1 NCEP Radiative Transfer Model Status Paul van Delst.

Slides:



Advertisements
Similar presentations
ECMWF/EUMETSAT NWP-SAF Satellite data assimilation Training Course
Advertisements

Upper Tropospheric Humidity: A Comparison of Satellite, Radiosonde, Lidar and Aircraft Measurements Satellite Lidar Aircraft Radiosonde.
Characterization of ATMS Bias Using GPSRO Observations Lin Lin 1,2, Fuzhong Weng 2 and Xiaolei Zou 3 1 Earth Resources Technology, Inc.
Data Assimilation Andrew Collard. Overview Introduction to Atmospheric Data Assimilation Control Variables Observations Background Error Covariance Summary.
Handling Cloud-Affected Infrared Radiances in the GSI Will McCarty GSFC/Global Modeling and Assimilation Office JCSDA Workshop 10 October 2012.
ECMWF CO 2 Data Assimilation at ECMWF Richard Engelen European Centre for Medium-Range Weather Forecasts Reading, United Kingdom Many thanks to Phil Watts,
Extracting Atmospheric and Surface Information from AVIRIS Spectra Vijay Natraj, Daniel Feldman, Xun Jiang, Jack Margolis and Yuk Yung California Institute.
Improved NCEP SST Analysis
1 JCSDA Community Radiative Transfer Model (CRTM) Framework JCSDA 3 rd Workshop on Satellite Data Assimilation, April 2005 Yong Han, Quanhua Liu,
A new prototype AMSR-E SWE operational algorithm M. Tedesco The City College of New York, CUNY, NYC With contributions from : Chris Derksen, Jouni Pulliainen,
A discussion of Radiative Transfer Models Thomas J. Kleespies NOAA/NESDIS.
Diagnosing Climate Change from Satellite Sounding Measurements – From Filter Radiometers to Spectrometers William L. Smith Sr 1,2., Elisabeth Weisz 1,
00/XXXX1 RTTOV-7: A satellite radiance simulator for the new millennium What is RTTOV Latest developments from RTTOV-6 to 7 Validation results for RTTOV-7.
Status of MIRS Updates Kevin Garrett and Sid-Ahmed Boukabara MIRS Meeting February 15, 2008.
Satellite Bias Correction for CFSRR Haixia Liu, Russ Treadon, Robert Kistler, John Derber, Suru Saha and Hua-lu Pan Nov. 7, 2007 with input from Jack Woollen,
Simulation of Observation Simulation of Conventional Observations Jack Woollen (NCEP/EMC) Considerations Data distribution depends on atmospheric conditions.
Validation workshop, Frascati, 13 December 2002Page 1 SCIAMACHY products quality and recommendations Based on presentations and discussions during this.
1 Geostationary Cloud Algorithm Testbed (GEOCAT) Processing Mike Pavolonis and Andy Heidinger (NOAA/NESDIS/STAR) Corey Calvert and William Straka III (UW-CIMSS)
P1.85 DEVELOPMENT OF SIMULATED GOES PRODUCTS FOR GFS AND NAM Hui-Ya Chuang and Brad Ferrier Environmental Modeling Center, NCEP, Washington DC Introduction.
Integrating Community RT Components into JCSDA CRTM Yong Han, Paul van Delst, Quanhua Liu, Fuzhong Weng, Thomas J. Kleespies, Larry M. McMillin.
Challenges and practical applications of data assimilation in numerical weather prediction Data Assimilation Education Forum Part I: Overview of Data Assimilation.
March 18, 2003 MODIS Atmosphere, St. Michaels MD Infrared Retrieval of Temperature, Moisture, Ozone, and Total Precipitable Water: Recent Update and Status.
IGARSS 2011, July 24-29, Vancouver, Canada 1 A PRINCIPAL COMPONENT-BASED RADIATIVE TRANSFER MODEL AND ITS APPLICATION TO HYPERSPECTRAL REMOTE SENSING Xu.
USE OF AIRS/AMSU DATA FOR WEATHER AND CLIMATE RESEARCH Joel Susskind University of Maryland May 12, 2005.
Hyperspectral Infrared Alone Cloudy Sounding Algorithm Development Objective and Summary To prepare for the synergistic use of data from the high-temporal.
Water Vapour & Cloud from Satellite and the Earth's Radiation Balance
Slide 1 VAISALA Award Lecture Characterising the FY-3A Microwave Temperature Sounder Using the ECMWF Model Qifeng Lu, William Bell, Peter Bauer, Niels.
AIRS (Atmospheric Infrared Sounder) Regression Retrieval (Level 2)
AIRS Radiance and Geophysical Products: Methodology and Validation Mitch Goldberg, Larry McMillin NOAA/NESDIS Walter Wolf, Lihang Zhou, Yanni Qu and M.
ITSC 12 5 March 2002 The Merger of OPTRAN and RTTOV: The Best of Both Worlds Thomas J. Kleespies NOAA/NESDIS Camp Springs MD USA.
