Research Testbed Research Articles

An Implementation of the Particle Flow Filter in an Atmospheric Model

Abstract The particle flow filter (PFF) shows promise for fully nonlinear data assimilation (DA) in high-dimensional systems. However, its application in atmospheric models has been relatively unexplored. In this study, we develop a new algorithm, PFF-DART, in order to conduct DA for high-dimensional atmospheric models. PFF-DART combines the PFF and the two-step ensemble filtering algorithm in the Data Assimilation Research Testbed (DART), exploiting the highly parallel structure of DART. To evaluate the performance of PFF-DART, we conduct an observing system simulation experiment (OSSE) in a simplified atmospheric general circulation model and compare the performance of PFF-DART with an existing linear and Gaussian DA method. Using the PFF-DART algorithm, we demonstrate, for the first time, the capability of the PFF to yield stable results in a yearlong cycling DA OSSE. Moreover, PFF-DART retains the important ability of the PFF to improve the assimilation of nonlinear and non-Gaussian observations. Finally, we emphasize that PFF-DART is a versatile algorithm that can be integrated with numerous other non-Gaussian DA techniques. This quality makes it a promising method for further investigation within a more sophisticated numerical weather prediction model in the future. Significance Statement Data assimilation is an important tool in weather forecasting. However, fundamental issues arising from linear and Gaussian approximations still exist, which can limit our utilization of observations. The particle flow filter is a promising method that avoids these approximations, but efficient algorithms for this method have yet to be developed. In this study, we develop an algorithm for the particle flow filter that can be implemented in high-dimensional geophysical models. We have demonstrated, for the first time, that the particle flow filter runs efficiently for a yearlong experiment in an atmospheric model. The new algorithm also improves the results over a commonly used data assimilation method. The results in this work demonstrate the potential usage of the particle flow filter in the weather forecasting and other high-dimensional forecasting problems.

Enhanced regional ocean ensemble data assimilation through atmospheric coupling in the SKRIPS model

We investigate the impact of ocean data assimilation using the Ensemble Adjustment Kalman Filter (EAKF) from the Data Assimilation Research Testbed (DART) on the oceanic and atmospheric states of the Red Sea. Our study extends the ocean data assimilation experiment performed by Sanikommu et al. (2020) by utilizing the SKRIPS model coupling the MITgcm ocean model and the Weather Research and Forecasting (WRF) atmosphere model. Using a 50-member ensemble, we assimilate satellite-derived sea surface temperature and height and in situ temperature and salinity profiles every three days for one year, starting January 01 2011. Atmospheric data are not assimilated in the experiments. To improve the ensemble realism, perturbations are added to the WRF model using several physics options and the stochastic kinetic energy backscatter (SKEB) scheme. Compared with the control experiments using uncoupled MITgcm with ECMWF ensemble forcing, the EAKF ensemble mean oceanic states from the coupled model are better or insignificantly worse (root-mean-square errors are 23% to −1.3% smaller), especially when the atmospheric model uncertainties are accounted for with stochastic perturbations. We hypothesize that the ensemble spreads of the air–sea fluxes are better represented in the downscaled WRF ensembles when uncertainties are well accounted for, leading to improved representation of the ensemble oceanic states from the new experiments with the coupled model. This indicates the ocean model assimilation will be improved with coupled models and may relax the need for operational centers to provide atmospheric ensembles to drive ocean forecasts. Although the feedback from ocean to atmosphere is included in this two-way regional coupled configuration, we find no significant effect of ocean data assimilation on the ensemble mean latent heat flux and 10-m wind speed over the Red Sea. This suggests that the improved skill using the coupled model is not from the two-way coupling, but from downscaling the ensemble atmospheric forcings (one-way coupled) to drive the ocean model.

Design and validation of the ADNI MR protocol.

