Common Modeling Errors Research Articles

Assessment of Warm and Dry Bias over ARM SGP Site in E3SMv2 and E3SM-MMF

Abstract Many climate models exhibit a dry and warm bias over the central United States during the summer months, including the Energy Exascale Earth System Model (E3SM) and its Multiscale Modeling Framework (MMF) configuration. Understanding the causes of this bias is important to shine a light on this common model error and reduce the uncertainty in future projections. In this study, we use E3SMv2 and E3SM-MMF to assess how parameterized and resolved convection affect temperature and precipitation biases over the Southern Great Plains site of the Atmospheric Radiation Measurement program. Both configurations overestimate near-surface temperature and underestimate precipitation at the ARM SGP site. The bias is associated with a lack of low-level clouds during days without precipitation and too much incoming solar radiation causing the surface to warm. Low-level cloud fraction in E3SM-MMF during the nonprecipitating days is lower in comparison to E3SMv2 and observation, consistent with the larger warm bias. We also find that the underestimated precipitation can be characterized as “too frequent, too weak” in E3SMv2 and “too rare, too intense” in E3SM-MMF. These deficiencies conspire to sustain the warm and dry bias over the central United States.

Study on systematic errors of BDS-3 broadcast ephemeris and their effects with Helmert transformation

Previous studies have not evaluated the systematic errors implied in the third generation of BeiDou-3 Navigation Satellite System (BDS-3) broadcast ephemeris. In this paper we evaluate the systematic pattern described by the Helmert transformation parameters, including translations, rotations, and scale. BDS-3 broadcast and precise ephemerides from December 2019 to 2022 are collected, and the characteristics of the transformation parameters as well as their effects on the signal in space error are analysed. The annual variation in the z-translation is obtained, and the similar amplitudes of 5.5 cm and phases of approximate 300 days are obtained for different years. When the rotation parameters are considered in the orbit comparison, the Root Mean Square (RMS) errors of the along- and cross-track orbital differences decrease from 29.1 to 12.5 cm and from 30.6 to 9.2 cm, respectively, because the three rotation parameters compensate for the majority of the errors in the BDS-3 broadcast ephemeris. Moreover, the high correlations in the obtained rotation parameters among the three orbital planes suggest that the orientation of the BDS-3 broadcast ephemeris is influenced by common model errors, i.e., uncertainty of Earth Rotation Parameters (ERPs). Further research is required because an offset of 1.5 × 10–9 for the scale parameter is observed. A degraded User Range Error (URE) for epochs of up to 84% is attained when the systematic pattern is considered, though the impact of the systematic pattern indicated by the z-translation and rotation parameters on the URE is less than 5.0 cm. With the refinement of the ERPs implemented in the new generation of broadcast ephemeris, we anticipate that the broadcast ephemeris performance of BDS-3 will be improved.

Rapid Development of Systematic ENSO‐Related Seasonal Forecast Errors

AbstractClimate models exhibit known systematic errors in their representation of the El Niño‐Southern Oscillation (ENSO). In this study, we show that such simulation errors are largely present in tropical seasonal prediction, even for short lead times. Regressing monthly forecast errors from 11 different operational models upon the observed ENSO state, we find that predicted ENSO‐related sea surface temperature anomalies (of either sign) for winter/spring are significantly extended or shifted to the west and are also too persistent during the ENSO decay phase, both common climate model errors. There are also corresponding precipitation forecast errors, most notably a robust westward shift of the ENSO‐related precipitation dipole that may impact predictions of extratropical teleconnections. These ENSO‐related errors develop within days after initialization regardless of month, including significant errors appearing in anomalous surface trade winds, and saturate so rapidly that they primarily depend upon the seasonal cycle rather than lead time.

