Using deep learning to emulate and accelerate a radiative-transfer model

2021 ◽  
Author(s):  
Ryan Lagerquist ◽  
David Turner ◽  
Imme Ebert-Uphoff ◽  
Jebb Stewart ◽  
Venita Hagerty
Keyword(s):  
Deep Learning ◽  
Radiative Transfer ◽  
Heating Rate ◽  
Weather Prediction ◽  
Computing Time ◽  
Single Layer ◽  
Radiative Heating ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Heating Rates

AbstractThis paper describes the development of U-net++ models, a type of neural network that performs deep learning, to emulate the shortwave Rapid Radiative-transfer Model (RRTM). The goal is to emulate the RRTM accurately in a small fraction of the computing time, creating a U-net++ that could be used as a parameterization in numerical weather prediction (NWP). Target variables are surface downwelling flux, top-of-atmosphere upwelling flux (), net flux, and a profile of radiative-heating rates. We have devised several ways to make the U-net++ models knowledge-guided, recently identified as a key priority in machine learning (ML) applications to the geosciences. We conduct two experiments to find the best U-net++ configurations. In Experiment 1, we train on non-tropical sites and test on tropical sites, to assess extreme spatial generalization. In Experiment 2, we train on sites from all regions and test on different sites from all regions, with the goal of creating the best possible model for use in NWP. The selected model from Experiment 1 shows impressive skill on the tropical testing sites, except four notable deficiencies: large bias and error for heating rate in the upper stratosphere, unreliable for profiles with single-layer liquid cloud, large heating-rate bias in the mid-troposphere for profiles with multi-layer liquid cloud, and negative bias at lowzenith angles for all flux components and tropospheric heating rates. The selected model from Experiment 2 corrects all but the first deficiency, and both models run ~104 times faster than the RRTM. Our code is available publicly.

scholarly journals A Discrete Ordinate, Multiple Scattering, Radiative Transfer Model of the Venus Atmosphere from 0.1 to 260 μm

2011 ◽  
Vol 68 (6) ◽  
pp. 1323-1339 ◽  
Author(s):  
Christopher Lee ◽  
Mark Ian Richardson
Keyword(s):  
Radiative Transfer ◽  
Radiative Heating ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Discrete Ordinate ◽  
Heating Rates ◽  
Venus Express ◽  
Reference Atmosphere ◽  
Venus Atmosphere ◽  
Good Agreement

Abstract The authors describe a new radiative transfer model of the Venus atmosphere (RTM) that includes optical properties from nine gases and four cloud modes between 0.1 and 260 μm. A multiple-stream discrete ordinate flux solver is used to calculate solar and atmospheric infrared fluxes with a prescribed temperature profiles and calculate radiative–convective equilibrium temperatures using the model. Components of the RTM are validated using observations from Pioneer Venus and Venus Express. A visible bond albedo of 0.74 and subsolar surface visible flux of 50 W m−2 [4.0% of the top-of-atmosphere (TOA) insolation] are calculated for a suitable temperature and composition profile derived from the Venus International Reference Atmosphere. Solar fluxes are simulated over a range of latitudes and good agreement is found with results from the Pioneer Venus probes and Venera landers. TOA infrared fluxes are compared with Venus Express observations and found to compare well at all observed wavelengths. The RTM is used to calculate radiative heating rates and these calculated heating rates are compared with those prescribed in a modern Venus GCM. Modifications are suggested to improve the prescribed thermal forcing used in recent GCMs. Using a small family of numerical and physical configurations, little sensitivity to vertical resolution is found in the model. For suitable global mean solar forcing a surface temperature of 750 K at radiative–convective equilibrium is calculated, in good agreement with observations and other recent modeling efforts.

scholarly journals Characteristic Atmospheric Radiative Heating Rate Profiles in Arctic Clouds as Observed at Barrow, Alaska

2018 ◽  
Vol 57 (4) ◽  
pp. 953-968 ◽  
Author(s):  
D. D. Turner ◽  
M. D. Shupe ◽  
A. B. Zwink
Keyword(s):  
Heating Rate ◽  
Liquid Water ◽  
Single Layer ◽  
Radiative Heating ◽  
Radiative Transfer Model ◽  
Radiative Cooling ◽  
Transfer Model ◽  
Liquid Water Path ◽  
Water Path ◽  
Order Of Magnitude

