Search NASA⌕ Search

SEARCH · Search NASA

Results for “forecast error”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 631 records · Page 35

Forte: An Interactive Visual Analytic Tool for Trust-Augmented Net-Load Forecasting

Accurate net-load forecasting is vital for energy planning, aiding decisions on trade and load distribution. However, assessing the performance of forecasting models across diverse input variables, like temperature and humidity, remains challenging, particularly for eliciting a high degree of trust in the model outcomes. In this context, there is a growing need for data-driven technological interventions to aid scientists in comprehending how models react to both noisy and clean input variables, thus shedding light on complex behaviors and fostering confidence in the outcomes. In this paper, we present Forte, a visual analytics-based application to explore deep probabilistic net-load forecasting models across various input variables and understand the error rates for different scenarios. With carefully designed visual interventions, this web-based interface empowers scientists to derive insights about model performance by simulating diverse scenarios, facilitating an informed decision-making process. We discuss observations made using Forte and demonstrate the effectiveness of visualization techniques to provide valuable insights into the correlation between weather inputs and net-load forecasts, ultimately advancing grid capabilities by improving trust in forecasting models.

Bhattacharjee, Kaustav↗

Analysis of Random Forest Modeling Strategies for Multi-Step Wind Speed Forecasting

Although the random forest (RF) model is a powerful machine learning tool that has been utilized in many wind speed/power forecasting studies, there has been no consensus on optimal RF modeling strategies. This study investigates three basic questions which aim to assist in the discernment and quantification of the effects of individual model properties, namely: (1) using a standalone RF model versus using RF as a correction mechanism for the persistence approach, (2) utilizing a recursive versus direct multi-step forecasting strategy, and (3) training data availability on model forecasting accuracy from one to six hours ahead. These questions are investigated utilizing data from the FINO1 offshore platform and Atmospheric Radiation Measurement (ARM) Southern Great Plains (SGP) C1 site, and testing results are compared to the persistence method. At FINO1, due to the presence of multiple wind farms and high inter-annual variability, RF is more effective as an error-correction mechanism for the persistence approach. The direct forecasting strategy is seen to slightly outperform the recursive strategy, specifically for forecasts three or more steps ahead. Finally, increased data availability (up to ~8 equivalent years of hourly training data) appears to continually improve forecasting accuracy, although changing environmental flow patterns have the potential to negate such improvement. We hope that the findings of this study will assist future researchers and industry professionals to construct accurate, reliable RF models for wind speed forecasting.

54 ENVIRONMENTAL SCIENCES↗

Using Ensemble Short-Term Initialized Coupled NASA GEOS5 Climate Model Integrations to Study Convective Bias Growth

The representation of convective processes, particularly deep convection in the tropics, remains a persistent problem in climate models. In fact structural biases in the distribution of tropical rainfall in the CMIP5 models is hardly different than that of the CMIP3 versions. Given that regional climate change at higher latitudes is sensitive to the configuration of tropical forcing, this persistent bias is a major issue for the credibility of climate change projections. In this study we use model output from integrations of the NASA Global Earth Observing System Five (GEOS5) climate modeling system to study the evolution of biases in the location and intensity of convective processes. We take advantage of a series of hindcast experiments done in support of the US North American Multi-Model Ensemble (NMME) initiative. For these experiments a nine-month forecast using a coupled model configuration is made approximately every five days over the past 30 years. Each forecast is started with an updated analysis of the ocean, atmosphere and land states. For a given calendar month we have approximately 180 forecasts with daily means of various quantities. These forecasts can be averaged to essentially remove "weather scales" and highlight systematic errors as they evolve. Our primary question is to ask how the spatial structure of daily mean precipitation over the tropics evolves from the initial state and what physical processes are involved. Errors in parameterized convection, various water and energy fluxes and the divergent circulation are found to set up on fast time scales (order five days) compared to errors in the ocean, although SST changes can be non-negligible over that time. For the month of June the difference between forecast day five versus day zero precipitation looks quite similar to the difference between the June precipitation climatology and that from the Global Precipitation Climatology Project (GPCP). We focus much of our analysis on the influence of SST gradients, associated PBL baroclinicity enabled by turbulent mixing, the ensuing PBL moisture convergence, and how changes in these processes relate to convective precipitation bias growth over this short period.

