IRENE: A Convolutional GRU Ensemble Model for Radar Precipitation Nowcasting over Italy
Abstract
We present IRENE (Italian Radar Ensemble Nowcasting Experiment), a deep learning model for probabilistic short-range precipitation nowcasting over the Italian domain at \SI{1}{km} spatial and 5 min temporal resolution. IRENE adopts an encoder--forecaster architecture built on multi-scale Convolutional Gated Recurrent Units (ConvGRUs), trained on the national radar composite produced by the Italian Civil Protection Department (DPC). An importance-sampling scheme focuses training on precipitation-relevant events, while the almost-fair Continuous Ranked Probability Score (afCRPS) is adopted as the primary probabilistic loss function. Two additional training configurations are proposed: an adversarial (GAN) variant, IRENE-GAN, designed to improve the spatial sharpness of the generated forecasts, and a spectrally constrained variant, IRENE-GAN-RAPSD, in which the adversarial objective is complemented by an explicit penalty on the radially averaged power spectral density. The three configurations are evaluated against the stochastic extrapolation method STEPS and the pre-trained deep learning model DGMR. All IRENE configurations attain a lower Continuous Ranked Probability Score than both benchmarks at every lead time and rank histograms closer to uniformity, indicating better probabilistic skill and ensemble calibration. In terms of ensemble-mean mean absolute error the advantage is confined to the first 90 min, beyond which the strongly damped DGMR fields and, to a lesser extent, STEPS become competitive. Spectral analysis shows that the adversarial training removes the progressive loss of small-scale variance exhibited by IRENE, at the cost of an excess of fine-scale power at long lead times that the spectral penalty only partially controls.
Related
- MW-Nowcast: Six-hour ensemble nowcasting of extreme precipitation
- Precipitation nowcasting model based on multi-source data fusion and efficient additive attention
- Explainable Deep Learning for Probabilistic Nowcasting of Radar Reflectivity in Tornadic Storms
- Diffusion-Based Refinement for Kilometer-Scale Probabilistic Precipitation Nowcasting
- Wet-Biased Data Pipelines Mislead the Training and Evaluation of Generative Precipitation Nowcasting
BibTeX
@misc{camilletti2026irene,
title = {{IRENE: A Convolutional GRU Ensemble Model for Radar Precipitation Nowcasting over Italy}},
author = {Alessandro Camilletti and Gabriele Franch and Elena Tomasi and Marco Cristoforetti},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.17175},
eprint = {2609.17175},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.17175},
note = {CC BY-SA 4.0},
}
RIS
TY - UNPB
TI - IRENE: A Convolutional GRU Ensemble Model for Radar Precipitation Nowcasting over Italy
AU - Alessandro Camilletti
AU - Gabriele Franch
AU - Elena Tomasi
AU - Marco Cristoforetti
PY - 2026
DA - 2026/09/15
PB - arXiv
DO - 10.48550/arxiv.2609.17175
UR - https://arxiv.org/abs/2609.17175
AB - We present IRENE (Italian Radar Ensemble Nowcasting Experiment), a deep learning model for probabilistic short-range precipitation nowcasting over the Italian domain at \SI{1}{km} spatial and 5 min temporal resolution. IRENE adopts an encoder--forecaster architecture built on multi-scale Convolutional Gated Recurrent Units (ConvGRUs), trained on the national radar composite produced by the Italian Civil Protection Department (DPC). An importance-sampling scheme focuses training on precipitation-relevant events, while the almost-fair Continuous Ranked Probability Score (afCRPS) is adopted as the primary probabilistic loss function. Two additional training configurations are proposed: an adversarial (GAN) variant, IRENE-GAN, designed to improve the spatial sharpness of the generated forecasts, and a spectrally constrained variant, IRENE-GAN-RAPSD, in which the adversarial objective is complemented by an explicit penalty on the radially averaged power spectral density. The three configurations are evaluated against the stochastic extrapolation method STEPS and the pre-trained deep learning model DGMR. All IRENE configurations attain a lower Continuous Ranked Probability Score than both benchmarks at every lead time and rank histograms closer to uniformity, indicating better probabilistic skill and ensemble calibration. In terms of ensemble-mean mean absolute error the advantage is confined to the first 90 min, beyond which the strongly damped DGMR fields and, to a lesser extent, STEPS become competitive. Spectral analysis shows that the adversarial training removes the progressive loss of small-scale variance exhibited by IRENE, at the cost of an excess of fine-scale power at long lead times that the spectral penalty only partially controls.
ER -