AbstractA 2-yr cloud microphysical property dataset derived from ground-based remote sensors at the Atmospheric Radiation Measurement site near Barrow, Alaska, was used as input into a radiative transfer model to compute radiative heating rate (RHR) profiles in the atmosphere. Both the longwave (LW; 5–100 μm) and shortwave (SW; 0.2–5 μm) RHR profiles show significant month-to-month variability because of seasonal dependence in the vertical profiles of cloud liquid and ice water contents, with additional contributions from the seasonal dependencies of solar zenith angle, water vapor amount, and temperature. The LW and SW RHR profiles were binned to provide characteristic profiles as a function of cloud type and liquid water path (LWP). Single-layer liquid-only clouds are shown to have larger (10–30 K day−1) LW radiative cooling rates at the top of the cloud layer than single-layer mixed-phase clouds; this is due primarily to differences in the vertical distribution of liquid water between the two classes. However, differences in SW RHR profiles at the top of these two classes of clouds are less than 3 K day−1. The absolute value of the RHR in single-layer ice-only clouds is an order of magnitude smaller than in liquid-bearing clouds. Furthermore, for double-layer cloud systems, the phase and condensed water path of the upper cloud strongly modulate the radiative cooling both at the top and within the lower-level cloud. While sensitivity to cloud overlap and phase has been shown previously, the characteristic RHR profiles are markedly different between the different cloud classifications.

scholarly journals Towards a better representation of the solar cycle in general circulation models

2007 ◽  
Vol 7 (20) ◽  
pp. 5391-5400 ◽  
Author(s):  
K. M. Nissen ◽  
K. Matthes ◽  
U. Langematz ◽  
B. Mayer
Keyword(s):  
Solar Cycle ◽  
Heating Rate ◽  
General Circulation ◽  
Temperature Response ◽  
General Circulation Models ◽  
Short Wave ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Heating Rates ◽  
Radiation Scheme

Abstract. We introduce the improved Freie Universität Berlin (FUB) high-resolution radiation scheme FUBRad and compare it to the 4-band standard ECHAM5 SW radiation scheme of Fouquart and Bonnel (FB). Both schemes are validated against the detailed radiative transfer model libRadtran. FUBRad produces realistic heating rate variations during the solar cycle. The SW heating rate response with the FB scheme is about 20 times smaller than with FUBRad and cannot produce the observed temperature signal. A reduction of the spectral resolution to 6 bands for solar irradiance and ozone absorption cross sections leads to a degradation (reduction) of the solar SW heating rate signal by about 20%. The simulated temperature response agrees qualitatively well with observations in the summer upper stratosphere and mesosphere where irradiance variations dominate the signal. Comparison of the total short-wave heating rates under solar minimum conditions shows good agreement between FUBRad, FB and libRadtran up to the middle mesosphere (60–70 km) indicating that both parameterizations are well suited for climate integrations that do not take solar variability into account. The FUBRad scheme has been implemented as a sub-submodel of the Modular Earth Submodel System (MESSy).

Development of a Fast Three-Dimensional Dynamic Radiative Transfer Solver for Numerical Weather Prediction Models


2021 ◽  
Author(s):  
Richard Maier ◽  
Bernhard Mayer ◽  
Claudia Emde ◽  
Aiko Voigt
Keyword(s):  
Radiative Transfer ◽  
Numerical Weather Prediction ◽  
Prediction Models ◽  
Weather Prediction ◽  
Three Dimensional ◽  
Transfer Model ◽  
Heating Rates ◽  
Time Step ◽  
Numerical Weather ◽  
Numerical Weather Prediction Models

<div> <div> <div> <div> <p>The increasing resolution of numerical weather prediction models makes 3D radiative effects more and more important. These effects are usually neglected by the simple 1D independent column approximations used in most of the current models. On top of that, these 1D radiative transfer solvers are also called far less often than the model’s dynamical core.</p> <p>To address these issues, we present a new „dynamic“ approach of solving 3D radiative transfer. Building upon the existing TenStream solver (Jakub and Mayer, 2015), radiation in this 3D model is not solved completely in each radiation time step, but is rather only transported to adjacent grid boxes. For every grid box, outgoing fluxes are then calculated from the incoming fluxes from the neighboring grid cells of the previous time step. This allows to reduce the computational cost of 3D radiative transfer models to that of current 1D solvers.</p> <p>Here, we show first results obtained with this new solver with a special emphasis on heating rates. Furthermore, we demonstrate issues related to the dynamical treatment of radiation as well as possible solutions to these problems.</p> <p>In the future, the speed of this newly developed 3D dynamic TenStream solver will be further increased by reducing the number of spectral bands used in the radiative transfer calculations with the aim to get a 3D solver that can be called even more frequently than the 1D two-stream solvers used nowadays.</p> <p>Reference:<br><span>Jakub, F. and Mayer, B. (2015), A three-dimensional parallel radiative transfer model for atmospheric heating rates for use in cloud resolving models—The TenStream solver, Journal of Quantitative Spectroscopy and Radiative Transfer, Volume 163, 2015, Pages 63-71, ISSN 0022-4073, . </span></p> </div> </div> </div> </div>