Cohen, Charlie↗

A Physical Validation Program for the GPM Mission

The GPM mission is currently planned for start in the late 2007 - early 2008 time frame. Its main scientific goal is to help answer pressing scientific problems arising within the context of global and regional water cycling. These problems cut across a hierarchy of scales and include climate-water cycle interactions, techniques for improving weather and climate predictions, and better methods for combining observed precipitation with hydrometeorological prediction models for applications to hazardous flood-producing storms, seasonal flood draught conditions, and fresh water resource assessments. The GPM mission will expand the scope of precipitation measurement through the use of a constellation of some 9 satellites, one of which will be an advanced TRMM-like core satellite carrying a dual-frequency Ku-Ka band precipitation radar and an advanced, multifrequency passive microwave radiometer with vertical-horizontal polarization discrimination. The other constellation members will include new dedicated satellites and co-existing operational/research satellites carrying similar (but not identical) passive microwave radiometers. The goal of the constellation is to achieve approximately 3-hour sampling at any spot on the globe -- continuously. The constellation's orbit architecture will consist of a mix of sun-synchronous and non-sun-synchronous satellites with the core satellite providing measurements of cloud-precipitation microphysical processes plus calibration-quality rainrate retrievals to be used with the other retrieval information to ensure bias-free constellation coverage. A major requirement before the retrieved rainfall information generated by the GPM mission can be used effectively by prognostic models to improve weather forecasts, hydrometeorological forecasts, and climate model reanalysis simulations is a capability to quantify the error characteristics of the retrievals. A solution for this problem has been upheld in past precipitation missions because of the lack of suitable error modeling systems incorporated into the validation programs and data distribution systems. An overview of how NASA intends to overcome this problem for the GPM mission using a physically-based error modeling approach within a multi-faceted validation program is described. The solution is to first identify specific user requirements and then determine the most stringent of these requirements that embodies all essential error characterization information needed by the entire user community. In the context of NASA s scientific agenda for the GPM mission, the most stringent user requirement is found within the data assimilation community. The fundamental theory of data assimilation vis-a-vis ingesting satellite precipitation information into the pre-forecast initializations is based on quantifying the conditional bias and precision errors of individual rain retrievals, and the space-time structure of the precision error (i.e., the spatial-temporal error covariance). By generating the hardware and software capability to produce this information in a near real-time fashion, and to couple the derived quantitative error properties to the actual retrieved rainrates, all key validation users can be satisfied. The talk will describe the essential components of the hardware and software systems needed to generate such near real-time error properties, as well as the various paradigm shifts needed within the validation community to produce a validation program relevant to the precipitation user community.

Smith, Eric A.↗

Neural-network learning of SPOD latent dynamics

Here, we aim to reconstruct the latent space dynamics of high dimensional, quasi-stationary systems using model order reduction via the spectral proper orthogonal decomposition (SPOD). The proposed method is based on three fundamental steps: in the first, once that the mean flow field has been subtracted from the realizations (also referred to as snapshots), we compress the data from a high-dimensional representation to a lower dimensional one by constructing the SPOD latent space; in the second, we build the time-dependent coefficients by projecting the snapshots containing the fluctuations onto the SPOD basis and we learn their evolution in time with the aid of recurrent neural networks; in the third, we reconstruct the high-dimensional data from the learnt lower -dimensional representation. The proposed method is demonstrated on two different test cases, namely, a compressible jet flow, and a geophysical problem known as the Madden-Julian Oscillation. An extensive comparison between SPOD and the equivalent POD-based counterpart is provided and differences between the two approaches are highlighted. The numerical results suggest that the proposed model is able to provide low rank predictions of complex statistically stationary data and to provide insights into the evolution of phenomena characterized by specific range of frequencies. The comparison between POD and SPOD surrogate strategies highlights the need for further work on the characterization of the interplay of error between data reduction techniques and neural network forecasts.

97 MATHEMATICS AND COMPUTING↗

Magnetic field calculation for a 10 MeV positron emission tomography cyclotron

The magnetic field calculation and correction for a 10 MeV positron emission tomography cyclotron is presented. 3D TOSCA analysis results are compared with the measured data, and the calculation error is used to calibrate the B-H curve to obtain a very precise finite element method estimator, which is used to predict the correction of the magnet pole for achieving the isochronous field. The isochronous field error is approximated with the effects of a set of standard patches. On the assumption that the effect of each small patch is proportional to its surface, the correction of the magnet pole is found by solving a system of equations using the least square scheme. The magnet shimming is performed and the measured magnetic field is found in good agreement with the prediction, with an error less than 2 G.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Full waveform inversion of solar interior flows

