Intermediate Store Research Articles

Hydrological concept formation inside long short-term memory (LSTM) networks

Abstract. Neural networks have been shown to be extremely effective rainfall-runoff models, where the river discharge is predicted from meteorological inputs. However, the question remains: what have these models learned? Is it possible to extract information about the learned relationships that map inputs to outputs, and do these mappings represent known hydrological concepts? Small-scale experiments have demonstrated that the internal states of long short-term memory networks (LSTMs), a particular neural network architecture predisposed to hydrological modelling, can be interpreted. By extracting the tensors which represent the learned translation from inputs (precipitation, temperature, and potential evapotranspiration) to outputs (discharge), this research seeks to understand what information the LSTM captures about the hydrological system. We assess the hypothesis that the LSTM replicates real-world processes and that we can extract information about these processes from the internal states of the LSTM. We examine the cell-state vector, which represents the memory of the LSTM, and explore the ways in which the LSTM learns to reproduce stores of water, such as soil moisture and snow cover. We use a simple regression approach to map the LSTM state vector to our target stores (soil moisture and snow). Good correlations (R2>0.8) between the probe outputs and the target variables of interest provide evidence that the LSTM contains information that reflects known hydrological processes comparable with the concept of variable-capacity soil moisture stores. The implications of this study are threefold: (1) LSTMs reproduce known hydrological processes. (2) While conceptual models have theoretical assumptions embedded in the model a priori, the LSTM derives these from the data. These learned representations are interpretable by scientists. (3) LSTMs can be used to gain an estimate of intermediate stores of water such as soil moisture. While machine learning interpretability is still a nascent field and our approach reflects a simple technique for exploring what the model has learned, the results are robust to different initial conditions and to a variety of benchmarking experiments. We therefore argue that deep learning approaches can be used to advance our scientific goals as well as our predictive goals.

Explorative Supercooling Technology for Prevention of Freeze Damages in Vaccines

Most freeze-sensitive vaccines are stored between 2 °C and 8 °C upon manufacturing and until they are eventually administered in intermediate vaccine stores and health facilities. This so-called “cold chain” of vaccine distribution is strictly regulated at these specific temperatures to avoid freeze damage. Liquid formulations of particular vaccines (e.g., aluminum-adsorbed tetanus toxoid (TT)) will irreversibly lose their immunogenicity once frozen. Using an oscillating magnetic field (OMF), supercooling can inhibit ice crystal nucleation effectively; water is susceptible to influence by a strong magnetic field, allowing normal water dynamics even in subzero freezing conditions. This recently developed technology—composed of a custom-designed electromagnet unit producing an optimal field strength (50 mT) at a specific frequency (1 Hz)—was successfully used to inhibit the formation of ice crystals in aluminum adjuvant TT vaccines, therefore preventing any visible damage in the vaccines’ microscopic structure. Despite being subject to temperatures far below their freezing point (up to −14 °C) for up to seven days, the TT vaccines showed no freeze damage on physical appearances. Results were further validated using shake tests and light microscopy. As storage and freeze-protection become more critical during times of increased vaccination efforts—particularly against COVID-19—this supercooling technology can be a promising solution to distribution problems by removing concern for temperature abuse or shock-induced freezing.

Seasonal fluxes and sediment routing in tropical catchments affected by nickel mining