Evaluation of shortwave radiation fluxes in the multi-layer SPARTACUS-Urban scheme using DART

2021 ◽  
Author(s):  
Megan Stretton ◽  
William Morrison ◽  
Robin Hogan ◽  
Sue Grimmond
Keyword(s):  
Radiative Transfer ◽  
Urban Form ◽  
Climate Models ◽  
Weather Prediction ◽  
Computational Cost ◽  
Three Dimensional ◽  
Shortwave Radiation ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Urban Structure

<p>The heterogenous structure of cities impacts radiative exchanges (e.g. albedo and heat storage). Numerical weather prediction (NWP) models often characterise the urban structure with an infinite street canyon – but this does not capture the three-dimensional urban form. SPARTACUS-Urban (SU) - a fast, multi-layer radiative transfer model designed for NWP - is evaluated using the explicit Discrete Anisotropic Radiative Transfer (DART) model for shortwave fluxes across several model domains – from a regular array of cubes to real cities .</p><p>SU agrees with DART (errors < 5.5% for all variables) when the SU assumptions of building distribution are fulfilled (e.g. randomly distribution). For real-world areas with pitched roofs, SU underestimates the albedo (< 10%) and shortwave transmission to the surface (< 15%), and overestimates wall-plus-roof absorption (9-27%), with errors increasing with solar zenith angle. SU should be beneficial to weather and climate models, as it allows more realistic urban form (cf. most schemes) without large increases in computational cost.</p>

scholarly journals Evaluation of MWHS-2 Using a Co-located Ground-Based Radar Network for Improved Model Assimilation

2019 ◽  
Vol 11 (20) ◽  
pp. 2338 ◽  
Author(s):  
Liu ◽  
Chu ◽  
Yin ◽  
Liu
Keyword(s):  
Radiative Transfer ◽  
Brightness Temperature ◽  
Weather Prediction ◽  
Order Relation ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Radar Network ◽  
Intense Precipitation ◽  
Standard Deviations ◽  
Light Precipitation

Accurate precipitation detection is one of the most important factors in satellite data assimilation, due to the large uncertainties associated with precipitation properties in radiative transfer models and numerical weather prediction (NWP) models. In this paper, a method to achieve remote sensing of precipitation and classify its intensity over land using a co-located ground-based radar network is described. This method is intended to characterize the O−B biases for the microwave humidity sounder -2 (MWHS-2) under four categories of precipitation: precipitation-free (0–5 dBZ), light precipitation (5–20 dBZ), moderate precipitation (20–35 dBZ), and intense precipitation (>35 dBZ). Additionally, O represents the observed brightness temperature (TB) of the satellite and B is the simulated TB from the model background field using the radiative transfer model. Thresholds for the brightness temperature differences between channels, as well as the order relation between the differences, exhibited a good estimation of precipitation. It is demonstrated that differences between observations and simulations were predominantly due to the cases in which radar reflectivity was above 15 dBZ. For most channels, the biases and standard deviations of O−B increased with precipitation intensity. Specifically, it is noted that for channel 11 (183.31 ± 1 GHz), the standard deviations of O−B under moderate and intense precipitation were even smaller than those under light precipitation and precipitation-free conditions. Likewise, abnormal results can also be seen for channel 4 (118.75 ± 0.3 GHz).

scholarly journals RTTOV-gb – Adapting the fast radiative transfer model RTTOV for the assimilation of ground-based microwave radiometer observations

