GRAS SAF Climate Products
description
Transcript of GRAS SAF Climate Products
11
GRAS SAF Climate Products
Hans Gleisner & Kent B. Lauritsen
Danish Meteorological Institute
-----
Contents- GRAS SAF offline profiles and climate gridded data- Status of products- Monitoring and a priori assessment- 1D-Var diagnostics- Prototype data from Metop NRT data
22
Overview of GRAS SAF climate gridded data
Climate data product 2D zonal grid: 1
climate + errors Time
resolution Spatial 2
resolutionFormats,graphical
Formats,numerical
CBA: bending angle yes Monthly 5 deg latitude PNG, JPGASCII, netCDF
CRG: refractivity yes Monthly 5 deg latitude PNG, JPGASCII, netCDF
CTE: temperature yes Monthly 5 deg latitude PNG, JPGASCII, netCDF
CHG: spec. humidity yes Monthly 5 deg latitude PNG, JPGASCII, netCDF
CZG: geopot. height yes Monthly 5 deg latitude PNG, JPGASCII, netCDF
1 A latitude-height grid where the height can be expressed in MSL height, geopotential height, or in terms of pressure.
2 The maximum resolution in height is determined by the height resolution of the profiles.
Offline Climate gridded data products – an enhancement of GRAS offline data.
33
Some recent achievements
• Delivered a complete RO climate data set based on CHAMP data Sept 2001 - Sept 2008 for the international ROtrends comparison study, November 2010;
• Participated in the Mid-term meeting of the ESA DUE GlobVapour project, ESRIN, Frascati, Italy, 7 March 2011;
• Participated in the GEWEX/ESA DUE GlobVapour workshop on long term water vapour data sets and their quality assessment, ESRIN, Frascati, Italy, 8-10 March 2011;
44
NRT, Offline, and Climate processing overviewPhase, amplitude,
ground station observations,near-real time orbits
Phase, amplitude,ground station observations,
NRT/offline data
Bending angle profiles(L1, L2, LC)
Refractivity profiles
1D-Var algorithm
Ancillarytemperature, pressure,
and humidity, fromECMWF forecasts
Temperature,pressure, and
humidity profiles
GRAS SAFNRT Products
Bending angle profiles(ionosphere corrected and
statistically optimized)
Refractivity profiles
1D-Var algorithm
Temperature,pressure, and
humidity profiles
GRAS SAFOffline Products
Level 1a
Level 1b
Level 2
Produced byEUMETSAT
Geometric optics inversion algorithm
Bending angle profiles(statistically optimized)
Abel transformalgorithm
CT2algorithm
Abel transformalgorithm
Level 2
Level 2
Re-processed data(zero/single/double diff.):
Phase, amplitude
Other RO data(COSMIC, CHAMP, ...):
Phase, amplitude
Bending angle profiles(ionosphere corrected and
statistically optimized)
Refractivity profiles
1D-Var algorithm
CT2algorithm
Abel transformalgorithm
Temperature,pressure, and
humidity profiles
Climate algorithms
Bending angle, refractivity,temperature, humidity, andgeopotential height grids
GRAS SAFClimate/Gridded Data
55
Status of offline profile and gridded products
Offline products from Metop: offline profiles and gridded climate data- first version based on prototype offline GRAS/Metop-A data via ftp from
EUMETSAT CAF planned from: June 2011- consolidation of new format, netCDF-3 or 4: June - August 2011- demonstration product may be made available for download: Q1, 2012
Offline products from COSMIC: offline gridded climate data- offline processing based on ROPP_PP has started- internal validation planned for Sept 2011- review for offline gridded data: end 2011- operational product made available for download and monitoring: Q1, 2012
66
- During 2010 (following the PCR-2 review) we have implemented algorithms to:
- monitor noise on the bending angles- monitor stability of observed errors in refractivity- monitor stability of the errors the 1D-Var a priori- quantify the relative importance of a priori in the refractivity- quantify the relative importance of a priori in 1DVar temperature & humidity
- The climate algorithms are now described in the Algorithm Theoretical Baseline Document (ATBD): Climate Algorithms, ver. 2.1.
Status of offline climate data products
77
Monitoring noise on bending angle: Algorithm
The neutral-atmosphere bending angle is contaminated with noise that increases exponentiallywith altitude. The noise is of both instrumental and ionospheric origin and varies considerablyfrom occultation to occultation.