The inference of flows of material in the interior of the Sun is a subject of major interest in helioseismology. Here, we apply techniques of full waveform inversion (FWI) to synthetic data to test flow inversions. In this idealized setup, we do not model seismic realization noise, training the focus entirely on the problem of whether a chosen supergranulation flow model can be seismically recovered. We define the misfit functional as a sum of L {sub 2} norm deviations in travel times between prediction and observation, as measured using short-distance filtered f and p {sub 1} and large-distance unfiltered p modes. FWI allows for the introduction of measurements of choice and iteratively improving the background model, while monitoring the evolution of the misfit in all desired categories. Although the misfit is seen to uniformly reduce in all categories, convergence to the true model is very slow, possibly because it is trapped in a local minimum. The primary source of error is inaccurate depth localization, which, due to density stratification, leads to wrong ratios of horizontal and vertical flow velocities ({sup c}ross talk{sup )}. In the present formulation, the lack of sufficient temporal frequency and spatial resolution makes it difficult to accurately localize flow profiles at depth. We therefore suggest that the most efficient way to discover the global minimum is to perform a probabilistic forward search, involving calculating the misfit associated with a broad range of models (generated, for instance, by a Monte Carlo algorithm) and locating the deepest minimum. Such techniques possess the added advantage of being able to quantify model uncertainty as well as realization noise (data uncertainty).

79 ASTRONOMY AND ASTROPHYSICS↗

A composite likelihood approach for inference under photometric redshift uncertainty

ABSTRACT Obtaining accurately calibrated redshift distributions of photometric samples is one of the great challenges in photometric surveys like LSST, Euclid, HSC, KiDS, and DES. We present an inference methodology that combines the redshift information from the galaxy photometry with constraints from two-point functions, utilizing cross-correlations with spatially overlapping spectroscopic samples, and illustrate the approach on CosmoDC2 simulations. Our likelihood framework is designed to integrate directly into a typical large-scale structure and weak lensing analysis based on two-point functions. We discuss efficient and accurate inference techniques that allow us to scale the method to the large samples of galaxies to be expected in LSST. We consider statistical challenges like the parametrization of redshift systematics, discuss and evaluate techniques to regularize the sample redshift distributions, and investigate techniques that can help to detect and calibrate sources of systematic error using posterior predictive checks. We evaluate and forecast photometric redshift performance using data from the CosmoDC2 simulations, within which we mimic a DESI-like spectroscopic calibration sample for cross-correlations. Using a combination of spatial cross-correlations and photometry, we show that we can provide calibration of the mean of the sample redshift distribution to an accuracy of at least 0.002(1 + z), consistent with the LSST-Y1 science requirements for weak lensing and large-scale structure probes.

(cosmology:) large-scale structure of Universe↗

Polarization of prompt J/{psi} in proton-proton collisions at RHIC

Within the framework of the nonrelativistic QCD (NRQCD) factorization approach, we compute the polarization of prompt J/{psi} produced at the Brookhaven's Relativistic Heavy-Ion Collider from proton-proton collisions at the center-of-momentum energy {radical}(s)=200 GeV. The perturbative contributions are computed at leading order in the strong coupling constant. The prediction reveals that the color-singlet contribution severely underestimates the PHENIX preliminary data for the differential cross section integrated over the rapidity range |y|<0.35 and its contribution is strongly transversely polarized, which disagrees with the PHENIX preliminary data. After including the color-octet contributions, we find that the NRQCD predictions for both the cross section and polarization over the transverse-momentum range 1.5 GeV<p{sub T}<5 GeV (1.5 GeV<p{sub T}<2 GeV) integrated over the rapidity range |y|<0.35 (1.2<|y|<2.2) agree with the data within errors.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Hadronic light by light contribution to the (g-2){sub {mu}} with holographic models of QCD