Jinlong Li 1, Jun Li 1, Christopher C. Schmidt 1, Timothy J. Schmit 2, and W. Paul Menzel 2 1 Cooperative Institute for Meteorological Satellite Studies.
As components of the GOES-R ABI Air Quality products, a multi-channel algorithm similar to MODIS/VIIRS for NOAA’s next generation geostationary satellite.
The Hyperspectral Environmental Suite (HES) and Advanced Baseline Imager (ABI) will be flown on the next generation of NOAA Geostationary Operational Environmental.
Layered Water Vapor Quick Guide by NASA / SPoRT and CIRA Why is the Layered Water Vapor Product important? Water vapor is essential for creating clouds,
Validation of OMPS-LP Radiances P. K. Bhartia, Leslie Moy, Zhong Chen, Steve Taylor NASA Goddard Space Flight Center Greenbelt, Maryland, USA.
© Crown copyright Met Office Assimilating cloud affected infrared radiances at the Met Office Ed Pavelin and Roger Saunders, Met Office, Exeter.
A step toward operational use of AMSR-E horizontal polarized radiance in JMA global data assimilation system Masahiro Kazumori Numerical Prediction Division.
MDTERP MarylanD TErrestial Radiation Package A Narrow-Band Longwave Radiation Model with a Graphical User Interface Robert G. Ellingson and Ezra Takara.
1 JCSDA Infrared Sea Surface Emissivity Model Status Paul van Delst 2 nd MURI Workshop April 2004 Madison WI.
Cloudy Radiance Assimilation in the NCEP Global Forecast System NOAA/NCEP/EMC 4 ESSIC, University of Maryland,
Retrieval of cloud parameters from the new sensor generation satellite multispectral measurement F. ROMANO and V. CUOMO ITSC-XII Lorne, Victoria, Australia.
UCLA Vector Radiative Transfer Models for Application to Satellite Data Assimilation K. N. Liou, S. C. Ou, Y. Takano and Q. Yue Department of Atmospheric.
1 Atmospheric Radiation – Lecture 13 PHY Lecture 13 Remote sensing using emitted IR radiation.
1 Radiative Transfer Models and their Adjoints Paul van Delst.
Matthew Lagor Remote Sensing Stability Indices and Derived Product Imagery from the GOES Sounder
AIRS Land Surface Temperature and Emissivity Validation Bob Knuteson Hank Revercomb, Dave Tobin, Ken Vinson, Chia Lee University of Wisconsin-Madison Space.
NCEP Assessment of ATMS Radiances Andrew Collard 1, John Derber 2 and Russ Treadon 2 1 IMSG at NOAA/NCEP/EMC 2 NOAA/NCEP/EMC 1NPP ATMS SDR Product Review13th.
The assimilation of satellite radiances in LM F. Di Giuseppe, B. Krzeminski,R. Hess, C. Shraff (1) ARPA-SIM Italy (2) IMGW,Poland (3)DWD, Germany.
Measurements and modeling of water vapor from solar spectral irradiance during ATTREX Bruce Kindel, Peter Pilewskie, Sebastian Schmidt, Troy Thornberry,
Radiance Simulation System for OSSE  Objectives  To evaluate the impact of observing system data under the context of numerical weather analysis and.
NOAA Near Real-Time AIRS Processing and Distribution System Walter Wolf Mitch Goldberg Lihang Zhou NESDIS/ORA/CRAD.
June 20, 2005Workshop on Chemical data assimilation and data needs Data Assimilation Methods Experience from operational meteorological assimilation John.
Visible vicarious calibration using RTM
Indirect impact of ozone assimilation using Gridpoint Statistical Interpolation (GSI) data assimilation system for regional applications Kathryn Newman1,2,
ECMWF/EUMETSAT NWP-SAF Satellite data assimilation Training Course
What is atmospheric radiative transfer?
ECMWF/EUMETSAT NWP-SAF Satellite data assimilation Training Course
Summer 2014 Group Meeting August 14, 2014 Scott Sieron
Example for a Satellite: Meteosat Second Generation
Impact of Traditional and Non-traditional Observation Sources using the Grid-point Statistical Interpolation Data Assimilation System for Regional Applications.
Vicarious calibration by liquid cloud target
RECENT INNOVATIONS IN DERIVING ATMOSPHERIC MOTION VECTORS AT CIMSS
Meteosat Second Generation
GIFTS-IOMI Clear Sky Forward Model
Clear Sky Forward Model & Its Adjoint Model
Use of ATOVS at DAO Joanna Joiner, Donald Frank, Arlindo da Silva,
Update on GSICS Product Development
Initialization of Numerical Forecast Models with Satellite data
Infrared Satellite Data Assimilation at NCAR
MURI Hyperspectral Workshop Madison WI, June 7
Presentation transcript:

