The EXospheric TEMperatures on a PoLyhedrAl gRid (EXTEMPLAR) method predicts the neutral densities in the thermosphere. The performance of this model has been evaluated through a comparison with the Air Force High Accuracy Satellite Drag Model (HASDM). The Space Environment Technologies (SET) HASDM database that was used for this test spans the 20 years 2000 through 2019, containing densities at 3 hr time intervals at 25 km altitude steps, and a spatial resolution of 10° latitude by 15° longitude. The upgraded EXTEMPLAR that was tested uses the newer Naval Research Laboratory MSIS 2.0 model to convert global exospheric temperature values to neutral density as a function of altitude. The revision also incorporated time delays that varied as a function of location, between the total Poynting flux in the polar regions and the exospheric temperature response. The density values from both models were integrated on spherical shells at altitudes ranging from 200 to 800 km. These sums were compared as a function of time. The results show an excellent agreement at temporal scales ranging from hours to years. The EXTEMPLAR model performs best at altitudes of 400 km and above, where geomagnetic storms produce the largest relative changes in neutral density. In addition to providing an effective method to compare models that have very different spatial resolutions, the use of density totals at various altitudes presents a useful illustration of how the thermosphere behaves at different altitudes, on time scales ranging from hours to complete solar cycles.
The Mass Spectrometer and Incoherent Scatter radar (MSIS) model family has been developed and improved since the early 1970's. The most recent version of MSIS is the Naval Research Laboratory (NRL) MSIS 2.0 empirical atmospheric model. NRLMSIS 2.0 provides species density, mass density, and temperature estimates as function of location and space weather conditions. MSIS models have long been a popular choice of thermosphere model in the research and operations community alike, but—like many models—does not provide uncertainty estimates. In this work, we develop an exospheric temperature model based in machine learning that can be used with NRLMSIS 2.0 to calibrate it relative to high‐fidelity satellite density estimates directly through the exospheric temperature parameter. Instead of providing point estimates, our model (called MSIS‐UQ) outputs a distribution which is assessed using a metric called the calibration error score. We show that MSIS‐UQ debiases NRLMSIS 2.0 resulting in reduced differences between model and satellite density of 25% and is 11% closer to satellite density than the Space Force's High Accuracy Satellite Drag Model. We also show the model's uncertainty estimation capabilities by generating altitude profiles for species density, mass density, and temperature. This explicitly demonstrates how exospheric temperature probabilities affect density and temperature profiles within NRLMSIS 2.0. Another study displays improved post‐storm overcooling capabilities relative to NRLMSIS 2.0 alone, enhancing the phenomena that it can capture.
more » « less- PAR ID:
- 10390072
- Publisher / Repository:
- DOI PREFIX: 10.1029
- Date Published:
- Journal Name:
- Space Weather
- Volume:
- 20
- Issue:
- 11
- ISSN:
- 1542-7390
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
more » « less
Abstract -
more » « less
Abstract NRLMSIS® 2.0 is an empirical atmospheric model that extends from the ground to the exobase and describes the average observed behavior of temperature, eight species densities, and mass density via a parametric analytic formulation. The model inputs are location, day of year, time of day, solar activity, and geomagnetic activity. NRLMSIS 2.0 is a major, reformulated upgrade of the previous version, NRLMSISE‐00. The model now couples thermospheric species densities to the entire column, via an effective mass profile that transitions each species from the fully mixed region below ~70 km altitude to the diffusively separated region above ~200 km. Other changes include the extension of atomic oxygen down to 50 km and the use of geopotential height as the internal vertical coordinate. We assimilated extensive new lower and middle atmosphere temperature, O, and H data, along with global average thermospheric mass density derived from satellite orbits, and we validated the model against independent samples of these data. In the mesosphere and below, residual biases and standard deviations are considerably lower than NRLMSISE‐00. The new model is warmer in the upper troposphere and cooler in the stratosphere and mesosphere. In the thermosphere, N2and O densities are lower in NRLMSIS 2.0; otherwise, the NRLMSISE‐00 thermosphere is largely retained. Future advances in thermospheric specification will likely require new in situ mass spectrometer measurements, new techniques for species density measurement between 100 and 200 km, and the reconciliation of systematic biases among thermospheric temperature and composition data sets, including biases attributable to long‐term changes.