We study the anomalous electromagnetic pion form factor F{sub {pi}}{sup 0}{sub {gamma}}{sup *}{sub {gamma}}{sup *} with a set of holographic models. By comparing with the measured value of the linear slope, some of these models can be ruled out. From the remaining models, we obtain predictions for the low-energy quadratic slope parameters of F{sub {pi}}{sup 0}{sub {gamma}}{sup *}{sub {gamma}}{sup *}, currently out of experimental reach but testable in the near future. We find it particularly useful to encode this low-energy information in a form factor able to satisfy also QCD short-distance constraints. We choose the form factor introduced by D'Ambrosio, Isidori, and Portoles in kaon decays, which has the right short distance for a particular value of the quadratic slope, which is later shown to be compatible with our holographic predictions. We then turn to a determination of the (dominant) pion exchange diagram in the hadronic light by light scattering contribution to the muon anomalous magnetic moment. We quantify the theoretical uncertainty in (g-2){sub {mu}} coming from the different input we use: QCD short distances, experimental input, and low-energy holographic predictions. We also test the pion-pole approximation. Our final result is a{sub {mu}}{sup {pi}0}=6.54(25)x10{sup -10}, where the error is driven by the linear slope of F{sub {pi}}{sup 0}{sub {gamma}}{sup *}{sub {gamma}}{sup *}, soon to be measured with precision at KLOE-2. Our numerical analysis also indicates that large values of the magnetic susceptibility {chi}{sub 0} are disfavored, therefore pointing at a mild effect from the pion off-shellness. However, in the absence of stronger bounds on {chi}{sub 0}, an additional (10-15)% systematic uncertainty on the previous value for a{sub {mu}}{sup {pi}0} cannot be excluded.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

A Machine Learning Correction Model of the Winter Clear-Sky Temperature Bias over the Arctic Sea Ice in Atmospheric Reanalyses

Atmospheric reanalyses are widely used to estimate the past atmospheric near-surface state over sea ice. They provide boundary conditions for sea ice and ocean numerical simulations and relevant information for studying polar variability and anthropogenic climate change. Previous research revealed the existence of large near-surface temperature biases (mostly warm) over the Arctic sea ice in the current generation of atmospheric reanalyses, which is linked to a poor representation of the snow over the sea ice and the stably stratified boundary layer in the forecast models used to produce the reanalyses. These errors can compromise the employment of reanalysis products in support of polar research. Here, we train a fully connected neural network that learns from remote sensing infrared temperature observations to correct the existing generation of uncoupled atmospheric reanalyses (ERA5, JRA-55) based on a set of sea ice and atmospheric predictors, which are themselves reanalysis products. The advantages of the proposed correction scheme over previous calibration attempts are the consideration of the synoptic weather and cloud state, compatibility of the predictors with the mechanism responsible for the bias, and a self-emerging seasonality and multidecadal trend consistent with the declining sea ice state in the Arctic. The correction leads on average to a 27% temperature bias reduction for ERA5 and 7% for JRA-55 if compared to independent in situ observations from the MOSAiC campaign (respectively, 32% and 10% under clear-sky conditions). These improvements can be beneficial for forced sea ice and ocean simulations, which rely on reanalyses surface fields as boundary conditions.

54 ENVIRONMENTAL SCIENCES↗

Reinvestigation into Closure Predictions of Room D at the Waste Isolation Pilot Plant

Room D was an in-situ, isothermal, underground experiment conducted at the Waste Isolation Pilot Plant between 1984 and 1991. The room was carefully instrumented to measure the horizontal and vertical closure immediately upon excavation and for several years thereafter. Early finite element simulations of salt creep around Room D under-predicted the vertical closure by 4.5×, causing investigators to explore a series of changes to the way Room D was modeled. Discrepancies between simulations and measurements were resolved through a series of adjustments to model parameters, which were openly acknowledged in published reports. Interest in Room D has been rekindled recently by the U.S./German Joint Project III and Project WEIMOS, which seek to improve the predictions of rock salt constitutive models. Joint Project participants calibrate their models solely against laboratory tests, and benchmark the models against underground experiments, such as room D. This report describes updating legacy Room D simulations to today’s computational standards by rectifying several numerical issues. Subsequently, the constitutive model used in previous modeling is recalibrated two different ways against a suite of new laboratory creep experiments on salt extracted from the repository horizon of the Waste Isolation Pilot Plant. Simulations with the new, laboratory-based, calibrations under-predict Room D vertical closure by 3.1×. A list of potential improvements is discussed.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

A four-dimensional analysis exactly satisfying equations of motion

A four-dimensional analysis is applied to spectral nonlinear models of the atmosphere. The experiment reveals that the four-dimensional analysis errors are smaller than measurement errors, the method is stable in an assimilation cycle, and an accurate estimate of the velocity field is maintained using only temperature observations. It is concluded that the four-dimensional analyses display rapid initial error growth and therefore are better than ordinary forecasts from observations for only the first 24 hours.

