Iterative Placement of Decoupling Capacitors using Optimization Algorithms and Machine Learning
Zouhair Nezhi, Nima Ghafarian Shoaee, and Marcus Stiemer
Adv. Radio Sci., 21, 123–132, https://doi.org/10.5194/ars-21-123-2024, 2024
An optimum placement and dimensioning of decaps on a printed circuit board is determined by a Genetic Algorithm (GA). The use of an artificial neural network as surrogate model to compute fitness values for the GA significantly reduces computation time. With the optimization framework at hand, the risk of a redesign that would take several weeks can be significantly reduced by a computation that just needs a few minutes.
Abstract
Specification and forecast of ionospheric parameters, such as ionospheric electron density (Ne), have been an important topic in space weather and ionospheric research. Neural networks (NNs) emerge as a powerful modeling tool for Ne prediction. However, heavy manual adjustments are time consuming to determine the optimal NN structures. In this work, we propose to use neural architecture search (NAS), an automatic machine learning method, to mitigate this problem. NAS aims to find the optimal network structure through the alternate optimization of the hyperparameters and the corresponding network parameters within a pre-defined hyperparameter search space. A total of 16-year data from Millstone Hill incoherent scatter radar (ISR) are used for the NN models. One single-layer NN (SLNN) model and one deep NN (DNN) model are both trained with NAS, namely SLNN-NAS and DNN-NAS, for Ne prediction and compared with their manually tuned counterparts (SLNN and DNN) based on previous studies. Our results show that SLNN-NAS and DNN-NAS outperformed SLNN and DNN, respectively. These NN predictions of Ne daily variation patterns reveal a 27-day mid-latitude topside Ne variation, which cannot be reasonably represented by traditional empirical models developed using monthly averages. DNN-NAS yields the best prediction accuracy measured by quantitative metrics and rankings of daily pattern prediction, especially with an improvement in mean absolute error more than 10% compared to the SLNN model. The limited improvement of NAS is likely due to the network complexity and the limitation of fully connected NN without the time histories of input parameters.
Abstract
To gain deeper insights into radiation belt loss into the atmosphere, a statistical study of MeV electron precipitation during radiation belt dropout events is undertaken. During these events, electron intensities often drop by an order of magnitude or more within just a few hours. For this study, dropouts are defined as a decrease by at least a factor of five in less than 8 hours. Van Allen probe measurements are employed to identify dropouts across various parameters, complemented by precipitation data from the CALorimetric Electron Telescope instrument on the International Space Station. A temporal analysis unveils a notable increase in precipitation occurrence and intensity during dropout onset, correlating with the decline of SYM-H, the north-south component of the interplanetary magnetic field, and the peak of the solar wind dynamic pressure. Moreover, dropout occurrences show correlations with the solar cycle, exhibiting maxima at the spring and autumn equinoxes. This increase during equinoxes reflects the correlation between equinoxes and the SYM-H index, which itself exhibits a correlation with precipitation during dropouts. Spatial analysis reveals that dropouts with precipitation penetrate into lower L-star regions, mostly reaching L-star <4, while most dropouts without precipitation don't penetrate deeper than L-star 5. This is consistent with the larger average dimensions of dropouts associated with precipitation. During dropouts, precipitation is predominantly observed in the dusk-midnight sector, coinciding with the most intense precipitation events. The results of this study provide insight into the contribution of precipitation to radiation belt dropouts by deciphering when and where precipitation occurred.
Abstract
The aurora and associated high energy particles and currents pose a space weather hazard to communication networks and ground-based infrastructure. Forecasting the location of the auroral oval forms an integral component of daily space weather operations. We evaluate a version of the OVATION-Prime 2013 auroral forecast model that was implemented for operational use at the UK Met Office Space Weather Operations Cent. Building on our earlier studies, we evaluate the ability of the OVATION-Prime 2013 model to predict the location of the auroral oval in all latitude and local time sectors under different levels of geomagnetic activity, defined by Kp. We compare the model predictions against auroral boundaries determined from IMAGE FUV data. Our analysis shows that the model performs well at predicting the equatorward extent of the auroral oval, particularly as the equatorward auroral boundary expands to lower latitudes for increasing Kp levels. The model performance is reduced in the high latitude region near the poleward auroral boundary, particularly in the nightside sectors where the model does not accurately capture the expansion and contraction of the polar cap as the open flux content of the magnetosphere changes. For increasing levels of geomagnetic activity (Kp ≥ 3), the performance of the model decreases, with the poleward edge of the auroral oval typically observed at lower latitudes than forecast. As such, the forecast poleward edge of the auroral oval is less reliable during more active and hazardous intervals.
