Comparison of ARIMA and LSTM for prediction of hemorrhagic fever at different time scales in China

Objectives This study intends to build and compare two kinds of forecasting models at different time scales for hemorrhagic fever incidence in China. Methods Autoregressive Integrated Moving Average (ARIMA) and Long Short-Term Memory Neural Network (LSTM) were adopted to fit monthly, weekly and daily incidence of hemorrhagic fever in China from 2013 to 2018. The two models, combined and uncombined with rolling forecasts, were used to predict the incidence in 2019 to examine their stability and applicability. Results ARIMA (2, 1, 1) (0, 1, 1)12, ARIMA (1, 1, 3) (1, 1, 1)52 and ARIMA (5, 0, 1) were selected as the best fitting ARIMA model for monthly, weekly and daily incidence series, respectively. The LSTM model with 64 neurons and Stochastic Gradient Descent (SGDM) for monthly incidence, 8 neurons and Adaptive Moment Estimation (Adam) for weekly incidence, and 64 neurons and Root Mean Square Prop (RMSprop) for daily incidence were selected as the best fitting LSTM models. The values of root mean square error (RMSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) of the models combined with rolling forecasts in 2019 were lower than those of the direct forecasting models for both ARIMA and LSTM. It was shown from the forecasting performance in 2019 that ARIMA was better than LSTM for monthly and weekly forecasting while the LSTM was better than ARIMA for daily forecasting in rolling forecasting models. Conclusions Both ARIMA and LSTM could be used to build a prediction model for the incidence of hemorrhagic fever. Different models might be more suitable for the incidence prediction at different time scales. The findings can provide a good reference for future selection of prediction models and establishments of early warning systems for hemorrhagic fever.

Optimising Energy Management in Hybrid Microgrids

This article deals with the optimization of the operation of hybrid microgrids. Both the problem of controlling the management of load sharing between the different generators and energy storage and possible solutions for the integration of the microgrid into the electricity market will be discussed. Solar and wind energy as well as hybrid storage with hydrogen, as renewable sources, will be considered, which allows management of the energy balance on different time scales. The Machine Learning method of Decision Trees, combined with ensemble methods, will also be introduced to study the optimization of microgrids. The conclusions obtained indicate that the development of suitable controllers can facilitate a competitive participation of renewable energies and the integration of microgrids in the electricity system.

The Empirical Adaptive Wavelet Decomposition (EAWD): An adaptive decomposition for the variability analysis of observation time series in atmospheric science

Most observational data sequences in geophysics can be interpreted as resulting from the interaction of several physical processes at several time and space scales. As a consequence, measurements time series have often characteristics of non-linearity and non-stationarity and thereby exhibit strong fluctuations at different time-scales. The variability analysis of a time series consists in decomposing it into several mode of variability, each mode representing the fluctuations of the original time series at a specific time-scale. Such a decomposition enables to obtain a time-frequency representation of the original time series and turns out to be very useful to estimate the dimensionality of the underlying dynamics. Decomposition techniques very well suited to non-linear and non-stationary time series have recently been developed in the literature. Among the most widely used of these technics are the empirical mode decomposition (EMD) and the empirical wavelet transformation (EWT). The purpose of this paper is to present a new adaptive filtering method that combines the advantages of the EMD and EWT technics, while remaining close to the dynamics of the original signal made of atmospheric observations, which means reconstructing as close as possible to the original time series, while preserving its variability at different time scales.

Escalas do tempo

Viral epidemics are processes in which temporality obviously constitutes an essential variable. But different time scales must be distinguished. To see the current pandemic as a singular event is but an illusion due to the “mesoscopic” timescale we are embracing. There is a microscopic scale — that of physiological processes —, a mesoscopic scale, which only allows to see the closest evidence, and a macroscopic scale, that of the ecological determinisms which explain the emergence of the disease in the history of the relationships between species. The article focuses on the mesoscopic level and highlights some semiotic specificities of today’s experience : a temporal suspension, the threat of pure, dramatic and final discontinuity, the behavior of a virus that appears to have “intentionality”, a strong intensity coupled with a long duration, a time of exception, drawn to a final end, and a victory which will only be achieved with great effort.

What's past is past, mostly: Brassicaceae host plants mask the feedback from the previous year's soil history on bacterial communities, except when the Brassicaceae hosts experience drought

Previous soil history and the current plant hosts are two plant-soil feedbacks that operate at different time-scales to influence the structure soil bacterial communities. In this study, we used a MiSeq metabarcoding strategy to describe the impact of five Brassicaceae host plant species, and three different soil histories, on the structure of their bacterial root and rhizosphere communities at full flower. We found that the Brassicaceae host plants were consistently significant in structuring the bacterial communities. Four host plants (Sinapis alba, Brassica napus, B. juncea, B. carinata) formed nearly the same bacterial communities, regardless of soil history. Camelina sativa host plants structured phylogenetically distinct bacterial communities compared to the other hosts, particularly in their roots. Soil history established the previous year was only a significant factor for bacterial community structure when the feedback of the Brassicaceae host plants was weakened, potentially due to limited soil moisture during a dry year. Understanding how plant-soil feedbacks operate at different time-scales and are involved in how microbial communities are structured is a pre-requisite for employing microbiome technologies in improving agricultural systems.

Coping strategies for extreme weather in dryland agriculture

Dryland areas account for 48% of area under food crop and 52% under non-food crop cultivation and contribute about 42% of total food grain production in India. Drought is the predominant weather extreme influencing the socio-economic structure of not only dry land regions but also the entire country. Various drought monitoring techniques and mechanisms aim at mitigating the drought impacts at different spatial scales. They are discussed briefly here with strategies to cope up this weather anomaly at different time scales. Dry land districts that are prone to frequent hail episodes are identified and measures to minimize damage to agriculture are also discussed.

Buckling versus crystal expulsion controlled by deformation rate of particle-coated air bubbles in oil

Oil foams stabilized by crystallizing agents exhibit outstanding stability and show promise for applications in consumer products. The stability and mechanics imparted by the interfacial layer of crystals underpin product shelf-life, as well as optimal processing conditions and performance in applications. Shelf-life is affected by the stability against bubble dissolution over a long time scale, which leads to slow compression of the interfacial layer. In processing flow conditions, the imposed deformation is characterized by much shorter time scales. In practical situations, the crystal layer is therefore subjected to deformation on extremely different time scales. Despite its importance, our understanding of the behavior of such interfacial layers at different time scales remains limited. To address this gap, here we investigate the dynamics of single, crystal-coated bubbles isolated from an oleofoam, at two extreme timescales: the diffusion-limited timescale characteristic of bubble dissolution 10,000 s, and a fast time scale characteristic of processing flow conditions, 0.001 s. In our experiments, slow deformation is obtained by bubble dissolution, and fast deformation in controlled conditions with real-time imaging is obtained using ultrasound-induced bubble oscillations. The experiments reveal that the fate of the interfacial layer is dramatically affected by the dynamics of deformation: after complete bubble dissolution, a continuous solid layer remains; while after fast, oscillatory deformation of the layer, small crystals are expelled from the layer. This observation shows promise towards developing stimuli-responsive systems, with sensitivity to deformation rate, in addition to the already known thermo- and photo-responsiveness of oleofoams.