Divergence and dispersion of global eddy propagation from satellite altimetry

It is well understood that isolated eddies are presumed to propagate westward intrinsically at the speed of the annual baroclinic Rossby wave. This classic description, however, is known to be frequently violated in both propagation speed and its direction in the real ocean. Here, we present a systematic analysis on the divergence of eddy propagation direction (i.e., global pattern of departure from due west) and dispersion of eddy propagation speed (i.e., zonal pattern of departure from Rossby wave phase speed). Our main findings include the following: 1) A global climatological phase map (the first of its kind to our knowledge) indicating localized direction of most likely eddy propagation has been derived from twenty-eight years (1993-2020) of satellite altimetry, leading to a leaf-like full-angle pattern in its overall divergence. 2) A meridional deflection map of eddy motion is created with prominent equatorward/poleward deflecting zones identified, revealing that it is more geographically correlated rather than polarity determined as previously thought (i.e., poleward for cyclonic eddies and equatorward for anticyclonic ones). 3) The eddy-Rossby wave relationship has a duality nature (waves riding by eddies) in five subtropical bands centered around 27°N and 26°S in the two hemispheres, outside which their relationship has a dispersive nature with dominant waves (eddies) propagating faster in the tropical (extratropical) oceans. Current, wind and topographic effects are major external forcings responsible for the observed divergence and dispersion of eddy propagations. These results are expected to make a significant contribution to eddy trajectory prediction using physically based and/or data-driven models.

Long short-term memory networks enhance rainfall-runoff modelling at the national scale of Denmark

This study explores the application of long short-term memory (LSTM) networks to simulate runoff at the national scale of Denmark using data from 301 catchments. This is the first LSTM application on Danish data. The results were benchmarked against the Danish national water resources model (DK-model), a physically based hydrological model. The median Kling-Gupta Efficiency (KGE), a common metric to assess performance of runoff predictions (optimum of 1), increased from 0.7 (DK-model) to 0.8 (LSTM) when trained against all catchments. Overall, the LSTM outperformed the DK-model in 80% of catchments. Despite the compelling KGE evaluation, the water balance closure was modelled less accurately by the LSTM. The applicability of LSTM networks for modelling ungauged catchments was assessed via a spatial split-sample experiment. A 20% spatial hold-out showed poorer performance of the LSTM with respect to the DK model. However, after pre-training, that is, weight initialisation obtained from training against simulated data from the DK-model, the performance of the LSTM was effectively improved. This formed a convincing argument supporting the knowledge-guided machine learning (ML) paradigm to integrate physically based models and ML to train robust models that generalise well.

Comparison of Rainfall-Runoff Simulation between Support Vector Regression and HEC-HMS for a Rural Watershed in Taiwan

To better understand the effect and constraint of different data lengths on the data-driven model training for the rainfall-runoff simulation, the support vector regression (SVR) approach was applied to the data-driven model as the core algorithm in the present study. Various features selection strategies and different data lengths were employed in the training phase of the model. The validated results of the SVR were compared with the rainfall-runoff simulation derived from a physically based hydrologic model, the Hydrologic Modeling System (HEC-HMS). The HEC-HMS was considered a conventional approach and was also calibrated with a dataset period identical to the SVR. Our results showed that the SVR and HEC-HMS models could be adopted for short and long periods of rainfall-runoff simulation. However, the SVR model estimated the rainfall-runoff relationship reasonably well even if the observational data of one year or one typhoon event was used. In contrast, the HEC-HMS model needed more parameter optimization and inference processes to achieve the same performance level as the SVR model. Overall, the SVR model was superior to the HEC-HMS model in the performance of the rainfall-runoff simulation.

Calculation of the Wöhler (S-N) Curve Using a Two-Scale Model

This chapter deals with the initiation of a short crack and subsequent growth of the long crack in a carbon steel under cyclic loading, concluded with the estimation of the complete lifetime represented by the Wöhler (S-N) curve. A micro-model containing the microstructure of the material is generated using the Finite Element Method and the according non-uniform stress distribution is calculated afterwards. The number of cycles needed for crack initiation is estimated on the basis of the stress distribution in the microstructural model and by applying the physically-based Tanaka-Mura model. The long crack growth is handled using the Paris law. The analysis yields good agreement with experimental results from literature.