Phase four of the Alzheimer's Disease Neuroimaging Initiative (ADNI4) magnetic resonance imaging (MRI) protocols aim to maintain longitudinal consistency across two decades of data acquisition, while adopting new technologies. Here we describe and justify the study's design and targeted biomarkers. The ADNI4 MRI protocol includes nine MRI sequences. Some sequences require the latest hardware and software system upgrades and are continuously rolled out as they become available at each site. The main sequence additions/changes in ADNI4 are: (1) compressed sensing (CS) T1-weighting, (2) pseudo-continuous arterial spin labeling (ASL) on all three vendors (GE, Siemens, Philips), (3) multiple-post-labeling-delay ASL, (4) 1 mm3 isotropic 3D fluid-attenuated inversion recovery, and (5) CS 3D T2-weighted. ADNI4 aims to help the neuroimaging community extract valuable imaging biomarkers and provide a database to test the impact of advanced imaging strategies on diagnostic accuracy and disease sensitivity among individuals lying on the cognitively normal to impaired spectrum. HIGHLIGHTS: A summary of MRI protocols for phase four of the Alzheimer's Disease Neuroimaging Initiative (ADNI 4). The design and justification for the ADNI 4 MRI protocols. Compressed sensing and multi-band advances have been applied to improve scan time. ADNI4 protocols aim to streamline safety screening and therapy monitoring. The ADNI4 database will be a valuable test bed for academic research.

Impact of Assimilating Uncrewed Aircraft System Observations on River Valley Fog Prediction

Abstract The impact of assimilating targeted Uncrewed Aircraft System (UAS) observations on the prediction of radiation and river-valley fog is assessed using Observing System Experiments (OSEs). Multi-rotor UAS were deployed during FOGMAP (Frequent in situ Observations Above Ground for Modeling and Advanced Prediction of fog) which took place during the summer of 2022 in northern Kentucky. Targeted UAS missions were flown to sample the spatio-temporal variability of temperature and moisture in the vicinity of the Cincinnati/Northern Kentucky International Airport. During each mission the UAS performed near-continuous profiling at two locations between the surface to 120 m AGL throughout the night. Data denial experiments were performed using the Ensemble Kalman Adjustment Filter available in NSF NCAR’s Data Assimilation Research Testbed (DART) to determine the impact of assimilating UAS observations on the skill of analyses and forecasts issued during potential fog events. Simulations that only assimilated conventional observations tended to have a dry bias in the analyses and forecasts. The dry bias in the analyses was reduced in experiments that assimilated UAS observations leading to improved probabilistic predictions of fog. Sensitivity tests revealed that the ensemble mean analyses were improved when assimilating UAS observations of specific humidity rather than relative humidity (RH) due to the existence of a cold bias near the surface and the negative covariance between RH and temperature. It was also found that either the assumed observation error variance of (1 g kg−1)2 or the ensemble spread of the background specific humidity was too large since their sum tended to overestimate the Root Mean Squared Error (RMSE) of the predicted ensemble mean values.

Investigating Arctic Cyclone–Tropopause Polar Vortex Interactions with Idealized Observing System Simulation Experiments

Abstract Arctic cyclones (ACs) are a primary driver of surface weather in the Arctic, contributing to heat and moisture transport and forcing short-term sea ice variability. Still, our understanding of the processes that drive ACs, particularly their large scales and long lifetimes, is limited. ACs are commonly associated with one or more cyclonic tropopause polar vortices (TPVs), potential vorticity (PV) anomalies in the upper troposphere and lower stratosphere that may spur baroclinic development in the surface system, though the exact processes that link the two have yet to be fully explored. In this study, we investigate physical links between TPVs, especially their mesoscale structure and moisture profiles, and ACs with idealized observing system simulation experiments (OSSEs). Starting with a nature run, we simulate different types of dropsonde observations over a TPV during the nascent phase of a nearby AC. The Model for Prediction Across Scales (MPAS) and the Data Assimilation Research Testbed (DART) ensemble adjustment Kalman filter are then used to run experiments to test the impact of these detailed TPV observations. In addition to a control, five main experiments are conducted, assimilating new observations of temperature and humidity. All experiments reduce forecast errors at the surface and throughout the troposphere. Additional humidity observations alter vertical PV distributions, which in turn impact the development of the AC. Experiments with additional temperature observations exhibit improvements in TPV structure and surrounding PV features and produce stronger surface cyclones with skillful TPV forecasts for up to 36 h longer than the control. Significance Statement Arctic cyclones (ACs) are a weather feature that can produce high winds, precipitation, and changes to sea ice cover in the Arctic. As a result, forecasting these storms accurately is important for human and economic interests in the region; however, there are currently gaps in our understanding of how ACs strengthen and persist. In this study, we explore potential links between ACs and weather features higher up in the atmosphere called tropopause polar vortices (TPVs) using computer modeling experiments. This study shows that there are important connections between the characteristics of TPVs and the development of ACs. These findings will be useful for making more accurate forecasts of future events and advancing our knowledge of how sea ice changes relate to the atmosphere.