AbstractAn important gap in the management of land erosion in mining‐affected areas is the understanding of the entire sediment routing system and the links between sources and storage at the catchment scale. In this study, we examine sediment delivery and its seasonality in the nickel mining‐affected Santa Cruz and Pamalabawan catchments in the Philippines. We monitored discharge, suspended sediment concentrations and suspended sediment loads across 13 sub‐catchments with contrasting degrees of mining influence from June 2018 to July 2019. First, we show the importance of the size of the area that has been physically disturbed within our sub‐catchments, with as little as 10–22% of relative disturbance area being enough to generate four‐fold to eight‐fold increase in the sediment yield relative to less disturbed and pristine areas. We found that sub‐catchments with > 10% disturbance exhibit the highest sediment yields (15.5 ± 44.7 t km−2 d−1) compared with sub‐catchments with < 10% disturbance (3.6 ± 17.7 t km−2 d−1) and undisturbed catchments (2.0 ± 5.7 t km−2 d−1). We also show that sediment flushing predominantly occurs in the most disturbed sub‐catchments at the onset of the wet season. A small number of flood events transports the bulk of the sediment, with hysteresis effects being most pronounced in disturbed areas. Lastly, we show that floodplain sediment recycling exerts a key control on sediment delivery at both reach and catchment scales, with the relative contribution of floodplain sources to the sediment budget becoming dominant in the latter stages of the wet season‐ up to 89% of the total sediment export per storm event. This study highlights the importance of both degree of disturbance and sediment pathways in controlling sediment transport in mining‐disturbed areas, and that considering the entire sediment routing system including intermediate stores is crucial to optimizing existing and future measures against siltation and potential contamination of trace metals and metalloids downstream of mining areas.

Initiation and evolution of knickpoints and their role in cut-and-fill processes in active submarine channels

Abstract Submarine channels are the main conduits and intermediate stores for sediment transport into the deep sea, including organics, pollutants, and microplastics. Key drivers of morphological change in channels are upstream-migrating knickpoints whose initiation has typically been linked to episodic processes such as avulsion, bend cutoff, and tectonics. The initiation of knickpoints in submarine channels has never been described, and questions remain about their evolution. Sedimentary and flow processes enabling the maintenance of such features in non-lithified substrates are also poorly documented. Repeated high-resolution multibeam bathymetry between 2012 and 2018 in the Capbreton submarine canyon (southeastern Bay of Biscay, offshore France) demonstrates that knickpoints can initiate autogenically at meander bends over annual to multi-annual time scales. Partial channel clogging at tight bends is shown to predate the development of new knickpoints. We describe this initiation process and show a detailed morphological evolution of knickpoints over time. The gradients of knickpoint headwalls are sustained and can grow over time as they migrate through headward erosion. This morphology, associated plunge pools, and/or development of enhanced downstream erosion are linked herein to the formation and maintenance of hydraulic jumps. These insights of autogenically driven, temporally high-frequency knickpoints reveal that cut-and-fill cycles with depths of multiple meters can be the norm in submarine systems.

From somatic pain to psychic pain: The body in the psychoanalytic field

The integration of psyche and soma begins with a baby's earliest contact with his or her parents. With the help of maternal empathy and reverie, β-elements are transformed into α-elements. While we understand this to be the case, we would like to enquire what actually happens to those parts of the affect which have not been transformed? For the most part they may be dealt with by evacuation, but they can also remain within the body, subsequently contributing to psychosomatic symptoms.This paper describes how the body serves as an intermediate store between the psychic (inner) and outer reality. The authors focuses on the unconscious communicative process between the analyst and the analysand, and in particular on how psychosomatic symptoms can spread to the analyst's body. The latter may become sensitive to the analysand's psychosomatic symptoms in order to better understand the psychoanalytical process. Sensory processes (visual and auditory) and psychic mechanisms such as projective identification can serve as a means for this communication.

Evaluation of a Hybrid Energy Storage Module for Pulsed Power Applications

Before pulsed power systems can be fielded in either mobile or small footprint stationary applications, the prime power source must be optimized for both size and operational efficiency. In large footprint laboratories, prime power supplies are connected to a local utility grid to charge intermediate storage systems. In mobile platforms, alternative energy sources, such as electrochemical batteries or supercapacitors, must be used to backup smaller fossil fuel generators. The prime power source used in a pulsed power system must store high energy, to maximize the number of shots stored, and be able to source high power to recharge the intermediate store as fast as possible. Finding a single electrochemical energy storage device that has the right energy and power density for most applications is nearly impossible. Therefore, usage of batteries, which possess high energy density, along with electrochemical capacitors, which offer high power density, in a hybrid energy storage module (HESM) configuration is a promising way of combining both of these features into a single supply. Usage of this topology reduces the stress on the batteries, thereby prolonging their life, and also increases the instantaneous power capabilities of the system. This paper presents the design and validation of an actively controlled HESM built using commercial off the shelf power electronics and simple control strategies.

