Machine learning for accelerating 2D flood models: Potential and challenges

Two-dimensional hydrodynamic models numerically solve full Shallow Water Equations (SWEs). Despite their high accuracy, these models have long simulation run times and therefore are of limited use for exploratory or real-time flood predictions. We investigated the possibility of improving flood modelling speed using Machine Learning (ML). We propose a new method that replaces the computationally expensive parts of the hydrodynamic models with simple and efficient data-driven approximations. Our hypothesis is that by integrating ML with physics-based numerical methods, we can achieve improved generalization performance: that is, the trained model for one case study can be used in other studies without the need for new training. We tested two ML approaches: for the first, we integrated curve fitting, and, for the second, artificial neural networks (ANN) with a finite volume scheme to solve the local inertial approximation of the SWEs. The data-driven models approximated the Momentum Equation, which explicitly solved the time derivative of flow rates. Water depths were then updated by applying a water balance equation. We also tested two different training datasets: the simulated dataset, generated from the results of hydrodynamic model, and the random dataset, generated by directly solving the momentum equation on randomly sampled input data. Various combinations of input features, for example, water slope and depth, were explored. The proposed models were trained in a small hypothetical case and tested in a different hypothetical and in two real case studies. Results showed that the curve-fitting method can be implemented successfully, given sufficient training and input data. The ANN model trained with a random dataset was substantially more accurate than that of the model trained with the simulated dataset. However, it was not successful in the real case studies. The curve-fitting method resulted in better generalization performance and increased the simulation speed of the local inertial model by 23%. Future research should test the performance of ML in terms of an increase in stable time step size and approximation of the full SWEs.     

Application of Fuzzy Sets to the Determination of Sunspot Areas

The computation of sunspot areas is important for many applications in solar physics. Some uncertainty is, however, inherent to this determination, since there is uncertainty in defining the limits of sunspots, as well as their umbrae and penumbrae, particularly in high-resolution images where large variations in intensity are observed, mainly in the penumbrae and photosphere regions. In this paper a methodology based on the classical histogram method is presented that enables the generation of fuzzy sets that correspond to the umbrae and penumbrae of sunspots, as well as to the complete sunspots, which enable the representation of the uncertainty in their locations. The areas of the sunspots and of their umbrae and penumbrae are then obtained by computing the areas of the fuzzy sets used to represent them. Two operators are used: the Rosenfeld Area operator, which generates a real number for the area, and the Fuzzy Area operator, which generates a fuzzy number, providing more information about the uncertainty of the area. A comparison of the obtained results with the area values given by the classical cumulative histogram method is made and the differences analyzed.     

PAH transport by sinking particles in the open Mediterranean Sea: a 1 year sediment trap study

One year time series of sinking particles were collected at two depths in the open Mediterranean Sea and analysed for polycyclic aromatic hydrocarbons (PAH). Average total PAH concentrations were 593+/-284 ng g(-1) at 250 m and 551 +/- 198 ng g(-1) at 2850 m. Total PAH fluxes averaged 73 +/- 58 ng m(-2) d(-1) at 250 m and 53 +/- 39 ng m(-2) d(-1) at 2850 m. Contamination levels and, thus, exposure of marine organisms to PAH are comparable in surface and deep waters. Deep waters appear as a significant, yet overlooked, PAH sink. PAH temporal patterns show noticeable seasonality. This is partly due to varying levels of specific components such as the winter increase of pyrolytic PAH. Downward transport processes and the nature of sinking particles also impact on PAH fluxes, as inferred during periods of increasing productivity. Different phase-associations and interactions with particulate organic carbon for low-MW fossil PAH and high-MW pyrolytic PAH influence their downward transport efficiency.     

Formulation of a tropical town energy budget (t-TEB) scheme

This work describes the tropical town energy budget (t-TEB) scheme addressed to simulate the diurnal occurrence of the urban heat island (UHI) as observed in the Metropolitan Area of Rio de Janeiro (MARJ; ?22° S; ?44° W) in Brazil. Reasoning about the tropical urban climate have guided the scheme implementation, starting from the original equations from Masson (Bound-Lay Meteorol 94:357–397, 2000). The modifications include (a) local scaling approaches for obtaining flux–gradient relationships in the roughness sub-layer, (b) the Monin-Obukhov similarity framework in the inertial sub-layer, (c) increasing aerodynamic conductance toward more unstable conditions, and (d) a modified urban subsurface drainage system to transfer the intercepted rainwater by roofs to the roads. Simulations along 2007 for the MARJ are obtained and compared with the climatology. The t-TEB simulation is consistent with the observations, suggesting that the timing and dynamics of the UHI in tropical cities could vary significantly from the familiar patterns observed in mid-latitude cities—with the peak heat island intensity occurring in the morning than at night. The simulations are suggesting that the thermal phase shift of this tropical diurnal UHI is a response of the surface energy budget to the large amount of solar radiation, intense evapotranspiration, and thermal response of the vegetated surfaces over a very humid soil layer.     

Dimensionality reduction in drought modelling

For monitoring hydrological events characterized by high spatial and temporal variability, the number and location of recording stations must be carefully selected to ensure that the necessary information is collected. Depending on the characteristics of each natural process, certain stations may be spurious or redundant, whereas others may provide most of the relevant data. With the objective of reducing the costs of the monitoring system and, at the same time, improving its operational effectiveness, three procedures were applied to identify the minimum network of rain gauge stations able to capture the characteristics of droughts in mainland Portugal. Drought severity is characterized by the standardized precipitation index applied to the timescales of 1, 3, 6 and 12 consecutive months. The three techniques used to reduce the dimensionality of the network of rain gauges were as follows: (i) artificial neural networks with sensitivity analysis, (ii) application of the mutual information criterion and (iii) K‐means cluster analysis using Euclidean distances. The results demonstrated that the best dimensionality reduction method was case dependent in the three regions of Portugal (northern, central and southern) previously identified by cluster analysis. All the reduction techniques lead to the selection of a subset of rain gauges capable of reproducing the original temporal patterns of drought. For specific severe drought events in Portugal in the past, the comparison between drought spatial patterns obtained with the original stations and the selected subset indicated that the subset produced statistically satisfactory results (correlation coefficients higher than 0.6 and efficiency coefficients higher than 0.5). Copyright © 2012 John Wiley & Sons, Ltd.