Analysis of the common model error on velocity field under Colored noise model by GPS and InSAR: A case study in the Nepal and everest region

The accuracy of the velocity field will be affected by the noise model and common mode errors through GPS time series analysis. In order to analyze the influence of these two factors on the accuracy of the velocity field, two kinds of data are used, including the three-year observation from 20 permanent GPS stations with high spatial correlation in the Everest, which is about 650 km from north to south and 1068 km from east to west, and three-year 80 ascending images and 141 descending images from sentinel-1A, which are processed by GAMIT/GLOBK software and Small Baseline Subset-Interferometric Synthetic Aperture Radar method (SBAS-InSAR), respectively. The vertical deformation rate is solved by time series analysis through a self-made adaptive algorithm. In the analysis, the linear change rate, period, half period coefficient, and residual sequence of all stations are solved by using James L. Davis periodic model. The noise type of residual sequence is analyzed by the power spectrum model. The spatio-temporal correlated noise, Common Mode Error (CME), is extracted by the Principal Component Analysis (PCA) and Karhunen-Loeve (KLE) methods. The results show that noises can be best described by “flicker noise + white noise” model. After the removal of CME, the R2 estimates of all stations are above 0.8, with RMS value of velocity field decreasing from 1.428 mm/yr to 1.062 mm/yr and 1.063 mm/yr to 0.815 mm/yr, in N and E directions, respectively, indicating that the influence of CME can't be ignored in the extraction of the high-precision velocity field in the Nepal and Everest region.

Using a Cost-Effective Approach to Increase Background Ensemble Member Size within the GSI-Based EnVar System for Improved Radar Analyses and Forecasts of Convective Systems

Abstract A valid time shifting (VTS) method is explored for the GSI-based ensemble variational (EnVar) system modified to directly assimilate radar reflectivity at convective scales. VTS is a cost-efficient method to increase ensemble size by including subensembles before and after the central analysis time. Additionally, VTS addresses common time and phase model error uncertainties within the ensemble. VTS is examined here for assimilating radar reflectivity in a continuous hourly analysis system for a case study of 1–2 May 2019. The VTS implementation is compared against a 36-member control experiment (ENS-36), to increase ensemble size (3 × 36 VTS), and as a cost-savings method (3 × 12 VTS), with time-shifting intervals τ between 15 and 120 min. The 3 × 36 VTS experiments increased the ensemble spread, with largest subjective benefits in early cycle analyses during convective development. The 3 × 12 VTS experiments captured analysis with similar accuracy as ENS-36 by the third hourly analysis. Control forecasts launched from hourly EnVar analyses show significant skill increases in 1-h precipitation over ENS-36 out to hour 12 for 3 × 36 VTS experiments, subjectively attributable to more accurate placement of the convective line. For 3 × 12 VTS, experiments with τ ≥ 60 min met and exceeded the skill of ENS-36 out to forecast hour 15, with VTS-3 × 12τ90 maximizing skill. Sensitivity results demonstrate preference to τ = 30–60 min for 3 × 36 VTS and 60–120 min for 3 × 12 VTS. The best 3 × 36 VTS experiments add a computational cost of 45%–67%, compared to the near tripling of costs when directly increasing ensemble size, while best 3 × 12 VTS experiments save about 24%–41% costs over ENS-36. Significance Statement The purpose of this work is to study a valid time shifting method to improve the prediction of severe convective storm systems over the continental United States. This method improves ensemble-based radar reflectivity analyses by including ensemble member information at times before and after the analysis time, thereby increasing the ensemble size at just a fractional added computational cost. The results show the method can boost the accuracy of high-resolution convection prediction out to at least 12 h. This case study motivates future systematic testing in a real-time setting and potential implementation to enhance the U.S. operational ensemble-based convection-allowing forecast model and data assimilation system.

