Implementing the autonomous adaptive algorithm to manage ESP operation in harsh reservoir conditions

The well geometries with a shallow kick-off point in conjunction with surface infrastructure limitations have led to Electrical Submersible Pump (ESP) technologies' application as one of the most suitable artificial lift methods for the harsh reservoir conditions. However, the harsh reservoir conditions in terms of the low reservoir pressure, high reservoir temperature, scaling problems in various forms, and high gas content at the pump intake have reduced the ESP system run life. Therefore, this research represents the Autonomous Adaptive Algorithm (A3) as a holistic approach to integrate analytical and machine learning models to assist production engineers in the early detection of operating problems. The A3 relies on different data sources and uses unique, well diagnostics logic to generate valuable features and prepare data for training. Finally, the paper evaluates different classifiers and explores the possibilities of application A3 as a flexible edge solution. The research benefits will be demonstrated for several problematic ESP wells.

Electronic structure and signature of Tomonaga–Luttinger liquid state in epitaxial CoSb1−x nanoribbons

Recently, monolayer CoSb/SrTiO3 has been proposed as a candidate harboring interfacial superconductivity in analogy with monolayer FeSe/SrTiO3. Experimentally, while the CoSb-based compounds manifesting as nanowires and thin films have been realized on SrTiO3 substrates, serving as a rich playground, their electronic structures are still unknown and yet to be resolved. Here, we have fabricated CoSb1−x nanoribbons with quasi-one-dimensional stripes on SrTiO3(001) substrates using molecular beam epitaxy and investigated the electronic structure by in situ angle-resolved photoemission spectroscopy. Straight Fermi surfaces without lateral dispersions are observed. CoSb1−x/SrTiO3 is slightly hole doped, where the interfacial charge transfer is opposite to that in monolayer FeSe/SrTiO3. The spectral weight near the Fermi level exhibits power-law-like suppression and obeys a universal temperature scaling, serving as the signature of Tomonaga–Luttinger liquid (TLL) state. The obtained TLL parameter of ~0.21 shows the underlying strong correlations. Our results not only suggest CoSb1−x nanoribbon as a representative TLL system but also provide clues for further investigations on the CoSb-related interface.

Eliminating the “hook” in Precipitation-Temperature Scaling

Observational studies of extreme daily and subdaily precipitation-temperature sensitivities (apparent scaling) aim to provide evidence and improved understanding of how extreme precipitation will respond to a warming climate. However, interpretation of apparent scaling results is hindered by large variations in derived scaling rates and divergence from theoretical and modelled projections of systematic increases in extreme precipitation intensities (climate scaling). In warmer climatic regions, rainfall intensity has been reported to increase with temperature to a maximum before decreasing, creating a second order discontinuity or “hook” like structure. Here we investigate spatial and temporal discrepancies in apparent scaling results by isolating rainfall events and conditioning event precipitation on duration. We find that previously reported negative apparent scaling at higher temperatures which creates the hook structure, is the result of a decrease in the duration of the precipitation event, and not to the decrease in precipitation rate. We introduce standardized pooling using long records of Australian station data across climate zones, to show average precipitation intensities and 1-h peak precipitation intensities increase with temperature across all event durations and locations investigated. For shorter duration events (< 6-h), average precipitation intensity scaling is in line with the expected Clausius- Clapeyron (CC) relation at ~7 %/°C, and this decreases with increasing duration, down to 2 %/°C at 24-h duration. Consistent with climate scaling derived from model projections, 1-h peak precipitation intensities are found to increase with temperature at elevated rates compared to average precipitation intensities, with super-CC scaling (10 – 14 %/°C) found for short-duration events in tropical climates.

Comparing Kullback-Leibler Divergence and Mean Squared Error Loss in Knowledge Distillation

Knowledge distillation (KD), transferring knowledge from a cumbersome teacher model to a lightweight student model, has been investigated to design efficient neural architectures. Generally, the objective function of KD is the Kullback-Leibler (KL) divergence loss between the softened probability distributions of the teacher model and the student model with the temperature scaling hyperparameter τ. Despite its widespread use, few studies have discussed how such softening influences generalization. Here, we theoretically show that the KL divergence loss focuses on the logit matching when τ increases and the label matching when τ goes to 0 and empirically show that the logit matching is positively correlated to performance improvement in general. From this observation, we consider an intuitive KD loss function, the mean squared error (MSE) between the logit vectors, so that the student model can directly learn the logit of the teacher model. The MSE loss outperforms the KL divergence loss, explained by the penultimate layer representations difference between the two losses. Furthermore, we show that sequential distillation can improve performance and that KD, using the KL divergence loss with small τ particularly, mitigates the label noise. The code to reproduce the experiments is publicly available online at https://github.com/jhoon-oh/kd_data/.