-
more » « less
Abstract Machine learning (ML) models are universal function approximators and—if used correctly—can summarize the information content of observational data sets in a functional form for scientific and engineering applications. A benefit to ML over parametric models is that there are no a priori assumptions about particular basis functions which can potentially limit the phenomena that can be modeled. In this work, we develop ML models on three data sets: the Space Environment Technologies High Accuracy Satellite Drag Model (HASDM) density database, a spatiotemporally matched data set of outputs from the Jacchia‐Bowman 2008 Empirical Thermospheric Density Model (JB2008), and an accelerometer‐derived density data set from CHAllenging Minisatellite Payload (CHAMP). These ML models are compared to the Naval Research Laboratory Mass Spectrometer and Incoherent Scatter radar (NRLMSIS 2.0) model to study the presence of post‐storm cooling in the middle‐thermosphere. We find that both NRLMSIS 2.0 and JB2008‐ML do not account for post‐storm cooling and consequently perform poorly in periods following strong geomagnetic storms (e.g., the 2003 Halloween storms). Conversely, HASDM‐ML and CHAMP‐ML do show evidence of post‐storm cooling indicating that this phenomenon is present in the original data sets. Results show that density reductions up to 40% can occur 1–3 days post‐storm depending on the location and strength of the storm.
-
more » « less
Abstract The Starlink satellites launched on 3 February 2022 were lost before they fully arrived in their designated orbits. The loss was attributed to two moderate geomagnetic storms that occurred consecutively on 3–4 February. We investigate the thermospheric neutral mass density variation during these storms with the Multiscale Atmosphere‐Geospace Environment (MAGE) model, a first‐principles, fully coupled geospace model. Simulated neutral density enhancements are validated by Swarm satellite measurements at the altitude of 400–500 km. Comparison with standalone TIEGCM and empirical NRLMSIS 2.0 and DTM‐2013 models suggests better performance by MAGE in predicting the maximum density enhancement and resolving the gradual recovery process. Along the Starlink satellite orbit in the middle thermosphere (∼200 km altitude), MAGE predicts up to 150% density enhancement near the second storm peak while standalone TIEGCM, NRLMSIS 2.0, and DTM‐2013 suggest only ∼50% increase. MAGE also suggests altitudinal, longitudinal, and latitudinal variability of storm‐time percentage density enhancement due to height dependent Joule heating deposition per unit mass, thermospheric circulation changes, and traveling atmospheric disturbances. This study demonstrates that a moderate storm can cause substantial density enhancement in the middle thermosphere. Thermospheric mass density strongly depends on the strength, timing, and location of high‐latitude energy input, which cannot be fully reproduced with empirical models. A physics‐based, fully coupled geospace model that can accurately resolve the high‐latitude energy input and its variability is critical to modeling the dynamic response of thermospheric neutral density during storm time.
-
more » « less
Abstract Neural networks (NN) have become an important tool for prediction tasks—both regression and classification—in environmental science. Since many environmental-science problems involve life-or-death decisions and policy making, it is crucial to provide not only predictions but also an estimate of the uncertainty in the predictions. Until recently, very few tools were available to provide uncertainty quantification (UQ) for NN predictions. However, in recent years the computer-science field has developed numerous UQ approaches, and several research groups are exploring how to apply these approaches in environmental science. We provide an accessible introduction to six of these UQ approaches, then focus on tools for the next step, namely, to answer the question:
Once we obtain an uncertainty estimate (using any approach), how do we know whether it is good or bad? To answer this question, we highlight four evaluation graphics and eight evaluation scores that are well suited for evaluating and comparing uncertainty estimates (NN based or otherwise) for environmental-science applications. We demonstrate the UQ approaches and UQ-evaluation methods for two real-world problems: 1) estimating vertical profiles of atmospheric dewpoint (a regression task) and 2) predicting convection over Taiwan based onHimawari-8 satellite imagery (a classification task). We also provide Jupyter notebooks with Python code for implementing the UQ approaches and UQ-evaluation methods discussed herein. This article provides the environmental-science community with the knowledge and tools to start incorporating the large number of emerging UQ methods into their research.Significance Statement Neural networks are used for many environmental-science applications, some involving life-or-death decision-making. In recent years new methods have been developed to provide much-needed uncertainty estimates for NN predictions. We seek to accelerate the adoption of these methods in the environmental-science community with an accessible introduction to 1) methods for computing uncertainty estimates in NN predictions and 2) methods for evaluating such estimates.