GNSS-TS-NRS: An Open-Source MATLAB-Based GNSS Time Series Noise Reduction Software

The global navigation satellite system (GNSS) has seen tremendous advances in measurement precision and accuracy, and it allows researchers to perform geodynamics and geophysics studies through the analysis of GNSS time series. Moreover, GNSS time series not only contain geophysical signals, but also unmodeled errors and other nuisance parameters, which affect the performance in the estimation of site coordinates and related parameters. As the number of globally distributed GNSS reference stations increases, GNSS time series analysis software should be developed with more flexible format support, better human–machine interaction, and with powerful noise reduction analysis. To meet this requirement, a new software named GNSS time series noise reduction software (GNSS-TS-NRS) was written in MATLAB and was developed. GNSS-TS-NRS allows users to perform noise reduction analysis and spatial filtering on common mode errors and to visualize GNSS position time series. The functions’ related theoretical background of GNSS-TS-NRS were introduced. Firstly, we showed the theoretical background algorithms of the noise reduction analysis (empirical mode decomposition (EMD), ensemble empirical mode decomposition (EEMD)). We also developed three improved algorithms based on EMD for noise reduction, and the results of the test example showed our proposed methods with better effect. Secondly, the spatial filtering model supported five algorithms on a separate common model error: The stacking filter method, weighted stacking filter method, correlation weighted superposition filtering method, distance weighted filtering method, and principal component analysis, as well as with batch processing. Finally, the developed software also enabled other functions, including outlier detection, correlation coefficient calculation, spectrum analysis, and distribution estimation. The main goal of the manuscript is to share the software with the scientific community to introduce new users to the GNSS time series noise reduction and application.

GNSS time offset monitoring based on the single difference among systems

Time offset monitoring of multiple Global Navigation Satellite Systems is important for compatibility and interoperability. Unlike conventional signal space approaches, such as single-point positioning (SPP) and precise point positioning, an approach based on the principle of single difference among systems, single-difference point positioning (SDPP), is proposed to omit the receiver clock parameter and reduce the influence of the common model errors during estimation. The proposed approach was validated on different datasets, and results shown that the estimated time offset among systems using SDPP has no significant difference with that using SPP. Moreover, SDPP retrieves the best time offset estimation when considering it as a random walk process, as it provides smooth estimation with low noise. Furthermore, considering constraints from station coordinates and tropospheric delay further improves time offset estimation. The time offset estimation over time remains within a reasonable range, exhibits some periodicity, and estimation from different stations is consistent if disregarding system biases. In addition, the estimated time offset can be used to improve the SPP accuracy.

GPS and BDS combined PPP model with inter-system differenced observations

This study proposes a combined precise point positioning (PPP) model, called the inter-system differenced PPP model, in which it formed the satellite-differenced observation between different satellite systems. Compared with the traditional combined PPP model, the inter-system differenced PPP model has the following characteristics: (1) The satellite difference between various systems can eliminate the receiver clock bias parameter, it enhances the strength of the PPP model. (2) Inter-satellite differences can eliminate some common model errors and reduce the impact of observation errors. (3) The system time difference can be calculated directly to provide the basis for time offset monitoring. In order to verify the positioning results of the inter-system differenced model, static PPP and dynamic PPP experiments were carried out to test the positioning accuracy and convergence time, and the results were compared with those of the traditional combination PPP model. In addition, the offset characteristics introduced by the inter-system differenced model were analysed, and the optimal estimation strategy was determined by dynamic PPP tests. Preliminary results are as follows: (1) For inter-system differenced static PPP, the convergence time of stations is less than 20 min. The standard deviation (STD) of the position bias for the East (E), North (N), and Up (U) components are better than 2.5 cm, 1.0 cm and 3.0 cm, respectively, and the corresponding root mean square deviation (RMS) are better than 3.0 cm, 1.5 cm and 6.0 cm, respectively. Compared with the traditional combined PPP model, the average convergence time of the inter-system differenced model is nearly the same, but its overall positioning accuracy is better. (2) For the inter-system differenced dynamic PPP model, the convergence time of stations is better than 30 min. The STD of the position bias for the E, N, and U components is better than 3.0 cm, 2.0 cm and 4.5 cm respectively, and the RMS is better than 4.0 cm, 2.5 cm and 7.0 cm respectively. Compared with the traditional combined PPP model, the convergence of the inter-system differenced model is slightly better, and the positioning accuracy does not differ significantly. Moreover, the inter-system differenced model is much better than the traditional model when fewer satellites are available. (3) The offset between GPS and BDS corresponding with the precise products of GFZ is related to the type of receiver, and the daily standard of offset is less than 0.5 ns. In addition, we determined optimum process noise of the offset parameter to be a 10-3/3(m2/s) random walk after comparing several options.