2016 ◽  
Author(s):  
Francesco De Angelis ◽  
Domenico Cimini ◽  
James Hocking ◽  
Pauline Martinet ◽  
Stefan Kneifel
Keyword(s):  
Radiative Transfer ◽  
Liquid Water ◽  
Weather Prediction ◽  
Microwave Radiometer ◽  
Software Tool ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
60 Ghz ◽  
Variational Assimilation ◽  
Temperature And Humidity

Abstract. Ground-based microwave radiometers (MWR) offer a new capability to provide continuous observations of the atmospheric thermodynamic state in the planetary boundary layer. Thus, they are potential candidates to supplement radiosonde network and satellite data to improve numerical weather prediction (NWP) models through a variational assimilation of their data. However in order to assimilate MWR observations a fast radiative transfer model is required and such a model is not currently available. This is necessary for going from the model state vector space to the observation space at every observation point. The fast radiative transfer model RTTOV is well accepted in the NWP community, though it was developed to simulate satellite observations only. In this work, the RTTOV code has been modified to allow for simulations of ground-based upward looking microwave sensors. In addition, the Tangent Linear, Adjoint, and K-modules of RTTOV have been adapted to provide Jacobians (i.e. the sensitivity of observations to the atmospheric thermodynamical state) for ground-based geometry. These modules are necessary for the fast minimization of the cost function in a variational assimilation scheme. The proposed ground-based version of RTTOV, called RTTOV-gb, has been validated against accurate and less time-efficient line-by-line radiative transfer models. In the frequency range commonly used for temperature and humidity profiling (22–60 GHz), root-mean-square brightness temperature differences are smaller than typical MWR uncertainties (~ 0.5 K) at all channels used in this analysis. Brightness temperatures (TB) computed with RTTOV-gb from radiosonde profiles have been compared with nearly simultaneous and colocated ground-based MWR observations. Differences between simulated and measured TB are below 0.5 K for all channels except for the water vapor band, where most of the uncertainty comes from instrumental errors. The Jacobians calculated with the K-module of RTTOV-gb have been compared with those calculated with the brute force technique and those from the line-by-line model ARTS. Jacobians are found to be almost identical, except for liquid water content Jacobians for which a 10 % difference between ARTS and RTTOV-gb at transparent channels around 450 hPa is attributed to differences in liquid water absorption models. Finally, RTTOV-gb has been applied as the forward model operator within a 1-Dimensional Variational (1D-Var) software tool in an Observing-System Simulation Experiment (OSSE). For both temperature and humidity profiles, the 1D-Var with RTTOV-gb improves the retrievals with respect to NWP model in the first few kilometers from the ground.

scholarly journals Feasibility study of multi-pixel retrieval of optical thickness and droplet effective radius of inhomogeneous clouds using deep learning

2017 ◽  
Vol 10 (12) ◽  
pp. 4747-4759 ◽  
Author(s):  
Rintaro Okamura ◽  
Hironobu Iwabuchi ◽  
K. Sebastian Schmidt
Keyword(s):  
Deep Learning ◽  
Radiative Transfer ◽  
Optical Thickness ◽  
Cloud Droplet ◽  
Effective Radius ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Optical Remote Sensing ◽  
Transfer Effects ◽  
Cloud Properties

Abstract. Three-dimensional (3-D) radiative-transfer effects are a major source of retrieval errors in satellite-based optical remote sensing of clouds. The challenge is that 3-D effects manifest themselves across multiple satellite pixels, which traditional single-pixel approaches cannot capture. In this study, we present two multi-pixel retrieval approaches based on deep learning, a technique that is becoming increasingly successful for complex problems in engineering and other areas. Specifically, we use deep neural networks (DNNs) to obtain multi-pixel estimates of cloud optical thickness and column-mean cloud droplet effective radius from multispectral, multi-pixel radiances. The first DNN method corrects traditional bispectral retrievals based on the plane-parallel homogeneous cloud assumption using the reflectances at the same two wavelengths. The other DNN method uses so-called convolutional layers and retrieves cloud properties directly from the reflectances at four wavelengths. The DNN methods are trained and tested on cloud fields from large-eddy simulations used as input to a 3-D radiative-transfer model to simulate upward radiances. The second DNN-based retrieval, sidestepping the bispectral retrieval step through convolutional layers, is shown to be more accurate. It reduces 3-D radiative-transfer effects that would otherwise affect the radiance values and estimates cloud properties robustly even for optically thick clouds.

scholarly journals A fast visible-wavelength 3D radiative transfer model for numerical weather prediction visualization and forward modeling