Fidelity-aware large-scale distributed network emulation

The design and development of new network protocols, architectures, and technologies requires an evaluation phase where the researcher must provide empirical evidence for the performance of their contributions, potentially in comparison to existing solutions. In this context, network emulation has proven to be an attractive approach as it offers more flexibility compared to traditional testing platforms, and more realism compared to simulation. Network emulators provide contained, customisable, and scalable testing environments both for researchers to evaluate their contributions and for the community to reproduce their results. However, two limitations to network emulation have been identified and well documented in the literature: its scalability limits and its accuracy issues. This paper11This journal paper is an extension of a shorter paper (Elbouanani et al., 2023 [1]) previously presented in the TASIR (Testbeds for Advanced Systems Implementation and Research) workshop of Comsnets 2023 conference, and published in its proceedings. documents our attempts to address these concerns. Our findings are distilled into Hifinet: a lightweight scalable and fidelity-aware distributed network emulator. We particularly show how Hifinet outperforms its state-of-the-art counterparts in terms of scalability and efficiency by working around the flaws of their design principles and the technological limitations of the tools they rely on. Hifinet is also fidelity-enhanced, in that it implements a well-theorised fidelity monitoring framework, which passively monitors emulated packet delays to evaluate realism of network emulation and accuracy of results. Another asset of Hifinet is its ability to infer underlying causes in case of erroneous emulation and guide the user through fixing them. This is achieved by using delay tomography algorithms and heuristics.

Advantages of assimilating multispectral satellite retrievals of atmospheric composition: a demonstration using MOPITT carbon monoxide products

Abstract. The Measurements Of Pollution In The Troposphere (MOPITT) is an ideal instrument to understand the impact of (1) assimilating multispectral and joint retrievals versus single spectral products, (2) assimilating satellite profile products versus column products, and (3) assimilating multispectral and joint retrievals versus assimilating individual products separately. We use the Community Atmosphere Model with chemistry with the Data Assimilation Research Testbed (CAM-chem+DART) to assimilate different MOPITT carbon monoxide (CO) products to address these three questions. Both anthropogenic and fire CO emissions are optimized in the data assimilation experiments. The results are compared with independent CO observations from TROPOspheric Monitoring Instrument (TROPOMI), the Total Carbon Column Observing Network (TCCON), NOAA Carbon Cycle Greenhouse Gases (CCGG) sites, In-service Aircraft for a Global Observing System (IAGOS), and Western wildfire Experiment for Cloud chemistry, Aerosol absorption and Nitrogen (WE-CAN). We find that (1) assimilating the MOPITT joint (multispectral; near-IR and thermal IR) column product leads to better model–observation agreement at and near the surface than assimilating the MOPITT thermal-IR-only column retrieval. (2) Assimilating column products has a larger impact and improvement for background and large-scale CO compared to assimilating profile products due to vertical localization in profile assimilation. However, profile assimilation can outperform column assimilations in fire-impacted regions and near the surface. (3) Assimilating multispectral and joint products results in similar or slightly better agreement with observations compared to assimilating the single spectral products separately.

Assimilating Space‐Based Thermospheric Neutral Density (TND) Data Into the TIE‐GCM Coupled Model During Periods With Low and High Solar Activity

AbstractThe global estimation of Thermospheric Neutral Density (TND) and electron density (Ne) on various altitudes are provided by upper atmosphere models, however, the quality of their forecasts needs to be improved. In this study, we present the impact of assimilating space‐based TNDs, measured along Low Earth Orbit (LEO) mission, into the NCAR Thermosphere‐Ionosphere‐Electrodynamics General Circulation Model (TIE‐GCM). In these experiments, the Ensemble Kalman Filter (EnKF) merger of the Data Assimilation Research Testbed (DART) community software is applied. To cover various space‐based TND data and both low and high solar activity periods, we used the measurements of CHAMP (Challenging Minisatellite Payload) and Swarm‐C as assimilated observations. The TND forecasts are then validated against independent TNDs of GRACE (Gravity Recovery and Climate Experiment mission) and Swarm‐B, respectively. To introduce the impact of the thermosphere on estimating ionospheric parameters, the outputs of Ne are validated against the radio occultation data. The Data Assimilation (DA) results indicate that TIE‐GCM overestimates (underestimates) TND and Ne during low (high) solar activity. Considerable improvements are found in forecasting TNDs after DA, that is, the Root Mean Squared Error (RMSE) is reduced by 79% and 51% during low and high solar activity periods, respectively. The reduction values for Ne are found to be 52.3% and 40.4%, respectively.