Assessment of Variations in Runoff Due to Landcover Changes Using the SWAT Model in an Urban River in Dublin, Ireland

Investigating the impact of land cover change in hydrological modelling is essential for water resources management. This paper investigates the importance of landcover change in the development of a physically-based hydrological model called SWAT. The study area considered is the Dodder River basin located in southern Dublin, Ireland. Runoff at the basin outlet was simulated using SWAT for 1993–2019 using five landcover maps obtained for 1990, 2000, 2006, 2012 and 2018. Results indicate that, in general, the SWAT model-simulated runoff for a chosen time-period are closer to the real-world observations when the landcover data used for simulation was collated as close to the time-period for which the simulations were performed. For 23 (20) years (from 27 years period) the monthly mean (maximum) runoff for the Dodder River generated by the SWAT model had the least error when the nearby landcover data were used. This study indicates the necessity of considering dynamic and time-varying landcover data during the development of hydrological modelling for runoff simulation. Furthermore, two composite quantile functions were generated by using a kappa distribution for monthly mean runoff and GEV distribution for monthly maximum runoff, based on model simulations obtained using different landcover data corresponding to different time-period. Modelling landcover change patterns and development of projected landcover in the future for river basins in Ireland needs to be integrated with SWAT to simulate future runoff.

Hybrid Gravimetry to Map Water Storage Dynamics in a Mountain Catchment

In mountain areas, both the ecosystem and the local population highly depend on water availability. However, water storage dynamics in mountains is challenging to assess because it is highly variable both in time and space. This calls for innovative observation methods that can tackle such measurement challenge. Among them, gravimetry is particularly well-suited as it is directly sensitive–in the sense it does not require any petrophysical relationship–to temporal changes in water content occurring at surface or underground at an intermediate spatial scale (i.e., in a radius of 100 m). To provide constrains on water storage changes in a small headwater catchment (Strengbach catchment, France), we implemented a hybrid gravity approach combining in-situ precise continuous gravity monitoring using a superconducting gravimeter, with relative time-lapse gravity made with a portable Scintrex CG5 gravimeter over a network of 16 stations. This paper presents the resulting spatio-temporal changes in gravity and discusses them in terms of spatial heterogeneities of water storage. We interpret the spatio-temporal changes in gravity by means of: (i) a topography model which assumes spatially homogeneous water storage changes within the catchment, (ii) the topographic wetness index, and (iii) for the first time to our knowledge in a mountain context, by means of a physically based distributed hydrological model. This study therefore demonstrates the ability of hybrid gravimetry to assess the water storage dynamics in a mountain hydrosystem and shows that it provides observations not presumed by the applied physically based distributed hydrological model.

Deep Learning in Multi-step Forecasting of Chaotic Dynamics

The prediction of chaotic dynamical systems’ future evolution is widely debated and represents a hot topic in the context of nonlinear time series analysis. Recent advances in the field proved that machine learning techniques, and in particular artificial neural networks, are well suited to deal with this problem. The current state-of-the-art primarily focuses on noise-free time series, an ideal situation that never occurs in real-world applications. This chapter provides a comprehensive analysis that aims at bridging the gap between the deterministic dynamics generated by archetypal chaotic systems, and the real-world time series. We also deeply explore the importance of different typologies of noise, namely observation and structural noise. Artificial intelligence techniques turned out to provide robust predictions, and potentially represent an effective and flexible alternative to the traditional physically-based approach for real-world applications. Besides the accuracy of the forecasting, the domain-adaptation analysis attested the high generalization capability of the neural predictors across a relatively heterogeneous spatial domain.

Horizontal Grid Spacing Comparison Among Random Forest Algorithms to Nowcast Cloud-to-Ground Lightning Occurrence

This paper discusses the use of Random Forest (RF), a popular Machine Learning (ML) algorithm, to perform spatially explicit nowcasting of cloud-to-ground lightning occurrence. An application to the Italian territory and the surrounding seas is presented. Specifically, 1-hour ahead lightning occurrences over the months of August, September and October from 2017 to 2019 have been modelled using a dataset including geo-environmental features. Results obtained with three different spatial resolutions have been compared, for nowcasting both positive and negative strokes. The features’ importance resulting from the best RF models showed how datadriven models are able to identify the relationships between meteorological variables, in agreement with previous physically based knowledge of the phenomenon. The encouraging results obtained in terms of forecasting accuracy support the idea to use ML-based algorithms in early warning procedures for disaster risk management.