Hoffman, R. N.↗

Identification of Robust Terminal-Area Routes in Convective Weather

Convective weather is responsible for large delays and widespread disruptions in the U.S. National Airspace System, especially during summer. Traffic flow management algorithms require reliable forecasts of route blockage to schedule and route traffic. This paper demonstrates how raw convective weather forecasts, which provide deterministic predictions of the vertically integrated liquid (the precipitation content in a column of airspace) can be translated into probabilistic forecasts of whether or not a terminal area route will be blocked. Given a flight route through the terminal area, we apply techniques from machine learning to determine the likelihood that the route will be open in actual weather. The likelihood is then used to optimize terminalarea operations by dynamically moving arrival and departure routes to maximize the expected capacity of the terminal area. Experiments using real weather scenarios on stormy days show that our algorithms recommend that a terminal-area route be modified 30% of the time, opening up 13% more available routes that were forecast to be blocked during these scenarios. The error rate is low, with only 5% of cases corresponding to a modified route being blocked in reality, whereas the original route is in fact open. In addition, for routes predicted to be open with probability 0.95 or greater by our method, 96% of these routes (on average over time horizon) are indeed open in the weather that materializes

Pfeil, Diana Michalek↗

The Gmao Hybrid 4d-Envar Observing System Simulation Experiment Framework.

This work describes the extension of the Global Modeling and Assimilation Office (GMAO) Observing System Simulation Experiment (OSSE) framework to use a hybrid 4D-EnVar scheme instead of 3D-Var. The original 3D-Var and hybrid 4D-EnVar OSSEs use the same version of the data assimilation system (DAS) so that a direct comparison is possible in terms of the validation with respect to their corresponding real cases. Rather than quantifying the differences between the two data assimilation methodologies, a short inter-comparison of upgrading from a 3D- to a 4D-OSSE is provided to highlight aspects where this change matters to the OSSE community and to identify features of data assimilation that can only be explored in a four-dimensional OSSE framework. A short validation of the hybrid 4D-EnVar OSSE shows that conclusions from previous assessments of the 3D-Var OSSE in its ability to mimic the behavior of the real system still hold with the same caveats. Furthermore, some aspects of the ensemble configuration and behavior are discussed along with forecast sensitivity to observation impacts (FSOI). Estimates of error standard deviations are shown to be smaller in the hybrid 4D-EnVar OSSE but with little impact on the character of the error. A discussion on future work directions focuses on exploring the four-dimensional aspect such as the error distribution within the assimilation window or four-dimensional handling of high-temporal density observations.

Data assimilation↗

Forecasting Multi-Step-Ahead Street-Scale Nuisance Flooding using a seq2seq LSTM Surrogate Model for Real-Time Application in a Coastal-Urban City

In coastal-urban cities facing an elevated risk of nuisance flooding (by rain and tide) due to increased heavy rainfall, sea level rise, urbanization, and aging drainage systems, real-time flood forecasting at the street-scale can provide useful information to transportation decision-makers. Physics-Based Models (PBMs) that offer high accuracy come with high computational runtimes and costs that limit their application for real-time flood forecasting. To address this challenge, Machine Learning (ML) surrogate models trained from PBMs have been proposed to provide street-scale flood forecasts. Previous related studies have focused on using Long Short-Term Memory (LSTM) architectures to model hourly flood depth on streets. While LSTM models can capture input sequences effectively, they fall short in accurately preserving output sequences, limiting their suitability for multi-step-ahead forecasts. The seq2seq LSTM architecture offers a key advantage here by capturing the full sequence of input–output, making it potentially more suitable for multi-step-ahead flood forecasts compared to traditional LSTM models. However, seq2seq LSTM has not been tested for street-scale flood forecasting, particularly for rapidly fluctuating nuisance flooding events which require special attention to its temporal sequences. Hence, in this study, we applied the seq2seq LSTM model to explore multi-step-ahead street-scale nuisance flooding and compared its results to the traditional LSTM model as a benchmark model. LSTM and seq2seq LSTM surrogate models were applied to 22 flood-prone streets in Norfolk, Virginia, as a case study with a 4-hr (short-term) and 8-hr (long-term) lead time. The models were trained with environmental (rainfall and tide) and topographic (elevation, Topographic Wetness Index, and Depth-To-Water) features along with PBM-derived water depths for different storm events. The results demonstrated satisfactory performance of both LSTM and seq2seq LSTM surrogate models throughout the forecast period compared to the PBM. However, the seq2seq LSTM showed lower Mean Absolute Error (MAE)/ Root Mean Square Error (RMSE) and higher Nash–Sutcliffe Efficiency (NSE)/ correlation than the LSTM across most lead times, particularly for long-term forecasting due to its supremacy in handling both input–output sequences together, which is missing in the traditional LSTM. For example, in the long-term, the average RMSE ranges were 0.0268–0.0373 m for LSTM and 0.0226–0.0319 m for seq2seq LSTM, while in the short-term, they were 0.0263–0.0293 m and 0.0261–0.0283 m, respectively. Additionally, while both models exhibited similar performance in distinguishing flooded and non-flooded streets for flood depth ≥ 0.1 m, the seq2seq LSTM model demonstrated superior performance for higher flood depths (such as ≥ 0.2 m and ≥ 0.3 m). Once trained, inference took only 0.09 to 0.11 s (short-term) and 0.30 to 0.35 s (long-term) per storm event for the 22 streets, making the application highly suitable for real-time decision-making during nuisance flood events.