An Observing System Simulation Experiment Analysis of How Well Geostationary Satellite Trace‐Gas Observations Constrain NOx Emissions in the US

AbstractWe investigate the benefit of assimilating high spatial‐temporal resolution nitrogen dioxide (NO2) measurements from a geostationary (GEO) instrument such as Tropospheric Emissions: Monitoring of Pollution (TEMPO) versus a low‐earth orbit (LEO) platform like TROPOspheric Monitoring Instrument (TROPOMI) on the inverse modeling of nitrogen oxides (NOx) emissions. We generated synthetic TEMPO and TROPOMI NO2 measurements based on emissions from the COVID‐19 lockdown period. Starting with emissions levels prior to the lockdown, we use the Weather Research and Forecasting Model coupled with Chemistry/Data Assimilation Research Testbed (WRF‐Chem/DART) to assimilate these pseudo‐observations in Observing System Simulation Experiments to adjust NOx emissions and quantify how well the assimilation of TEMPO versus TROPOMI measurements recovers the lockdown‐induced emissions changes. We find that NOx emission biases can be ameliorated using half as many simulation days when assimilating GEO observations, and the estimated NOx emissions in 23 out of 29 major urban regions in the US are more accurate. The root mean square error and coefficient of determination of posterior NOx emissions are reduced by 12.5%–41.5% and 1.5%–17.1%, respectively, across different regions. We conduct sensitivity experiments that use different data assimilation (DA) configurations to assimilate synthetic GEO observations. Results demonstrate that the temporal width of the DA window introduces −10% to −20% biases in the emissions inversion and constraining both NOx concentrations and emissions simultaneously yields the most accurate NOx emissions estimates. Our work serves as a valuable reference on how to appropriately assimilate GEO observations for constraining NOx emissions in future studies.

SliceChip: a benchtop fluidic platform for organotypic culture and serial assessment of human and rodent pancreatic slices.

Enzymatically isolated pancreatic islets are the most commonly used ex vivo testbeds for diabetes research. Recently, precision-cut living slices of human pancreas are emerging as an exciting alternative because they maintain the complex architecture of the endocrine and exocrine tissues, and do not suffer from the mechanical and chemical stress of enzymatic isolation. We report a fluidic pancreatic SliceChip platform with dynamic environmental controls that generates a warm, oxygenated, and bubble-free fluidic pathway across singular immobilized slices with continuous deliver of fresh media and the ability to perform repeat serial perfusion assessments. A degasser ensures the system remains bubble-free while systemic pressurization with compressed oxygen ensures slice medium remains adequately oxygenated. Computational modeling of perfusion and oxygen dynamics within SliceChip guide the system's physiomimetic culture conditions. Maintenance of the physiological glucose dependent insulin secretion profile across repeat perfusion assessments of individual pancreatic slices kept under physiological oxygen levels demonstrated the culture capacity of our platform. Fluorescent images acquired every 4 hours of transgenic murine pancreatic slices were reliably stable and recoverable over a 5 day period due to the inclusion of a 3D-printed bioinert metallic anchor that maintained slice position within the SliceChip. Our slice on a chip platform has the potential to expand the useability of human pancreatic slices for diabetes pathogenesis and the development of new therapeutic approaches, while also enabling organotypic culture and assessment of other tissue slices such as brain and patient tumors.