Summer drought in Switzerland 1981‒2020: Events, trends and drivers

&lt;p&gt;The Alpine region has recently experienced several dry summers with negative impacts on the economy, society and ecology. Here, soil water, evapotranspiration and meteorological data from several observational and model-based data sources is used to assess events, trends and drivers of summer drought in Switzerland in the period 1981&amp;#8210;2020. 2003 and 2018 are identified as the driest summers followed by somewhat weaker drought conditions in 2020, 2015 and 2011. We find clear evidence for an increasing summer drying in Switzerland. The observed climatic water balance (-39.2&amp;#160;mm/decade) and 0-1 m soil water from reanalysis (ERA5-Land: -4.7&amp;#160;mm/decade; ERA5: -7.2&amp;#160;mm/decade) show a clear tendency towards summer drying with decreasing trends in most months. Increasing evapotranspiration (potential evapotranspiration: +21.0&amp;#160;mm/decade; ERA5-Land actual evapotranspiration: +15.1&amp;#160;mm/decade) is identified as important driver which scales excellently (+4 to +7%/K) with the observed strong warming of about 2&amp;#176;C. An insignificant decrease in precipitation further enhanced the tendency towards drier conditions. Most simulations of the EURO-CORDEX regional climate model ensemble underestimate the changes in summer drying. They underestimate both, the observed recent summer warming and the small decrease in precipitation. The changes in temperature and precipitation are negatively correlated, i.e. simulations with stronger warming tend to show (weak) decreases in precipitation. However, most simulations and the reanalysis overestimate the correlation between temperature and precipitation and the precipitation-temperature scaling on the interannual time scale. Our results emphasize that the analysis of the regional summer drought evolution and its drivers remains challenging especially with regional climate model data but considerable uncertainties also exist in reanalysis data sets.&lt;/p&gt;

Temperature scaling of extreme precipitation events – an application of the new DWD event catalogue

&lt;p&gt;According to the Clausius-Clapeyron equation on saturation vapour pressure a temperature increase of 1&amp;#160;K allows an atmospheric air mass to hold approximately 7&amp;#160;% more water vapour thus increasing its potential for heavy precipitation. Several published measurement studies on the relation between precipitation intensity and temperature, however, revealed an increase of even up to twofold the CC rate for short-term precipitation events. Model conceptions explain this scaling behaviour with increasing temperature by different intensification pathways of convective processes and/or a transition between stratiform and convective precipitation regimes that both can hardly be verified by point measurements alone. In this presentation, we present first results of the correlation between ambient air temperature and different attributes of the Catalogue of Radar-based Heavy Rainfall Events (CatRaRE) recently published by Deutscher Wetterdienst (DWD). This object-oriented event catalogue files and characterizes extreme precipitation events that have occurred on German territory since 2001. It is based on the high-resolution precipitation climate data set RADKLIM of DWD, i.e. contiguous radar-based reflectivity measurements adjusted to hourly station-based precipitation totals and corrected for typical measurement errors applying specific climatological correction methods. Our analysis gives new insights into potential explanations of the observed temperature scaling relating not only precipitation intensity but characteristic event properties like area, duration, and extremity indices with ambient temperature data. With this approach, extreme precipitation events can be analysed in a comprehensive way that is significant in the context of potential impact. The presented analysis moreover allows testing the hypothesis of regime changing based on objective precipitation event criteria that are typical for different precipitation types. We will briefly present the methodological background of CatRaRE with special focus on the event attributes used in the analysis of Clausius-Clapeyron scaling and give first results on the retrieved temperature dependencies of extreme precipitation events.&lt;/p&gt;

Radiative cooling by clouds affects the precipitation - temperature scaling derived from observations

&lt;p&gt;One direct effect of climate warming on hydrology is the increase in moisture holding capacity of atmosphere at the rate of 7%/&amp;#176;C as suggested by the Clausius Clapeyron equation. Extreme precipitation largely depends on the amount of precipitable water in the atmospheric column and is thus expected to scale with temperature at the same rate. Observations, however, show significant variability in precipitation - temperature scaling rates, with negative scaling dominating in the tropical regions. These scaling relationships assume a one way causality, i.e. temperature is independent of precipitation. However, we show here that temperatures strongly co-vary with precipitation through the effect that clouds have on surface radiation. The presence of clouds associated with precipitation events result in lower solar isolation at the surface, further leading to reduced temperatures. This induces a two-way causality and thus temperature is no longer independent of precipitation. To remove this cooling effect of clouds, we used a surface energy balance model with a thermodynamic constraint to derive clear sky temperatures during precipitation events. We then show using observations from India, that extreme precipitation scaled with clear sky temperatures shows an increase consistent with the CC rate. On contrary, the negative scaling obtained using observed temperatures misrepresent the precipitation response to warming as a result of the co-variation with the cloud radiative effect. Our findings reveal that scaling relationships not only show how precipitation changes with temperature but also how atmospheric conditions associated with precipitation affect temperature. Thus, this covariation needs to be taken into account when using observations to derive scaling relationships that are then used to infer the extreme precipitation response to climate change.&lt;/p&gt;

A Calibrated Multiexit Neural Network for Detecting Urothelial Cancer Cells

Deep convolutional networks have become a powerful tool for medical imaging diagnostic. In pathology, most efforts have been focused in the subfield of histology, while cytopathology (which studies diagnostic tools at the cellular level) remains underexplored. In this paper, we propose a novel deep learning model for cancer detection from urinary cytopathology screening images. We leverage recent ideas from the field of multioutput neural networks to provide a model that can efficiently train even on small-scale datasets, such as those typically found in real-world scenarios. Additionally, we argue that calibration (i.e., providing confidence levels that are aligned with the ground truth probability of an event) has been a major shortcoming of prior works, and we experiment a number of techniques to provide a well-calibrated model. We evaluate the proposed algorithm on a novel dataset, and we show that the combination of focal loss, multiple outputs, and temperature scaling provides a model that is significantly more accurate and calibrated than a baseline deep convolutional network.