Abstract
Measurements of Ar and N2 densities at 160–250 km altitude from the Mars Atmosphere and Volatile Evolution (MAVEN) Neutral Gas and Ion Mass Spectrometer (NGIMS) during February 2015–February 2023 are analyzed to provide a comprehensive analysis of their diurnal- and zonal-mean (DZM) structures, and ZM (solar-synchronous) diurnal (DW1) and semidiurnal (SW2) tides. After applying a solar flux trend correction, multi-year binning and averaging with respect to longitude, local solar time (LST), latitude and Ls at each height results in the first full global picture of these components of the ZM thermosphere for a single climatological Mars year. The following new observational insights into Mars thermosphere are obtained: The DZM N2 latitude versus Ls (latvsLs) structures contain a prominent latitudinally-symmetric annual component (∼±25%–35%) due to the eccentricity of Mars orbit around the Sun, and an antisymmetric component (∼±30%–45%) below about 190 km that is seasonally-symmetric and thus consistent with the tilt of Mars rotation axis. Aperiodic deviations from these symmetries increase with height and are tentatively attributed to dissipation of waves originating in the lower atmosphere. DW1 and SW2 maximize around 200–220 km altitude, suggesting existence of an unknown dissipation mechanism at higher altitudes. The DZM, DW1 and SW2 components of Ar generally exceed those of N2 by factors of 1.4–2.5. The scale heights of Ar and N2 between 205 and 245 km are also employed to derive DZM exosphere temperatures, which reflect aperiodic ∼±15K deviations from the annual-mean in the latvsLs frame.
Abstract
Pc5 ULF waves play an important role in transporting energy and particles in the coupled magnetospheric and ionospheric system. They are known to be initiated by dynamic pressure fluctuations upstream of the magnetopause, including those induced by hot flow anomalies (HFAs). However, the role of HFAs in generating magnetospheric and ground magnetic Pc5 ULF oscillations has not been investigated statistically yet. Thus, in this paper, we investigate the contribution of HFAs to ground magnetic Pc5 ULF oscillations and analyze how the characteristics of HFAs influence these oscillations, based on the coordinated observations between the THEMIS probes and the ground magnetometers at high latitudes during the years 2008, 2009 and 2019. We find that HFAs can serve as a notable source of ground magnetic Pc5 ULF oscillations, with about 18.9% of Interplanetary Magnetic Field (IMF) discontinuity-induced HFAs associated with discernible enhancements in Pc5 ULF wave power, whereas spontaneous HFAs play a comparatively minor role in generating these oscillations. Furthermore, we observe that the cores of HFAs are likely to contribute more significantly to modulating the induced ground magnetic Pc5 ULF oscillations than their compressed boundaries. More dynamic pressure reductions within HFA cores correspond to stronger ground magnetic Pc5 ULF oscillations. Additionally, HFAs can propagate with the IMF discontinuity along the bow shock, continuously generating ground magnetic Pc5 ULF oscillations during their propagation. This research sheds light on the mechanisms underlying Pc5 ULF wave generation and underscores the role of HFAs in driving magnetospheric-ionospheric interactions.
Objective identification of meteorological fronts and climatologies from ERA-Interim and ERA5
Philip G. Sansom and Jennifer L. Catto
Geosci. Model Dev., 17, 6137–6151, https://doi.org/10.5194/gmd-17-6137-2024, 2024
Weather fronts bring a lot of rain and strong winds to many regions of the mid-latitudes. We have developed an updated method of identifying these fronts in gridded data that can be used on new datasets with small grid spacing. The method can be easily applied to different datasets due to the use of open-source software for its development and shows improvements over similar previous methods. We present an updated estimate of the average frequency of fronts over the past 40 years.
ShellSet v1.1.0 parallel dynamic neotectonic modelling: a case study using Earth5-049
Jon B. May, Peter Bird, and Michele M. C. Carafa
Geosci. Model Dev., 17, 6153–6171, https://doi.org/10.5194/gmd-17-6153-2024, 2024
ShellSet is a combination of well-known geoscience software packages. It features a simple user interface and is optimised through the addition of a grid search input option (automatically searching for optimal models within a defined N-dimensional parameter space) and the ability to run multiple models in parallel. We show that for each number of models tested there is a performance benefit to parallel running, while two examples demonstrate a use case by improving an existing global model.
A Bayesian method for predicting background radiation at environmental monitoring stations
Jens Peter K. W. Frankemölle, Johan Camps, Pieter De Meutter, and Johan Meyers
Geosci. Model Dev. Discuss., https//doi.org/10.5194/gmd-2024-137,2024
Preprint under review for GMD (discussion: open, 0 comments)
To detect anomalous radioactivity in the environment, it is paramount that we understand the natural background level. In this work, we propose a statistical model to describe the most likely background level and the associated uncertainty in a network of dose rate detectors. We train, verify and validate the model using real environmental data. Using the model, we show that we can correctly predict the background level in a subset of the detector network during a known `anomalous’ event.