Mesospheric Water Vapor From SABER as a Tracer for the Residual Mean Circulation During SSW Events

AbstractThe sparse wind observations and reanalysis winds in the mesosphere make it challenging to accurately determine the residual mean meridional circulation (MMC). As winds distribute tracers, an alternative approach is to utilize long‐lived trace species such as water vapor (H2O). The recently released H2O data from Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) version 2.07, combined with simulations by the Whole Atmosphere Community Climate Model eXtension with specified dynamics (SD‐WACCM‐X) and with the Data Assimilation Research Testbed (WACCMX + DART), provide an opportunity to assess the accuracy of using H2O isopleths to derive the vertical component of the residual MMC during the 2009 sudden stratospheric warming (SSW) event. In winter, the impact of photochemistry and diffusion on the distribution of mesospheric H2O is negligible compared to advective processes. H2O poleward of 70°N accurately captures the anomalous ascent that occurs a few days prior to the onset of the SSW and the subsequent descent. The derived vertical velocity in SD‐WACCM‐X and WACCMX + DART is qualitatively consistent with SABER observations. However, the derivation of vertical motion from isopleth analysis has limitations at the beginning of 2009 when the meridional transport is stronger than the vertical transport. While measuring winds in the mesosphere is challenging, satellite observations of mesospheric H2O prove to be an effective dynamical tracer during time periods characterized by strong vertical coupling.

Data assimilation for the Model for Prediction Across Scales – Atmosphere with the Joint Effort for Data assimilation Integration (JEDI-MPAS 2.0.0-beta): ensemble of 3D ensemble-variational (En-3DEnVar) assimilations

Abstract. An ensemble of 3D ensemble-variational (En-3DEnVar) data assimilations is demonstrated with the Joint Effort for Data assimilation Integration (JEDI) with the Model for Prediction Across Scales – Atmosphere (MPAS-A) (i.e., JEDI-MPAS). Basic software building blocks are reused from previously presented deterministic 3DEnVar functionality and combined with a formal experimental workflow manager in MPAS-Workflow. En-3DEnVar is used to produce an 80-member ensemble of analyses, which are cycled with ensemble forecasts in a 1-month experiment. The ensemble forecasts approximate a purely flow-dependent background error covariance (BEC) at each analysis time. The En-3DEnVar BECs and prior ensemble-mean forecast errors are compared to those produced by a similar experiment that uses the Data Assimilation Research Testbed (DART) ensemble adjustment Kalman filter (EAKF). The experiment using En-3DEnVar produces a similar ensemble spread to and slightly smaller errors than the EAKF. The ensemble forecasts initialized from En-3DEnVar and EAKF analyses are used as BECs in deterministic cycling 3DEnVar experiments, which are compared to a control experiment that uses 20-member MPAS-A forecasts initialized from Global Ensemble Forecast System (GEFS) initial conditions. The experimental ensembles achieve mostly equivalent or better performance than the off-the-shelf ensemble system in this deterministic cycling setting, although there are many obvious differences in configuration between GEFS and the two MPAS ensemble systems. An additional experiment that uses hybrid 3DEnVar, which combines the En-3DEnVar ensemble BEC with a climatological BEC, increases tropospheric forecast quality compared to the corresponding pure 3DEnVar experiment. The JEDI-MPAS En-3DEnVar is technically working and useful for future research studies. Tuning of observation errors and spread is needed to improve performance, and several algorithmic advancements are needed to improve computational efficiency for larger-scale applications.

Global Scale Inversions from MOPITT CO and MODIS AOD

Top-down observational constraints on emissions flux estimates from satellite observations of chemical composition are subject to biases and errors stemming from transport, chemistry and prior emissions estimates. In this context, we developed an ensemble data assimilation system to optimize the initial conditions for carbon monoxide (CO) and aerosols, while also quantifying the respective emission fluxes with a distinct attribution of anthropogenic and wildfire sources. We present the separate assimilation of CO profile v9 retrievals from the Measurements of Pollution in the Troposphere (MOPITT) instrument and Aerosol Optical Depth (AOD), collection 6.1, from the Moderate Resolution Imaging Spectroradiometer (MODIS) instruments. This assimilation system is built on the Data Assimilation Research Testbed (DART) and includes a meteorological ensemble to assimilate weather observations within the online Community Atmosphere Model with Chemistry (CAM-chem). Inversions indicate an underestimation of CO emissions in CAMS-GLOB-ANT_v5.1 in China for 2015 and an overestimation of CO emissions in the Fire INventory from NCAR (FINN) version 2.2, especially in the tropics. These emissions increments are consistent between the MODIS AOD and the MOPITT CO-based inversions. Additional simulations and comparison with in situ observations from the NASA Atmospheric Tomography Mission (ATom) show that biases in hydroxyl radical (OH) chemistry dominate the CO errors.