Kanban number optimisation in a supermarket warehouse feeding a mixed-model assembly system

Following just-in-time principles, a growing number of manufacturers are adopting the so-called supermarket concept. Supermarkets are decentralised storage areas scattered throughout the shopfloor that serve as an intermediate store for parts required by nearby assembly lines. From these stores, a certain number of handling operators deliver parts from the supermarket to, and collect empty bins from, assembly stations. Finally, they return to the supermarket and are refilled for their next tours. The assembly stations are typically refilled from the supermarket through the constant replacement of the consumed parts pulled by the kanban system. Considering a mixed model assembly system composed of different assembly lines, feeding problems can occur as an effect of the replenishment lead time, of the production mix variation, of the commonality between the different models assembled. The aims of this paper are (i) to highlight how the supermarket/multi-mixed assembly-line system presents specific attributes that prohibit the simple application of well-known kanban dimensioning formulations and (ii) to provide an innovative procedure to optimally set all decision variables related to such a feeding system.

The catchment sediment budget as a management tool

There is increasing recognition that fine sediment represents an important diffuse source pollutant in surface waters, due to its role in governing the transfer and fate of many substances, including nutrients, heavy metals, pesticides and other organic contaminants, and because of its impacts on aquatic ecology. Catchment management strategies therefore frequently need to include provision for the control of sediment mobilisation and delivery. The sediment budget concept provides a valuable framework for assisting the management and control of diffuse source sediment pollution and associated problems, by identifying the key sources and demonstrating the importance of intermediate stores and the likely impact of upstream mitigation strategies on downstream suspended sediment and sediment-associated contaminant fluxes. Accordingly, the utility of the sediment budget concept for catchment management is further discussed, by introducing examples from several contrasting river basins.

Active facility management strategy for containment of ILW using risk-based service life modelling of reinforced concrete to incorporate in-service cracking

An Intermediate Level Waste Store facility for on-site waste storage is under construction to contain arisings from historic operations of the Hunterston 'A' NPS, Ayrshire. This paper updates previous a durability design strategy for reinforced concrete structures used in long-term storage (>100 years) of nuclear waste that incorporates design-limit flexural cracking. The approach, utilising probabilistic modelling of deterioration and reliability methods, gives greater flexibility to the decision-making process for management of concrete structures depending on the balance chosen between acceptable risk and maintenance philosophy. The results are compared with contemporary UK Codes concerning durability of reinforced concrete.

Debris flow-dominated and rockfall-dominated talus slopes: Genetic models derived from GPR measurements

Stratified talus deposits are reported from many different mountain environments. Numerous possible explanations are discussed in the literature; however, the sediment stores are rarely accessible as exposures are sparse. We applied ground-penetrating radar (25, 50 and 100 MHz antennae) to gain insight into the internal sediment structures of 23 alpine scree slopes; ten examples are presented in this paper. The study areas are spread over the Eastern European Alps at altitudes ranging from 1500 to 2900 m. The bedrock type is primarily limestone and dolostone; one area is composed of gneiss and mica–schist. GPR turned out to be highly suitable for investigating sediment structures of dry talus debris. The results showed that almost all of the deposits investigated are characterized by pronounced stratification. Several different types of layering were identified. Discordant layers which are restricted to confined parts of the talus are probably related to sediment redistribution processes like surficial debris flows or dry grain flows. These features frequently occur at the uppermost part of the slope caused by overland flow from the adjacent rock face, but may also develop in the downhill part of a talus. One talus in the Reintal area showed surface-parallel, persistent layers of different grain sizes which cannot be explained by any known models. We suggest a novel model of talus development which is driven by climatic fluctuations. In periods of enhanced freeze–thaw activity like the Little Ice Age, the delivery of coarse debris prevails. In warmer climate with a higher frequency of rainstorms, the depletion of finer-grained intermediate stores in less inclined rockwall positions leads to delivery of clasts smaller than 2 cm. The type of layering found within a talus is determined by rockwall parameters like height, steepness, topography and dissection of the rock face. The “storage depletion” model applies to high rockwalls with a considerable volume of intermediate storage.