Bridge Real-Time Damage Identification Method Using Inclination and Strain Measurements in the Presence of Temperature Variation

In this paper, a new real-time damage identification method has been presented for bridge structural health monitoring (SHM) considering temperature variation. The method utilizes model-based damage identification that involves three major steps: (1) efficient basis functions—extracted from finite-element (FE) models prior to real-time identification; (2) partial least-squares regression (PLSR) analyses; and (3) the fusion of different types of structural responses into damage indicator. By treating local damages as equivalent vertical loads and then cross-referencing global (inclinations) and local (strain) data, the hidden damage information in bridge structures can be detected and localized in a timely fashion, even in the presence of unknown temperature variation as well as vehicle loads. Inclinations alone cannot reflect local damages, but by fusing inclinations and strains (that represent local damage) into the proposed damage indicator, local damages can be identified. Numerical simulations on a medium-span continuous bridge demonstrate that the proposed method is insensitive to measurement noise and some common modeling errors, revealing the potential of real-time damage identification in bridge SHM applications.

Evaluation of SRE Scenarios for Penang, Selangor and Johor in Peninsular Malaysia using PRECIS Regional Climate Model (RCM)

Climate change is unambiguous as there is much evidence from around the world showing that changes have already occurred. This phenomenon is in response to an array of human activities, notably the release of greenhouse gases; an understanding of the rate, mode and scale of this change is now of literally vital importance to society. Researchers utilize climate models to study the dynamics of our changing climate and also to make future projections. Climate models are basic representation of many interactions within the Earth’s climate which includes the atmosphere, land surface, oceans and ice. These models are typically quantitative in nature and range from simple depictions of the climate to very complex ones. In this present study, downscaled PRECIS regional climate models (RCMs) were used to project the average minimum and average maximum temperatures and average precipitation for Penang, Selangor and Johor in Peninsular Malaysia. The RCM projections for these three states were developed based on ECHAM4 A2 and ECHAM5 A1B scenarios for the years 1980 to 2069 and ECHAM4 B2 scenario for the years 2010 to 2069. Bias correction will be applied to the simulated historical data to remove common systematic model errors. Historical observation data of monthly average minimum and maximum temperatures and monthly average rainfall from the Malaysian Meteorological Department (MMD) will be used in the bias correction. Finally, a RCM scenario which matches with the historical observation data of the three states for future projections will be recommended.

Unifying error structures in commonly used biotracer mixing models.

Mixing models are statistical tools that use biotracers to probabilistically estimate the contribution of multiple sources to a mixture. These biotracers may include contaminants, fatty acids, or stable isotopes, the latter of which are widely used in trophic ecology to estimate the mixed diet of consumers. Bayesian implementations of mixing models using stable isotopes (e.g., MixSIR, SIAR) are regularly used by ecologists for this purpose, but basic questions remain about when each is most appropriate. In this study, we describe the structural differences between common mixing model error formulations in terms of their assumptions about the predation process. We then introduce a new parameterization that unifies these mixing model error structures, as well as implicitly estimates the rate at which consumers sample from source populations (i.e., consumption rate). Using simulations and previously published mixing modeldatasets, we demonstrate that the new error parameterization outperforms existing models and provides an estimate of consumption. Our results suggest that the error structure introduced here will improve future mixing model estimates of animal diet.