Potential impact of all‐sky assimilation of visible and infrared satellite observations compared with radar reflectivity for convective‐scale numerical weather prediction

AbstractAlthough cloud‐affected satellite observations are heavily used for nowcasting applications, their use in regional data assimilation is very limited despite possible benefits for convective‐scale forecasts. In this article, we estimate the potential impact of assimilating cloud‐affected satellite observations of visible (0.6 m) and near thermal infrared wavelengths (6.2 and 7.3 m) relative to the impact of assimilating radar reflectivity observations. We employed observing‐system simulation experiments with a perfect‐model forecast for two cases of strong convective summertime precipitation. Observations are simulated using the radiative transfer model RTTOV/MFASIS and assimilated by the ensemble adjustment Kalman filter in the Data Assimilation Research Testbed. The Weather Research and Forecasting model at 2‐km grid resolution was used for forecasts. Results show that satellite observations can be nearly as beneficial as three‐dimensional radar reflectivity observations. Under favorable conditions, where the prior contains no error in the stage of storm development but only in horizontal position and strength, the assimilation of visible observations leads to 88% of the radar impact. Under more difficult conditions, the impact of visible and infrared observations still reached 50 and 79%, respectively.

Assimilation of the AMSU-A radiances using the CESM (v2.1.0) and the DART (v9.11.13)–RTTOV (v12.3)

Abstract. To improve the initial condition (“analysis”) for numerical weather prediction, we attempt to assimilate observations from the Advanced Microwave Sounding Unit-A (AMSU-A) on board the low-Earth-orbiting satellites. The data assimilation system, used in this study, consists of the Data Assimilation Research Testbed (DART) and the Community Earth System Model as the global forecast model. Based on the ensemble Kalman filter scheme, DART supports the radiative transfer model that is used to simulate the satellite radiances from the model state. To make the AMSU-A data available to be assimilated in DART, preprocessing modules are developed, which consist of quality control, spatial thinning, and bias correction processes. In the quality control, two sub-processes are included, outlier test and channel selection, depending on the cloud condition and surface type. The bias correction process is divided into scan-bias correction and air-mass-bias correction. Like input data used in DART, the observation errors are also estimated for the AMSU-A channels. In the trial experiments, a positive analysis impact is obtained by assimilating the AMSU-A observations on top of the DART data assimilation system that already makes use of the conventional measurements. In particular, the analysis errors are significantly reduced in the whole troposphere and lower stratosphere over the Northern Hemisphere. Overall, this study demonstrates a positive impact on the analysis when the AMSU-A observations are assimilated in the DART assimilation system.

Comparison of temperature and wind observations in the Tropics in a perfect‐model, global EnKF data assimilation system

Flow‐dependent errors in tropical analyses and short‐range forecasts are analysed using global observing‐system simulation experiments assimilating only temperature, only winds, and both data types using the ensemble Kalman filter (EnKF) Data Assimilation Research Testbed (DART) and a perfect model framework. The idealised, homogeneous observation network provides profiles of wind and temperature data from the nature run for January 2018 using the National Center for Atmospheric Research (NCAR) Community Earth System Model (CESM) forced by the observed sea‐surface temperature. The results show that the assimilation of abundant wind observations in a perfect model makes the temperature data in the Tropics largely uninformative. Furthermore, the assimilation of wind data reduces the background errors in specific humidity twice as much as the assimilation of temperature observations. In all experiments, the largest analysis uncertainties and the largest short‐term forecast errors are found in regions of strong vertical and longitudinal gradients in the background wind, especially in the upper troposphere and lower stratosphere over the Indian Ocean and Maritime Continent. The horizontal error correlation scales are on average short throughout the troposphere, just several hundred km. The correlation scales of the wind variables in precipitating regions are half of those in nonprecipitating regions. In precipitating regions, the correlations are elongated vertically, especially for the wind variables. Strong positive cross‐correlations between temperature and specific humidity in the precipitating regions are explained using the Clausius–Clapeyron equation.