Mechanisms of compound I formation in heme peroxidases

The formation of compound I is the first step in the reaction mechanism of plant heme peroxidases. This intermediate stores two oxidizing equivalents from hydrogen peroxide as an oxyferryl iron center and a radical, either on the porphyrin ring or on a tryptophan residue. Site-directed mutagenesis has proved to be a most useful tool for the identification of the intermediates involved and the resulting nature of the compound I formed. Although there is no doubt that an acid–base mechanism operates in heme peroxidase during the formation of compound I, the roles of several distal pocket residues are currently the subject of intensive research. It is now generally accepted that the conserved distal histidine in the active site of heme peroxidases is the acid–base catalyst that promotes the heterolytic cleavage of hydrogen peroxide. Other residues, such as the distal arginine and asparagine, participate in a range of roles assisting catalysis by the distal histidine. Recent advances in the elucidation of the mechanism at the molecular level are discussed. Another aspect related to the nature of compound I is the location of the radical center. Novel radical species have been detected in the reactions of ascorbate peroxidase, lignin peroxidase and several mutants of horseradish peroxidase. Detailed kinetic and spectroscopic studies of these radical species have provided important insights about the factors that control porphyrin–protein radical exchange. The wide range of data being obtained on compound I will lead to an understanding of its vital function in peroxidase catalysis and the physiological roles played by these enzymes.

PH: Signal and Messenger in Plant Cells

Abstract: Since water spontaneously ionizes, protons cannot be removed from the medium: their free concentration in cells must be regulated through actively controlling H+‐related transport across membranes, by active and passive buffering, and by setting a certain pH within the metabolic network. Whereas these are the basic tools that provide effective H+ homeostasis, cellular compartmentation serves as an intermediate store into which protons can be shifted temporarily and from which protons can be regained when required. On the other hand, intracellular compartments can also serve as a final proton sink. pH regulation is not confined to intracellular spaces, but also comprises the apoplast. Whereas the pH of the cytosol is kept slightly alkaline at 7.2 to 7.5, with an average buffer capacity of 20 to 80 mM H+ per pH unit, the apoplastic pH may vary among tissues but is always acidic, with values between pH 5 and 6 and with a buffer capacity in the lower millimolar range per pH unit. pH can be a signal and/or a messenger, a distinction not always clearly made. Here, “signal” should be understood as information about an ongoing or preceding process, whereas “messenger” would be the carrying of certain information that will lead to a change of state. As such, pH would signal light intensity changes, drought, lack of oxygen and the presence of symbiotic partners or microbial attackers. On the other hand, pH would be a messenger in situations where pH changes are preconditions for certain processes, e.g., the gravity response or for activation of certain transporters in stomatal movements, and possibly for growth. The function of pH as a cellular messenger raises the question of whether pH should be understood as a “second messenger” in the way this is done for Ca2+. In an effort to give a comprehensive answer to this problem, the different roles of Ca2+ and H+ in cellular signalling are discussed and a number of Ca2+/pH interactions are presented.

Design of a radiographic integrated test stand (RITS) based on a voltage adder, to drive a diode immersed in a high magnetic field