Model calibration and streamflow simulations for the extreme drought event of 2018 on the Rhine River Basin using WRF-Hydro 5.2.0
Andrea L. Campoverde, Uwe Ehret, Patrick Ludwig, and Joaquim G. Pinto
Geosci. Model Dev. Discuss., https//doi.org/10.5194/gmd-2024-134,2024
Preprint under review for GMD (discussion: open, 0 comments)
We looked at how well the model WRF-Hydro performed during the 2018 drought event in the River Rhine basin, even though it is typically used for floods. We used the meteorological ERA5 reanalysis dataset to simulate River Rhine’s streamflow and adjusted the model using parameters and actual discharge measurements. We focused on Lake Constance, a key part of the basin, but found issues with the model’s lake outflow simulation. By removing the lake module, we obtained more accurate results.
Numerical investigations on the modelling of ultrafine particles in SSH-aerosol-v1.3a: size resolution and redistribution
Oscar Jacquot and Karine Sartelet
Geosci. Model Dev. Discuss., https//doi.org/10.5194/gmd-2024-150,2024
Preprint under review for GMD (discussion: open, 0 comments)
As the health impact of ultrafine particles is better understood, modeling the size distribution and the number concentration becomes increasingly important. A new analytic formulation is presented to compute coagulation partition coefficients, allowing to lower down the numerical diffusion associated to the resolution of aerosol dynamics. The significance of this effect is assessed over Greater Paris with a chemistry transport model, using different size resolution of the particle distribution.
Abstract
How forests respond to accelerated climate change will affect the terrestrial carbon cycle. To better understand these responses, more examples are needed to assess how tree growth rates react to abrupt changes in growing-season temperatures. Here we use a natural experiment in which a glacier's fluctuations exposed a temperate rainforest to changes in summer temperatures of similar magnitude to those predicted to occur by 2050. We hypothesized that the onset of glacier-accentuated temperature trends would act to increase the variance in stand-level tree growth rates, a proxy for forest net primary productivity. Instead, dendrochronological records reveal that the growth rates of five, co-occurring conifer species became less synchronous, and this diversification of species responses acted to reduce the variance and to increase the stability of community-wide growth rates. These results warrant further inquiry into how climate-induced changes in tree-growth diversity may help stabilize future ecosystem services like forest carbon storage.
The Far-INfrarEd Spectrometer for Surface Emissivity (FINESSE) – Part 1: Instrument description and level 1 radiances
Jonathan E. Murray, Laura Warwick, Helen Brindley, Alan Last, Patrick Quigley, Andy Rochester, Alexander Dewar, and Daniel Cummins
Atmos. Meas. Tech., 17, 4757–4775, https://doi.org/10.5194/amt-17-4757-2024, 2024
The Far INfrarEd Spectrometer for Surface Emissivity, FINESSE, is designed to measure the ability of natural surfaces to emit infrared radiation. FINESSE combines a commercial instrument with custom-built optics to view a surface from different angles with complementary views of the sky. Its choice of internal components means it can cover a wide range of wavelengths, extending into the far-infrared. We characterize FINESSE’s uncertainty budget and provide examples of its measurement capability.
The Far-INfrarEd Spectrometer for Surface Emissivity (FINESSE) – Part 2: First measurements of the emissivity of water in the far-infrared
Laura Warwick, Jonathan E. Murray, and Helen Brindley
Atmos. Meas. Tech., 17, 4777–4787, https://doi.org/10.5194/amt-17-4777-2024, 2024
We describe a method for measuring the emissivity of natural surfaces using data from the new Far-INfrarEd Spectrometer for Surface Emissivity (FINESSE) instrument. We demonstrate our method by making measurements of the emissivity of water. We then compare our results to the emissivity predicted using a model and find good agreement. The observations from FINESSE are novel because they allow us to determine surface emissivity at longer wavelengths than have been routinely measured before.
Abstract
The precipitation of magnesium oxide (MgO) from the Earth's core has been proposed as a potential energy source to power the geodynamo prior to the inner core solidification. Yet, the stable phase and exact amount of MgO exsolution remain elusive. Here we utilize an iterative learning scheme to develop a unified deep learning interatomic potential for the Mg-Fe-O system valid over a wide pressure-temperature range. This potential enables direct, large-scale simulations of MgO exsolution processes at the Earth's core-mantle boundary. Our results suggest that Mg exsolve in the form of crystalline Fe-poor ferropericlase as opposed to a liquid MgO component presumed previously. The solubility of Mg in the core is limited, and the present-day core is nearly Mg-free. The resulting exsolution rate is small yet nonnegligible, suggesting that MgO exsolution may provide a potentially important energy source, although it alone may be difficult to drive an early geodynamo.