2020 ◽  
Vol 13 (6) ◽  
pp. 3235-3261
Author(s):  
Steven Albers ◽  
Stephen M. Saleeby ◽  
Sonia Kreidenweis ◽  
Qijing Bian ◽  
Peng Xian ◽  
...  
Keyword(s):  
Solar Radiation ◽  
Radiative Transfer ◽  
Numerical Weather Prediction ◽  
Land Surface ◽  
Weather Prediction ◽  
Lower Boundary ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Numerical Weather ◽  
Visible Wavelength

Abstract. Solar radiation is the ultimate source of energy flowing through the atmosphere; it fuels all atmospheric motions. The visible-wavelength range of solar radiation represents a significant contribution to the earth's energy budget, and visible light is a vital indicator for the composition and thermodynamic processes of the atmosphere from the smallest weather scales to the largest climate scales. The accurate and fast description of light propagation in the atmosphere and its lower-boundary environment is therefore of critical importance for the simulation and prediction of weather and climate. Simulated Weather Imagery (SWIm) is a new, fast, and physically based visible-wavelength three-dimensional radiative transfer model. Given the location and intensity of the sources of light (natural or artificial) and the composition (e.g., clear or turbid air with aerosols, liquid or ice clouds, precipitating rain, snow, and ice hydrometeors) of the atmosphere, it describes the propagation of light and produces visually and physically realistic hemispheric or 360∘ spherical panoramic color images of the atmosphere and the underlying terrain from any specified vantage point either on or above the earth's surface. Applications of SWIm include the visualization of atmospheric and land surface conditions simulated or forecast by numerical weather or climate analysis and prediction systems for either scientific or lay audiences. Simulated SWIm imagery can also be generated for and compared with observed camera images to (i) assess the fidelity and (ii) improve the performance of numerical atmospheric and land surface models. Through the use of the latter in a data assimilation scheme, it can also (iii) improve the estimate of the state of atmospheric and land surface initial conditions for situational awareness and numerical weather prediction forecast initialization purposes.

scholarly journals Clear-Sky Mask for the Advanced Clear-Sky Processor for Oceans

2010 ◽  
Vol 27 (10) ◽  
pp. 1609-1623 ◽  
Author(s):  
B. Petrenko ◽  
A. Ignatov ◽  
Y. Kihai ◽  
A. Heidinger
Keyword(s):  
High Resolution ◽  
Radiative Transfer ◽  
Weather Prediction ◽  
Radiative Transfer Model ◽  
Transfer Model ◽  
Clear Sky ◽  
Brightness Temperatures ◽  
Spatial Uniformity ◽  
Uniformity Test ◽  
Cross Platform

Abstract The Advanced Clear Sky Processor for Oceans (ACSPO) generates clear-sky products, such as SST, clear-sky radiances, and aerosol, from Advanced Very High Resolution Radiometer (AVHRR)-like measurements. The ACSPO clear-sky mask (ACSM) identifies clear-sky pixels within the ACSPO products. This paper describes the ACSM structure and compares the performances of ACSM and its predecessor, Clouds from AVHRR Extended Algorithm (CLAVRx). ACSM essentially employs online clear-sky radiative transfer simulations enabled within ACSPO with the Community Radiative Transfer Model (CRTM) in conjunction with numerical weather prediction atmospheric [Global Forecast System (GFS)] and SST [Reynolds daily high-resolution blended SST (DSST)] fields. The baseline ACSM tests verify the accuracy of fitting observed brightness temperatures with CRTM, check retrieved SST for consistency with Reynolds SST, and identify ambient cloudiness at the boundaries of cloudy systems. Residual cloud effects are screened out with several tests, adopted from CLAVRx, and with the SST spatial uniformity test designed to minimize misclassification of sharp SST gradients as clouds. Cross-platform and temporal consistencies of retrieved SSTs are maintained by accounting for SST and brightness temperature biases, estimated within ACSPO online and independently from ACSM. The performance of ACSM is characterized in terms of statistics of deviations of retrieved SST from the DSST. ACSM increases the amount of “clear” pixels by 30% to 40% and improves statistics of retrieved SST compared with CLAVRx. ACSM is also shown to be capable of producing satisfactory statistics of SST anomalies if the reference SST field for the exact date of observations is unavailable at the time of processing.

Sign in / Sign up

Export Citation Format

Share Document