Demonstrating the Potential Impacts of Assimilating FY-4A Visible Radiances on Forecasts of Cloud and Precipitation with a Localized Particle Filter

Abstract The Advanced Geostationary Radiation Imager (AGRI) on board the Fengyun-4A (FY-4A) satellite provides visible radiances that contain critical information on clouds and precipitation. In this study, the impact of assimilating FY-4A/AGRI all-sky visible radiances on the simulation of a convective system was evaluated with an observing system simulation experiment (OSSE) using a localized particle filter (PF). The localized PF was implemented into the Data Assimilation Research Testbed (DART) coupled with the Weather Research and Forecasting (WRF) Model. The results of a 2-day data assimilation (DA) experiment generated encouraging outcome at a synoptic scale. Assimilating FY-4A/AGRI visible radiances with the localized PF significantly improved the analysis and forecast of cloud water path (CWP), cloud coverage, rain rate, and rainfall areas. In addition, some positive impacts were produced on the temperature and water vapor mixing ratio in the vicinity of cloudy regions. Sensitivity studies indicated that the best results were achieved by the localized PF configured with a localization distance that is equivalent to the model grid spacing (20 km) and with an adequately short cycling interval (30 min). However, the localized PF could not improve cloud vertical structures and cloud phases due to a lack of related information in the visible radiances. Moreover, the localized PF was compared with the ensemble adjustment Kalman filter (EAKF) and it was indicated that the localized PF outperformed EAKF even when the number of ensemble members was doubled for the latter, indicating a great potential of the localized PF in assimilating visible radiances.

Official International Mahjong: A New Playground for AI Research

Games have long been benchmarks and testbeds for AI research. In recent years, with the development of new algorithms and the boost in computational power, many popular games played by humans have been solved by AI systems. Mahjong is one of the most popular games played in China and has been spread worldwide, which presents challenges for AI research due to its multi-agent nature, rich hidden information, and complex scoring rules, but it has been somehow overlooked in the community of game AI research. In 2020 and 2022, we held two AI competitions of Official International Mahjong, the standard variant of Mahjong rules, in conjunction with a top-tier AI conference called IJCAI. We are the first to adopt the duplicate format in evaluating Mahjong AI agents to mitigate the high variance in this game. By comparing the algorithms and performance of AI agents in the competitions, we conclude that supervised learning and reinforcement learning are the current state-of-the-art methods in this game and perform much better than heuristic methods based on human knowledge. We also held a human-versus-AI competition and found that the top AI agent still could not beat professional human players. We claim that this game can be a new benchmark for AI research due to its complexity and popularity among people.

Sketching sound in architectural design using Virtual Reality

A sketch is an important tool that architects use to visualize and formulate design decisions. The sketch process is a crucial stage where design iterations and decisions can be made fast, easy and abundant. Sketching facilitates idea spawning, design assessment, modification and communication. This exploratory study focuses on qualities with Virtual Reality that can be used for sketching sound in the architectural design process. The study was influenced by Contextual Inquiry Interviews and used trained architects as respondents. The main result of this study is a summary of the pros and cons on how Virtual Reality work today and discusses how to develop a test bed for future research. One of the main cons, as of now, is the bulkiness of the equipment and the fact that it is not intended for sketching sound, yet. Computing speeds and acoustic rendering is another issue on the technical side. A possible benefit of using Virtual Reality is that it gives architects the possibility of incorporating sound in the architectural design process, thus making it an integral part of the practice.

Deep Learning Based Object Attitude Estimation for a Laser Beam Control Research Testbed

ABSTRACT This paper presents an object attitude estimation method using a 2D object image for a Laser Beam Control Research Testbed (LBCRT). Motivated by emerging Deep Learning (DL) techniques, a DL model that can estimate the attitude of a rotating object represented by Euler angles is developed. Instead of synthetic data for training and validation of the model, customized data is experimentally created using the laboratory testbed developed at the Naval Postgraduate School. The data consists of Short Wave Infra-Red (SWIR) images of a 3D-printed Unmanned Aerial Vehicle (UAV) model with varying attitudes and associated Euler angle labels. In the testbed, the estimated attitude is used to aim a laser beam to a specific point of the rotating model UAV object. The attitude estimation model is trained with 1684 UAV images and validated with 421 UAV images not used in the model training. The validation results show the Root-Mean-Square (RMS) angle estimation errors of 6.51 degrees in pitch, 2.74 degrees in roll, and 2.51 degrees in yaw. The Extended Kalman Filter (EKF) is also integrated to show the reduced RMS estimation errors of 1.36 degrees in pitch, 1.20 degrees in roll, and 1.52 degrees in yaw.