The short-term prediction of precipitation is a difficult spatio-temporal task due to the non-uniform characterization of meteorological structures over time. Currently, neural networks such as convolutional LSTM have shown ability for the spatio-temporal prediction of complex problems. In this research, we propose an LSTM convolutional neural network (CNN-LSTM) architecture for immediate prediction of various short-term precipitation events using satellite data. The CNN-LSTM is trained with NASA Global Precipitation Measurement (GPM) precipitation data sets, each at 30-min intervals. The trained neural network model is used to predict the sixteenth precipitation data of the corresponding fifteen precipitation sequence and up to a time interval of 180 min. The results show that the increase in the number of layers, as well as in the amount of data in the training data set, improves the quality of the forecast.

We study the effect of local atomic- and nano-scale protrusions on field emission and, in particular, on the local field enhancement which plays a key role as known from the Fowler-Nordheim model of electronic emission. We study atomic size defects which consist of right angle steps forming an infinite length staircase on a tungsten surface. This structure is embedded in a 1 GV/m ambient electrostatic field. We perform calculations based upon density functional theory in order to characterize the total and induced electronic densities as well as the local electrostatic fields taking into account the detailed atomic structure of the metal. We show how the results must be processed to become comparable with those of a simple homogeneous tungsten sheet electrostatic model. We also describe an innovative procedure to extrapolate our results to nanoscale defects of larger sizes, which relies on the microscopic findings to guide, tune, and improve the homogeneous metal model, thus gaining predictive power. Furthermore, we evidence analytical power laws for the field enhancement characterization. The main physics-wise outcome of this analysis is that limited field enhancement is to be expected from atomic- and nano-scale defects.

The Ar3 system has been studied between T=0 K and T=40 K by means of a path-integral Monte Carlo (PIMC) method. The behavior of the average energy in terms of the temperature has been explained by comparison with results obtained with the thermal averaged rovibrational spectra estimated via: (i) a quantum mechanical method based on distributed Gaussian functions for the interparticle distances and (ii) an analytical model which precisely accounts for the participation of the dissociative continua Ar2+Ar and Ar+Ar+Ar. Beyond T∼20 K, the system explores floppier configurations than the rigid equilateral geometry, as linear and Ar–Ar2-like arrangements, and fragmentates around T∼40 K. A careful investigation of the specific heat in terms of a confining radius in the PIMC calculation seems to discard a proper phase transition as in larger clusters, in apparent contradiction with previous reports of precise values for a liquid-gas transition. The onset of this noticeable change in the dynamics of the trimer occurs, however, at a remarkably low value of the temperature in comparison with Arn systems formed with more Ar atoms. Quantum mechanical effects are found of relevance at T≤15 K, with both energies and radial distributions obtained with a quantum PIMC deviating from the corresponding classical results, thus precluding exclusively classical approaches for a precise description of the system at this low temperature range.

A detailed study of the rovibrational spectrum of the Ar trimer is performed by means of an exact hyperspherical coordinate (HC) method and a variational approach based on distributed Gaussian functions (DGFs) to describe the interparticle distances. The good agreement observed between the energy levels obtained with both procedures for high values of the total angular momentum ($J=15$ and 20) reveals the quality of the DGF method to describe the rotation of the title system. Rotational constants for the lowest bound states, obtained as averages for each vibrational state, have been obtained and compared to previous results. A detailed analysis of density probability functions obtained by means of the HC approach for rovibrational states at $J=0$ and 20 shows close similitudes thus supporting the vibration-rotation separation adopted within the DGF scheme for the ${\text{Ar}}_3$ system.

The effect of metallic surface contamination on the field electron emission is investigated for the first time using a three dimensional quantum mechanical model. The plane wave periodic version of the density functional theory is used to obtain wavefunctions and potentials. Local and averaged emitted current densities are obtained from them using the time dependent perturbation theory. This method is used to study the effect of the presence of carbon adsorbates on emission from tungsten surfaces. Fowler-Nordheim plots, which provide the dependence of the emitted current with the external electric field, show that carbon contamination inhibits emission. Significant differences with the results of the analytical Fowler-Nordheim model are observed. Emission images (i.e., the spatial dependence of the emitted current density) are presented to identify the important emission spots. These images are significantly different from the electronic density plots usually presented to model constant height scanning tunneling microscope images. Analysis of the emitted current density energy distributions in the light of the projected local density of states provides a deeper understanding of the emission process.