54 ENVIRONMENTAL SCIENCES↗

Subseasonal-to-Seasonal Hindcast Skill Assessment of Ridging Events Related to Drought Over the Western United States

Persistent atmospheric ridging events centered near the western United States are associated with widespread precipitation deficits and meteorological drought. Due to the relatively low skill of- dynamical models in forecasting precipitation on subseasonal‐to‐seasonal (S2S) time scales across this region, forecasts of ridging are explored in this study as a potential bridge for early warning drought prediction. To assess skill, we evaluate deterministic and probabilistic S2S hindcasts (out to 6 week lead time) of ridging events and geopotential height anomalies over the western United States in five ensemble hindcast systems. Prediction skill for ridging events is shown to be highly variable across models. For some models, longer‐time‐averaged patterns of geopotential height anomalies across the first 6 weeks can be skillfully simulated, when evaluated probabilistically against climatology. The most skillful models show modest skill in forecasting above normal ridging occurrences at lead times of Weeks 3–4 and Weeks 5–6,with some sensitivity to the specific ridge location and method for determining skill. Using the European Centre for Medium‐Range Weather Forecasts (ECMWF) model as a case study, longer lead time forecast busts are shown to often occur under extended periods of highly active La Nina‐like tropical convection. Model errors at simulating these tropical convection features may have a disproportionately large impact and degrade downstream forecasts over the western United States. Despite the documented model shortcomings, our results highlight an opportunity to target skillful ridging forecasts from dynamical models for improving early warning drought forecasting at S2S lead times.

S2S↗

Evaluation of the Near-Surface Variables in the HRRR Weather Model Using Observations from the ARM SGP Site

Abstract The performance of version 4 of the NOAA High-Resolution Rapid Refresh (HRRR) numerical weather prediction model for near-surface variables, including wind, humidity, temperature, surface latent and sensible fluxes, and longwave and shortwave radiative fluxes, is examined over the Atmospheric Radiation Measurement (ARM) Southern Great Plains (SGP) region. The study evaluated the model’s bias and bias-corrected mean absolute error relative to the observations on different time scales. Forecasts of near-surface geophysical variables at five SGP sites (HRRR at 3-km scale) were found to agree well with observations, but some consistent observation–forecast differences also occurred. Sensible and latent heat fluxes are the most challenging variables to be reproduced. The diurnal cycle is the main temporal scale affecting observation–forecast differences of the near-surface variables, and almost all of the variables showed different biases throughout the diurnal cycle. Results show that the overestimation of downward shortwave and the underestimation of downward longwave radiative flux are the two major biases found in this study. The timing and magnitude of downward longwave flux, wind speed, and sensible and latent heat fluxes are also different with contributions from model representations, data assimilation limitations, and differences in scales between HRRR and SGP sites. The positive bias in downward shortwave and negative bias in longwave radiation suggests that the model is underestimating cloud fraction in the study domain. The study concludes by showing a brief comparison with version 3 of the HRRR and shows that version 4 has better performance in almost all near-surface variables. Significance Statement A correct representation of the near-surface variables is important for numerical weather prediction models. This study investigates the capability of the latest NOAA High-Resolution Rapid Refresh (HRRRv4) model in simulating the near-surface variables by comparing against the Atmospheric Radiation Measurement (ARM) Southern Great Plains (SGP) in situ observations. Among others, we find that the surface heat fluxes, such as sensible and latent heat fluxes, are the most difficult variables to be reproduced. This study also shows that the diurnal cycle has the dominant impact on the model’s performance, which means the majority of the outputted near-surface variables have the strong diurnal cycle in their bias errors.

54 ENVIRONMENTAL SCIENCES↗