How Predictable is the Indian Ocean Dipole?

Abstract In light of the growing recognition of the role of surface temperature variations in the Indian Ocean for driving global climate variability, the predictive skill of the sea surface temperature (SST) anomalies associated with the Indian Ocean dipole (IOD) is assessed using ensemble seasonal forecasts from a selection of contemporary coupled climate models that are routinely used to make seasonal climate predictions. The authors assess predictions from successive versions of the Australian Bureau of Meteorology Predictive Ocean–Atmosphere Model for Australia (POAMA 15b and 24), successive versions of the NCEP Climate Forecast System (CFSv1 and CFSv2), the ECMWF seasonal forecast System 3 (ECSys3), and the Frontier Research Centre for Global Change system (SINTEX-F) using seasonal hindcasts initialized each month from January 1982 to December 2006. The lead time for skillful prediction of SST in the western Indian Ocean is found to be about 5–6 months while in the eastern Indian Ocean it is only 3–4 months when all start months are considered. For the IOD events, which have maximum amplitude in the September–November (SON) season, skillful prediction is also limited to a lead time of about one season, although skillful prediction of large IOD events can be longer than this, perhaps up to about two seasons. However, the tendency for the models to overpredict the occurrence of large events limits the confidence of the predictions of these large events. Some common model errors, including a poor representation of the relationship between El Niño and the IOD, are identified indicating that the upper limit of predictive skill of the IOD has not been achieved.

XMAS: Quick Formal Modeling of Communication Fabrics to Enable Verification

Although communication fabrics at the microarchitectural level are mainly composed of standard primitives such as queues and arbiters, to get an executable model one has to connect these primitives with glue logic to complete the description. In this paper we identify a richer set of microarchitectural primitives that allows us to describe complete systems by composition alone. This enables us to build models faster (since models are now simply wiring diagrams at an appropriate level of abstraction) and to avoid common modeling errors such as inadvertent loss of data due to incorrect timing assumptions. Our models are formal and they are used for model checking as well as dynamic validation and performance modeling. However, unlike other formalisms this approach leads to a precise yet intuitive graphical notation for microarchitecture that captures timing and functionality in sufficient detail to be useful for reasoning about correctness and for communicating microarchitectural ideas to RTL and circuit designers and validators.

Surface energy balance errors in AGCMs: Implications for ocean‐atmosphere model coupling

The oceanic poleward heat transport (TO) implied by an atmospheric General Circulation Model (GCM) can help evaluate a model's readiness for coupling with an ocean GCM. A well known problem is that the sub‐tropical Southern Hemisphere TO implied by many models is equatorward, contrary to most observationally‐based estimates. By correcting the TO of a diversity of models with satellite‐derived observations, an earlier study demonstrated that the dominant TO problems could be explained by model errors in cloud radiative effects. Such errors were argued to be primarily responsible for the erroneous Southern Hemisphere TO. However, systematic evaluation of one model in a later study suggested that the equatorward Southern Hemisphere TO might also be attributable to biases in the ocean surface latent heat flux. In this study we revisit the problem with a more recent suite of simulations, and demonstrate that collectively models have improved, but only slightly. We then use ocean surface fluxes estimates to examine the problem from a surface point of view. Our results clarify that common model errors in the surface latent heat flux can afflict TO as dramatically as errors in cloud radiative effects. To illustrate the relative importance of the two errors, diagnostic tests are introduced that should become increasingly useful as estimates of ocean surface fluxes are improved.