We estimate the upper-level bending angle noise by the smallest standard deviation ofthe bending angle difference obs-clim found over a scale height (here, taken to be 7.5 kilo-
meters) in the interval 60 to 80 kilometer:
n
n 1k
2kclim,kobs,1
1min
Here, n is the number of data points within a sliding window of 7.5 kilometer width. obs is the observed bending angle.
clim is the corresponding bending angle from the MSIS climatology.
88
Monitoring noise on bending angle: Example
Biases (left panel) and standard deviations (right panel) of the bendingangle differences obs-clim in CHAMP data during the year 2004.
99
For each grid-box and month we compute the quantity:
j2
jobs,2jbg,
2jobs,
iSO
1
MW
where obs and bg are estimates of the errors in the observed and background
bending angles, and index j loops over the Mi data points in grid box i.
This quantity provides a measure of the observational information in the optimizedBending angles. As a consequence of the error characteristics, it goes from 0(no background) at low altitudes to 1 (no observational information) at high altitudes.
Relative importance of a priori in refractivity: Algorithm
1010
Relative importance of a priori in refractivity: Example
1111
Relative importance of a priori in 1DVar T, q: Algorithm
We quantify the relative importance of the a priori information in the retrievedtemperature and humidity by the error standard deviations used in the 1DVar retrieval:
bgT,
solT,T 100
W
bgq,
solq,q 100
W
where index sol denotes solution and index bg denotes background. The factor 100normalizes the ratio to percent.
1212
Relative importance of a priori in 1DVar T and q: Example
1313
1D-Var minimization: Diagnostics
The state xs that minimizes J(x) is a valid estimate of the true atmosphere only if
the error covariance matrixes O and B provide good enough descriptions ofthe actual errors. These errors are not known perfectly known. Desroziers et al. [2005] described how information on the errors can be gainedfrom the statistics of the differences between observation, background, andsolution. In observation space:
))(())((2
1)()(
2
1)( o
1Tob
1Tb xyOxyxxBxxx HHJ
bobo xyd H
soso xyd H
bsbs xxd HH
1414
If the errors are unbiased, Gaussian, and accurately describe the true errors,then the following set of relations should hold:
The diagnosed error covariances are on the left hand side. The error covariances assumed in the 1D-Var retrieval are on the right hand side. H is the Jacobian of H(x).
Consistency criteria for 1D-Var errors
1515
Diagnosing 1D-Var errors
The consistency criteria provide a means to monitor the stability of errors, e.g. by aregular diagnosis of the mean error covariance diagonal elements.
i
1jjb,jo,jb,js,
i
2ib,
1 M
yyyyM
i
1jjb,jo,js,jo,
i
2io,
1 M
yyyyM
Here, index i denotes a latitude band and index j loops over the Mj observations in
latitude band i. If the observed quantity is refractivity, which falls off exponentiallywith height, the diagnosed errors are more conveniently expressed in relative terms:
ib,2ib,
diagib, 100 y
io,2io,
diagio, 100 y
1616
Diagnosed background and observational errors for COSMIC-FM4 in March 2008.
Diagnosing 1D-Var errors
1717
Metop EUM-BA O-B/B
1818
Metop EUM-BA O-B/B
1919
Metop REF O-B/B
2020
Metop REF & T: Polar bias
2121
Metop REF & T: Polar bias May 2011
2222
Reprocessing and construction of climate data sets
Major activity in the next phase of the GRAS SAF: Reprocessing of all RO data planned in CDOP-2 in 2014 and 2016 with new algorithms and improved input data:-GRAS data from EUMETSAT CAF-ERA-Clim RO data from EUMETSAT CAF (CHAMP, COSMIC, …)-RO data from CDAAC
Applications and data:- Include QC info, error estimate and usable range; comparisons of reprocessed datasets within a reanalysis system;- Produce RO datasets for testing forecast and climate models (this is more difficult for radiances because they are bias corrected to the models; RO data is assimilated without bias correction);- Produce RO datasets for climate monitoring (RO information content is highest in the upper troposphere/lower-mid stratosphere);
2323
Overview of reprocessing and validation
Overview of GRAS SAF reprocessing, interfaces, and validation:
WebsiteUsers
Phases, amplitudes, orbits(GRAS, ERA-Clim)
ReprocessingGRAS SAF
EUMETSAT CAF
RO Processing Centers
ECMWF NWP/Reanalysis
ROtrends intercomparison
SAF’s (CM SAF)
ESA DUE GlobVapour
GEWEX Radiation Panel
Intercomparison:- BA, REF, T, P, q (profiles)- Climate data (grids)
Intercomparison:- BA (level 1b)
Offline and climate data
RO Data providers
Phases, amplitudes, orbits
2424
fin