Brief Communication: Rapid high-resolution flood impact-based early warning is possible with RIM2D: a showcase for the 2023 pluvial flood in Braunschweig
Shahin Khosh Bin Ghomash, Heiko Apel, Kai Schröter, and Max Steinhausen
Nat. Hazards Earth Syst. Sci. Discuss., https//doi.org/10.5194/nhess-2024-139,2024
Preprint under review for NHESS (discussion: open, 0 comments)
This work introduces RIM2D, a hydrodynamic model for precise and rapid flood predictions, ideal for early warning systems. We demonstrate RIM2D's ability to deliver detailed and localized flood forecasts using the June 2023 flood in Braunschweig, Germany, as a case study. This research highlights the readiness of RIM2D and the required hardware for integration into operational flood warning and impact-based forecasting systems.
Development of operational decision support tools for mechanized ski guiding using avalanche terrain modelling, GPS tracking, and machine learning
John Sykes, Pascal Haegeli, Roger Atkins, Patrick Mair, and Yves Bühler
Nat. Hazards Earth Syst. Sci. Discuss., https//doi.org/10.5194/nhess-2024-147,2024
Preprint under review for NHESS (discussion: open, 0 comments)
We develop decision support tools to assist professional ski guides in determining safe terrain each day based on current conditions. To understand the decision-making process we collaborate with professional guides and build three unique models to predict their decisions. The models accurately capture the real world decision-making outcomes in 85–93 % of cases. Our conclusions focus on strengths and weaknesses of each model and discuss ramifications for practical applications in ski guiding.
Statistical and neural network assessment of climatological features of fog and mist at Pula airport in Croatia: from local to synoptic scale
Marko Zoldoš, Tomislav Džoić, Jadran Jurković, Frano Matić, Sandra Jambrošić, Ivan Ljuština, and Maja Telišman Prtenjak
Nonlin. Processes Geophys. Discuss., https//doi.org/10.5194/npg-2024-18,2024
Preprint under review for NPG (discussion: open, 0 comments)
Fog can disrupt aviation by causing accidents and delays due to low visibility, yet it remains under-researched in Croatia. This study examined fog and mist at Pula Airport using 20 years of data and machine learning techniques. There is a declining trend in fog, linked to changing weather patterns. Fog mainly occurs from October to March. These findings enhance knowledge about fog in Croatia and can improve weather forecasts, increasing safety at the airport.
Abstract
When charged particles are accelerated from Earth's magnetosphere and precipitate into the atmosphere, their impact with neutral gas creates the aurora. Structured electric fields drive the acceleration processes but they are also passed down to the ionosphere, meaning that turbulence can in part be embedded into the ionosphere rather than emerge through instability processes locally. Applying a point-cloud analysis technique adapted from observational cosmology, we show how observed turbulence in the ionosphere matches electrical current signatures in the pulsating aurora in a series of conjunctions between space- and ground-based instruments. We propose that the temporal spectrum of pulsations in the pulsating aurora is the driver of a clearly observed energy injection into the ionosphere's unstable bottomside. Precipitating electrons produce electric fields through charge deposition, and we observe wave characteristics that are present in this pattern. Next, the relative electron-ion drifts excite the Farley-Buneman instability, the distribution of whose waves are organized according to the local electric field. It is the temporal characteristics of chorus wave interactions in the magnetosphere that is imparted, via precipitating electrons, to the pulsating aurora, and so we propose that chorus wave interactions are capable of embedding turbulent structure into the ionosphere. This structure (now pressure gradients) dissipate energy in the E-region through turbulent processes, observed by the icebear coherent scatter radar.
Abstract
The frequency change of 100-year flood events is often determined by fitting extreme value distributions to annual maximum discharge from a historical base period. This study demonstrates that this approach may significantly bias the computed flood frequency change. An idealized experiment shows frequency bias exceeding 100% for a 50-year base period. Further analyses using Monte Carlo simulations, mathematical derivations, and hydrological model outputs reveal that bias magnitude inversely relates to base period length and is weakly influenced by the generalized extreme value distribution's shape parameter. The bias, persisting across different estimation methods, implies floods may exceed local defenses designed based on short historical records more often than expected, even without climate change. We introduce a frequency bias adjustment method, which significantly reduces the projected rise in global flood occurrence. This suggests a substantial part of the earlier projected increase in flood occurrence and impacts is not attributable to climate change.