Evaluation of water and energy budgets in regional climate models applied over Europe

This study presents a model intercomparison of four regional climate models (RCMs) and one variable resolution atmospheric general circulation model (AGCM) applied over Europe with special focus on the hydrological cycle and the surface energy budget. The models simulated the 15 years from 1979 to 1993 by using quasi-observed boundary conditions derived from ECMWF re-analyses (ERA). The model intercomparison focuses on two large atchments representing two different climate conditions covering two areas of major research interest within Europe. The first is the Danube catchment which represents a continental climate dominated by advection from the surrounding land areas. It is used to analyse the common model error of a too dry and too warm simulation of the summertime climate of southeastern Europe. This summer warming and drying problem is seen in many RCMs, and to a less extent in GCMs. The second area is the Baltic Sea catchment which represents maritime climate dominated by advection from the ocean and from the Baltic Sea. This catchment is a research area of many studies within Europe and also covered by the BALTEX program. The observed data used are monthly mean surface air temperature, precipitation and river discharge. For all models, these are used to estimate mean monthly biases of all components of the hydrological cycle over land. In addition, the mean monthly deviations of the surface energy fluxes from ERA data are computed. Atmospheric moisture fluxes from ERA are compared with those of one model to provide an independent estimate of the convergence bias derived from the observed data. These help to add weight to some of the inferred estimates and explain some of the discrepancies between them. An evaluation of these biases and deviations suggests possible sources of error in each of the models. For the Danube catchment, systematic errors in the dynamics cause the prominent summer drying problem for three of the RCMs, while for the fourth RCM this is related to deficiencies in the land surface parametrization. The AGCM does not show this drying problem. For the Baltic Sea catchment, all models similarily overestimate the precipitation throughout the year except during the summer. This model deficit is probably caused by the internal model parametrizations, such as the large-scale condensation and the convection schemes.

Feedforward Control under the Presence of Uncertainty*

In this paper, we study the effect of model errors on the performance of feedforward controllers. In accordance with the sensitivity function for feedback control, we define the feedforward sensitivities, Sff(feedforward from disturbance) and Sff,r (feedforward from set-point), as measures for the reduction in the output error obtained by the feedforward control. For “ideal” feedforward controllers based on the inverted nominal model, the feedforward sensitivities equal the relative model errors, which must thus remain less than 1 for feedforward control to have a positive (dampening) effect. For some common model error classes we provide rules for when the feedforward controller is effective, and we also design μ-optimal feedforward controllers.

General asymptotic analysis of the generalized likelihood ratio test for a Gaussian point source under statistical or spatial mismodeling

This paper investigates the robustness of the generalized likelihood ratio test (GLRT) for a far-field Gaussian point source. Given measurements from an array of sensors, the performance of the GLRT under two types of common modeling errors is investigated. The first type is spatial mismodeling, which relates to errors due to multipath effects or errors in the assumed number of sources, i.e., deviation from the single point source assumption. The second type is statistical mismodeling, which relates to errors due to non-Gaussianity in either the noise or the signal, i.e., deviation from the Gaussian assumption. It is shown that for some types of modeling errors, the detector's performance improves, and general conditions for such an improvement are found. Moreover, for both types of errors, the change in performance is analyzed and quantified. This analysis shows that for a distributed source with small spatial spreading, the degradation in performance is significant, whereas for a constant modulus point source, the performance improves. Simulations of various cases are shown to verify the analytical results.

A Computational Model for the Stereoscopic Optics of a Head-Mounted Display

For stereoscopic photography or telepresence, orthostereoscopy occurs when the perceived size, shape, and relative position of objects in the three-dimensional scene being viewed match those of the physical objects in front of the camera. In virtual reality, the simulated scene has no physical counterpart, so orthostereoscopy must be defined in this case as constancy, as the head moves around, of the perceived size, shape, and relative positions of the simulated objects. Achieving this constancy requires that the computational model used to generate the graphics matches the physical geometry of the head-mounted display being used. This geometry includes the optics used to image the displays and the placement of the displays with respect to the eyes. The model may fail to match the geometry because model parameters are difficult to measure accurately, or because the model itself is in error. Two common modeling errors are ignoring the distortion caused by the optics and ignoring the variation in interpupillary distance across different users. A computational model for the geometry of a head-mounted display is presented, and the parameters of this model for the VPL EyePhone are calculated.