Recent experiments have adapted existing magnetically insulated induction voltage adders (Sabre, Hermes III) to drive a 10-MV diode immersed in magnetic fields as high as 50 T. In such a diode, an electron beam of tens of kiloamperes can be confined by the magnetic field to a diameter of about 1 mm, and when it strikes a high-Z anode, it can create a bremsstrahlung X-ray source intense enough to radiograph massive objects with high resolution. The radiographic integrated test stand (RITS) is an adder system designed specially to drive such diodes, and it will be used to develop and exploit them. As in other adder-based pulsers, such as Sabre, Hermes III, and Kalif-Helia, the induction cells have amorphous-iron cores, and the pulse-forming system consists of water dielectric pulselines and self-closing water switches that are pulse-charged from Marx-charged intermediate water capacitors through laser-triggered Rimfire switches. An oil prepulse switch in series with each pulseline is designed to reduce cathode prepulse to less than /spl plusmn/5 kV, and a means is provided to bias the cathode and avoid negative prepulse entirely. The RITS pulse-forming system consists of two modules. Each module has one Marx that charges two 3-MV intermediate stores, each of which charges three 7.8 /spl Omega/ pulselines, making six pulselines per module. The two modules in concert can supply 1.35-MV, 50-ns pulses to a 12-cell adder and thus drive a 16-MV diode with a single pulse. The 1,35-MV induction cells each have a single-point feed, from which a single, slotted azimuthal oil transmission line distributes energy uniformly around the cell. The modules can also be pulsed separately at different times, either to power two 8-MV adders that each drive one of two closely spaced cathodes immersed in a common magnetic field or to provide two separate pulses to a common six-cell adder and a single 8-MV diode; in these two-pulse modes, the spacing of the two 50-ns pulses may be chosen to be anything from a few hundred nanoseconds upward. The use of only one pulseline per cell has been shown to increase the extent to which the cell voltages can vary with the timing of closure of the water switches. This and all other functions of RITS have been simulated in detail, and a conservative electrical design has been developed. This will be illustrated, along with the conceptual design of a pulse sorting network that can couple two pulselines efficiently to one cell when the two RITS modules drive a common adder in two-pulse mode.

The inclusion of storage in a queueing system

Service operations involving an intermediate store are modelled using an extended definition of the system state which is a conventional definition of state supplemented with an indicator of the live capacity in the store. Additional to the parameters of mean service rate and mean arrival rate, a mean discharge rate of the store is introduced and the steady state probabilities are calculated in terms of these three quantities. Both finite source and infinite source cases are considered under the assumption of exponential distributions. The models permit production comparisons for systems with and for systems without storage devices.

Simplified evaluation and documentation of data from glucose-controlled insulin infusion systems. Part II: Artefact handling.

handling A computerized method for handling artefacts occurring during recording and evaluation of data with glucose-controlled insulin infusion systems (GCIIS) is demonstrated: 1) Recording phase: - »Interrupt«-Mo de: Maintenance of the original linear data-time relation is achieved by using an additional time channel. Analogous structuring of time- and parameter data files permits correct correlation. - »Communication I/O error«: An error-handlin g subroutine prevents the interruption of the program caused by lead characters which occur when restarting the GCIIS thereby disturbing GCIIS-microcomputer communication. - Risk of data loss caused by power failure is minimized by an appropriate interaction between intermediate store and floppy disk. 2) Evaluation phase: - Implausible data: Obstruction of double lumen catheter as well as loss of glucose sensor sensitivity result in inappropriate data with consecutive incorrect computation of glycemic indices such as MAGE and M-value. - Zero-values: In the »HOLD«-mode, all blood glucose values are set to zero by the GCIIS causing forbidden arithmetic operations. These artefact periods are marked in the prepared time-file, deleted and interpolated subsequently. In conclusion, our method permits automatic artefact handling during the recording phase and complete correct data processing during the evaluation phase.

Deacidification in a plant with crassulacean acid metabolism associated with anion-cation balance

In plant tissues using crassulacean acid metabolism (CAM), malate accumulates during the dark and declines in the subsequent light period. The malate is an intermediate store of CO2 which is fixed photosynthetically during the subsequent day and for CAM plants the amplitude of the fluctuation is significant for plant carbon assimilation1. Because they are so considerable, the maximum values for acid content aroused early comment2. The maximum seems to be related not just to the rate of acidification but also to internal factors which may include the water and cation content of tissues3. Changes in the distribution of cations have recently been linked to the diel fluctuation in malate4. I report here that the malate measured in leaves at the end of the day can be correlated with an imbalance between total cations and the organic acids other than malate in the leaves of Kalanchoe daigremontiana. There occurs a greater imbalance in leaves of plants grown in a long compared with a short day and this may explain the finding that leaves contain considerably more malate at the end of a long photoperiod.