method proposed by Hypermeteo (x-axis) versus the actual diameter (y-axis) considering the total of 6000 hail and non-hail events.

The results obtained allow us to conclude that the proposed index effectively identifies hail with diameters between 0 and 5 cm.

Case Study

To illustrate the functioning of the index, a case study was selected to show the range of hail sizes in Italy during the documented events. The map below (Figure 2) shows a hail event affecting areas of Abruzzo and Molise on 4 July 2023, at about 13:30 UTC.

The data obtained show that the model accurately detects the hail size and the geographical area actually affected by the hail events.

Figure 2:  Hail size estimation using the method proposed by Hypermeteo for 4 July 2023, at 13:30 UTC. PRETEMPS documented hail with a diameter of 2 cm in L’Aquila, Abruzzo region, and 1 cm in Capracotta, Molise region, at 13:20 UTC and 13:25 UTC respectively.

Request more information

It might interest you

Hypermeteo’s customers