1 NCEP Radiative Transfer Model Status Paul van Delst

John Derber, NCEP/EMC Yoshihiko Tahara, JMA/NCEP/EMC Larry McMillin, NESDIS/ORA Tom Kleespies, NESDIS/ORA Hal Woolf, CIMSS/SSEC/UWisc Others involved

Introduction RT model “system” – Application – the RT model used in NCEP GDAS Offline tests of RT code Parallel runs of assimilation system using new RT code Issues with some coefficients – Transmittance regression coefficient generation New algorithm from Yoshihiko Tahara – Line-by-line transmittance generation Issues – Atmospheric profile inputs and manipulation – Regression coefficient quality control

RT application software

NCEP Radiative Transfer Model (RTM) All components completed: – Forward, tangent-linear, adjoint, K-matrix. – Parallel testing of updated code in GDAS ongoing. Memory usage and timing are same (even with 2-3x more calculations) for effectively unoptimised code. – Code supplied to NASA DAO, NOAA ETL and FSL. Code availablility – v1.3  Forward and K_matrix code available at – GOES, POES, and AIRS(*) coefficients available. Code comments – ANSI standard Fortran90; no vendor extensions – Platform testbeds: Linux (PGI compilers), IBM SP/RS6000, SGI Origin, Sun SPARC. – Code prototyped in IDL. Not the best choice but allows for simple in situ visualisation and easy detection/rectification of floating point errors.

TL and AD models used in tandem for testing Unit perturbations applied Floating point precision and underflow a concern with transmittance predictor formulation. – Some integrated predictors require the 3 rd and 4 th powers of absorber amount in the denominator. This is a problem for low absorber (e.g. water) amounts. – Current operational code will not run with floating point error handling enabled. Offline tests of RTM

TL N16 HIRS channel radiances wrt T(p)

AD N16 HIRS channel radiances wrt T(p)

|TL-AD| difference for N16 HIRS wrt T(p)

|TL-AD| difference for N16 AMSU wrt W(p)

Integrated absorber formulation Floating point underflow issues Integrated predictor formulation – X == Temperature or Pressure. – Denominator can get very small at high altitudes

Upgraded RTM improves bias in some channels, degrades it in others. Variability is better in some channels with upgraded RTM, but differences are quite small. Biggest improvements are in the solar affected channels and the microwave channels where cosmic background is significant. RTM Comparison in GDAS: Operational and Parallel Analysis Runs

Operational Run Mean  T b HIRS Mean Observed – Guess  T b ; no bias correction All: Gross quality controlled data. Used: RT-dependent quality controlled data. (e.g. clear sky data for lower peaking channels) NOTE: Ch. 1, not assimilated

HIRS Mean Observed – Guess  T b ; no bias correction Parallel Run Mean  T b All: Gross quality controlled data. Used: RT-dependent quality controlled data. (e.g. clear sky data for lower peaking channels) NOTE: Ch. 1, not assimilated.

Operational Run Std. Dev.  T b HIRS Std. Dev. Observed – Guess  T b ; no bias correction All: Gross quality controlled data. Used: RT-dependent quality controlled data. (e.g. clear sky data for lower peaking channels) NOTE: Ch. 1, not assimilated.

Parallel Run Std.Dev.  T b HIRS Std. Dev. Observed – Guess  T b ; no bias correction All: Gross quality controlled data. Used: RT-dependent quality controlled data. (e.g. clear sky data for lower peaking channels) NOTE: Ch. 1, not assimilated.

-10 –2 – –2 – –1 – –1 – |  Tb(OP)| – |  Tb(NEW)| > 0  NEW is better|  Tb(OP)| – |  Tb(NEW)| < 0  NEW is worse  Tb(NEW) = Tb(NEW) – Tb(Obs)  Tb(OP) = Tb(OP) – Tb(Obs) HIRS Ch.18 comparison, no bias correction

-10 –2 – –2 – –1 – –1 – |  Tb(OP)| – |  Tb(NEW)| > 0  NEW is better|  Tb(OP)| – |  Tb(NEW)| < 0  NEW is worse  Tb(NEW) = Tb(NEW) – Tb(Obs)  Tb(OP) = Tb(OP) – Tb(Obs) HIRS Ch.18 comparison, with bias correction

RT transmittance coefficient generation

Memory requirement for OPTRAN coefficients becomes prohibitive for high resolution IR sensors. – Currently, OPTRAN requires 5400 available coefficients for each channel; 6 coefficients (offset + 5 predictors) for 300 absorber layers for each absorber (wet, dry, ozone). – Assimilation of 431 channels would require ~20MB memory simply for coefficient data. – Problem exacerbated if an increase in the number of absorber layers or predictors is warranted, or more channels assimilated. Mr. Yoshihiko Tahara, visiting scientist from JMA, is investigating a different method – within the OPTRAN framework – to predict absorption coefficient and transmittance profiles. New method fits the vertical absorption coefficient profile and this reduces the need for a large number of coefficients. New Transmittance Algorithm

