-
Implicit Assimilation of Sparse In Situ Data for Dense & Global Storm Surge Forecasting
Authors:
Patrick Ebel,
Brandon Victor,
Peter Naylor,
Gabriele Meoni,
Federico Serva,
Rochelle Schneider
Abstract:
Hurricanes and coastal floods are among the most disastrous natural hazards. Both are intimately related to storm surges, as their causes and effects, respectively. However, the short-term forecasting of storm surges has proven challenging, especially when targeting previously unseen locations or sites without tidal gauges. Furthermore, recent work improved short and medium-term weather forecastin…
▽ More
Hurricanes and coastal floods are among the most disastrous natural hazards. Both are intimately related to storm surges, as their causes and effects, respectively. However, the short-term forecasting of storm surges has proven challenging, especially when targeting previously unseen locations or sites without tidal gauges. Furthermore, recent work improved short and medium-term weather forecasting but the handling of raw unassimilated data remains non-trivial. In this paper, we tackle both challenges and demonstrate that neural networks can implicitly assimilate sparse in situ tide gauge data with coarse ocean state reanalysis in order to forecast storm surges. We curate a global dataset to learn and validate the dense prediction of storm surges, building on preceding efforts. Other than prior work limited to known gauges, our approach extends to ungauged sites, paving the way for global storm surge forecasting.
△ Less
Submitted 5 April, 2024;
originally announced April 2024.
-
Super-resolved rainfall prediction with physics-aware deep learning
Authors:
S. Moran,
B. Demir,
F. Serva,
B. Le Saux
Abstract:
Rainfall prediction at the kilometre-scale up to a few hours in the future is key for planning and safety. But it is challenging given the complex influence of climate change on cloud processes and the limited skill of weather models at this scale. Following the set-up proposed by the \emph{weather4cast} challenge of NeurIPS, we build a two-step deep-learning solution for predicting rainfall occur…
▽ More
Rainfall prediction at the kilometre-scale up to a few hours in the future is key for planning and safety. But it is challenging given the complex influence of climate change on cloud processes and the limited skill of weather models at this scale. Following the set-up proposed by the \emph{weather4cast} challenge of NeurIPS, we build a two-step deep-learning solution for predicting rainfall occurrence at ground radar high spatial resolution starting from coarser resolution weather satellite images. Our approach is designed to predict future satellite images with a physics-aware ConvLSTM network, which is then converted into precipitation maps through a U-Net. We find that our two-step pipeline outperforms the baseline model and we quantify the benefits of including physical information. We find that local-scale rainfall predictions with good accuracy starting from satellite radiances can be obtained for up to 4 hours in the future.
△ Less
Submitted 24 October, 2023;
originally announced October 2023.
-
Deep-Learning-based Change Detection with Spaceborne Hyperspectral PRISMA data
Authors:
J. F. Amieva,
A. Austoni,
M. A. Brovelli,
L. Ansalone,
P. Naylor,
F. Serva,
B. Le Saux
Abstract:
Change detection (CD) methods have been applied to optical data for decades, while the use of hyperspectral data with a fine spectral resolution has been rarely explored. CD is applied in several sectors, such as environmental monitoring and disaster management. Thanks to the PRecursore IperSpettrale della Missione operativA (PRISMA), hyperspectral-from-space CD is now possible. In this work, we a…
▽ More
Change detection (CD) methods have been applied to optical data for decades, while the use of hyperspectral data with a fine spectral resolution has been rarely explored. CD is applied in several sectors, such as environmental monitoring and disaster management. Thanks to the PRecursore IperSpettrale della Missione operativA (PRISMA), hyperspectral-from-space CD is now possible. In this work, we apply standard and deep-learning (DL) CD methods to different targets, from natural to urban areas. We propose a pipeline starting from coregistration, followed by CD with a full-spectrum algorithm and by a DL network developed for optical data. We find that changes in vegetation and built environments are well captured. The spectral information is valuable to identify subtle changes and the DL methods are less affected by noise compared to the statistical method, but atmospheric effects and the lack of reliable ground truth represent a major challenge to hyperspectral CD.
△ Less
Submitted 20 October, 2023;
originally announced October 2023.
-
IceCloudNet: Cirrus and mixed-phase cloud prediction from SEVIRI input learned from sparse supervision
Authors:
Kai Jeggle,
Mikolaj Czerkawski,
Federico Serva,
Bertrand Le Saux,
David Neubauer,
Ulrike Lohmann
Abstract:
Clouds containing ice particles play a crucial role in the climate system. Yet they remain a source of great uncertainty in climate models and future climate projections. In this work, we create a new observational constraint of regime-dependent ice microphysical properties at the spatio-temporal coverage of geostationary satellite instruments and the quality of active satellite retrievals. We ach…
▽ More
Clouds containing ice particles play a crucial role in the climate system. Yet they remain a source of great uncertainty in climate models and future climate projections. In this work, we create a new observational constraint of regime-dependent ice microphysical properties at the spatio-temporal coverage of geostationary satellite instruments and the quality of active satellite retrievals. We achieve this by training a convolutional neural network on three years of SEVIRI and DARDAR data sets. This work will enable novel research to improve ice cloud process understanding and hence, reduce uncertainties in a changing climate and help assess geoengineering methods for cirrus clouds.
△ Less
Submitted 5 October, 2023;
originally announced October 2023.
-
THRawS: A Novel Dataset for Thermal Hotspots Detection in Raw Sentinel-2 Data
Authors:
Gabriele Meoni,
Roberto Del Prete,
Federico Serva,
Alix De Beussche,
Olivier Colin,
Nicolas Longépé
Abstract:
Nowadays, most of the datasets leveraging space-borne Earth Observation (EO) data are based on high-end levels products, which are ortho-rectified, coregistered, calibrated, and further processed to mitigate the impact of noise and distortions. Nevertheless, given the growing interest to apply Artificial Intelligence (AI) onboard satellites for time-critical applications, such as natural disaster…
▽ More
Nowadays, most of the datasets leveraging space-borne Earth Observation (EO) data are based on high-end levels products, which are ortho-rectified, coregistered, calibrated, and further processed to mitigate the impact of noise and distortions. Nevertheless, given the growing interest to apply Artificial Intelligence (AI) onboard satellites for time-critical applications, such as natural disaster response, providing raw satellite images could be useful to foster the research on energy-efficient pre-processing algorithms and AI models for onboard-satellite applications. In this framework, we present THRawS, the first dataset composed of Sentinel-2 (S-2) raw data containing warm temperature hotspots (wildfires and volcanic eruptions). To foster the realisation of robust AI architectures, the dataset gathers data from all over the globe. Furthermore, we designed a custom methodology to identify events in raw data starting from the corresponding Level-1C (L1C) products. Indeed, given the availability of state-of-the-art algorithms for thermal anomalies detection on the L1C tiles, we detect such events on these latter and we then re-project them on the corresponding raw images. Additionally, to deal with unprocessed data, we devise a lightweight coarse coregisteration and georeferencing strategy. The developed dataset is comprehensive of more than 100 samples containing wildfires, volcanic eruptions, and event-free volcanic areas to enable both warm-events detection and general classification applications. Finally, we compare performances between the proposed coarse spatial coregistration technique and the SuperGlue Deep Neural Network method to highlight the different constraints in terms of timing and quality of spatial registration to minimise the spatial displacement error for a specific scene.
△ Less
Submitted 12 May, 2023;
originally announced May 2023.