The number of regression coefficients is significantly decreased. – For a polynomial order of 10, the number of coefficients is ~200 per channel. No interpolation required in generating regression coefficients or predicting absorption coefficient. Harder to fit LBL absorption coefficients at all levels. Polynomial fit to absorption coefficient

layer k’ (new) layer k (org) New absorption coefficient k’ k’ has smoother profiles than k. k’ can be negative. NOAA/HIRS Ch.3 Dry gas effective  Absorption coefficient

Absorption coefficients should be predicted accurately over highly sensitive layers for accurate radiance calculation. The sensitivity is used as the weight of regression coefficients. The weighting method saves LBL information lost by introducing polynomial fitting. NOAA/HIRS Ch.6, Dry Gas layer sample weight Weighting Regression Method

Index for predictor selection – RMSE of predicted transmittances against LBL has been found to be a better index for selecting predictors rather than that of predicted T b. Stable calculation – Many regression coefficients sometimes cause unstable calculation. – Careful selection amongst highly correlated predictors is needed. – Not always 5 predictors are needed for wet and ozone gas. RMSE Variation for Predictor Sets; NOAA/HIRS Ch.3 How to select predictors

Mean Error w/ No Bias Cor. Error Map w/ No Bias Cor. NOAA-14/HIRS Ch.9 New Original S.D. SD = 1.49k Mean Err. = -1.20kMean Err. = -1.76k SD = 1.64k

NOAA-14/HIRS Ch.4 New S.D. Mean Error w/ No Bias Cor. Error Map w/ No Bias Cor. Original SD = 1.54k Mean Err. = +0.22k SD = 1.42k Mean Err. = +0.71k

NOAA-14/HIRS Ch.17 New S.D. Mean Error w/ No Bias Cor. Error Map w/ No Bias Cor. Original New is BetterOriginal is Better

RT line-by-line transmittance generation

Gearing up system for generating line-by-line transmittances routinely. (  W recommendations?) Profiles anyone….? (please) HITRAN changes, LBL algorithm changes, profile set changes, etc. occur frequently. Goal is to make the operation as simple as possible. Software exists; just needs to be assembled. – Profile units conversion code – LBL input file generation code – LBL convolution code – Data readers (native LBL and netCDF formats) Moving dependent data (e.g. atmospheric profiles, instrument SRFs, final LBL and convolved transmittances) into netCDF format. LBL transmittances

Impact of spectroscopic changes Plot provided by Dave Tobin and Dave Turner at CIMSS/SSEC/UWisc. HIRS ch.10 FWHM

Issues to be addressed/further work

Profile data supplied with AIRS dependent set transmittances were layer column densities in kmol/cm 2. Dependent profile set “made up” so level profiles do not exist. These values were converted to ppmv and then exponentially interpolated to the level pressures. This introduced small, subtle but still significant differences. These level profile sets were used in generating OPTRAN coefficients for AIRS. RT performed using both profile sets on “truth” transmittances: Profile units conversion/interpolation (1)

Profile units conversion/interpolation (2) N 2 amount (ppmv x1.0e05) Pressure (hPa) N 2 profile (same for all climatologies) Layer Level Once level profile values in ppmv (or whatever units) are converted to column density (integrated layer quantity), the result should not be converted back to ppmv.

Profile units conversion/interpolation (3) Must ensure the column density calculation is consistent with the LBL code.

GOES Sounder channel 2 dry coefficients are responsible for artifacts for high absorber (== large zenith angle) Coefficient issues – GOES Sounder Effect only seen in the parallel GDAS runs using the updated RT code. Operational GDAS results o.k. Off-line tests show old RT code also exhibits problem. Problem is with top-of- atmosphere. Coefficient problems not seen below 2.4hPa. Large view angle causes transmittance anomaly to “migrate” down.

Coefficient issues – GOES Imager GOES Imager coefficients are not valid for large view angle (e.g. beyond 55 ° ) or high absorber amount. Known problem, but a lot of good data is getting thrown out. Plots provided by Xiujuan Su at NCEP/EMC. Zenith angle vs.  T for G10 IMGR ch4  T (obs-calc) T62 – lower TOA boundaryT254 – higher TOA boundary

Coefficient issues – GOES Imager GOES Imager channel 3: case where high absorber amount occurs at relatively small angles. Appears to have a TOA boundary component also. “Rings” appear in temperature residual images.  T images for G10 IMGR ch3 T62 – lower TOA boundaryT254 – higher TOA boundary