| Satellite radiance observation
data sources |
|
| |
| NOAA-15, NOAA-16, NOAA-17 and NOAA-18 ATOVS data
are received from NOAA/NESDIS
in level 1B format as described in the NOAA-KLM
user guide. Briefly level 1B data consists of raw counts, calibration
coefficients and navigation information as well as a complete set
of instrument "housekeeping" data. |
| |
| Data processing at the Met Office |
|
| |
- Level 1B ATOVS data are calibrated and mapped to the HIRS
instrument grid using the AAPP
software system which was developed by NMS in Europe, primarily
the Met Office, Météo-France and KNMI, in collaboration
with EUMETSAT. AAPP also runs tests which detect cloud and precipitation
effects on AMSU channels (see below)
and which provide an estimate of surface type which can be used
alongside a land/sea mask. The resulting dataset is known as
level 1D ATOVS and it is these radiances which are monitored
in these pages.
- Prior to assimilation in 4Dvar the level 1D ATOVS radiances
are used in a 1Dvar analysis. This allows refinement of the
quality control, especially for cloud in the infra-red (see
below) and analysis of skin temperature and temperature
above the current NWP model highest level. The 1Dvar preprocessor
also allows radiances to be monitored against the NWP background
(referred to as "observation-background") and against the 1Dvar
analysis (referred to as "retrieval-background").
|
| |
|
|
| |
| The time period for each point in the time series
is labelled according to the beginning of the 24 hour period. The
24 hour period is from 2100Z to 2100Z, i.e. the point for the 24th
covers data from the 24th 2100Z to the 25th 2100Z. |
| |
|
|
| |
- Clear: Clear in both the
microwave and infrared channels as defined below.
- Cloudy in microwave:Based
on the AMSU cloud cost generated in AAPP. Microwave cloud cost
is a maximum likelihood method based on microwave window channels
consistent with each other and a clear climatology. See reference
English et al. The AAPP module for identifying ice cloud, liquid
cloud and surface type on the AMSU-A grid, Technical Proceedings
of the Ninth International TOVS Study Conference, Igls, Austria,
20-16 February 1997
- Cloudy in infrared: Infra
red cloud cost is a maximum likelihood method based or HIRS
4-8,13-15, AMSU 4-5 being consistent with each other and a model
of short range forecasts, See reference English et al. A cloud
detection scheme for use with satellite sounding radiances in
the context of data assimilation for numerical weather prediction.
Q. J. R. Meteorological society, Vol. 125, pp 2359-2378.
|
| |
|
|
| |
- Land: A surface is classified
land if the altitude is > 0 and the ATOVPP surface classification
is land or when ATOVPP land classification is not sea and the
radiance classifier is not sea and the latitude is < 55 deg.
- Sea: A surface is classified sea if the altitude is
< 0 and the seaice fraction is < 0.2 and radiance classifier
and ATOVPP land classification is sea.
- Sea ice: A surface is classified sea ice if the sea
ice fraction is < 1 and > 0.2. or it is not land but < -72 deg.
latitude (to catch Ross & Weddel Sea ice shelves) or when the
ATOVPP land classification is sea and radiance classifier is
not sea and the latitude is greater 55 deg.
|
| |
|
|
| |
- Observed-Background: A comparison
is made for each available satellite of the difference between
observed (but preprocessed and in the case of TOVS adjusted
to nadir) brightness temperatures and the equivalent brightness
temperature calculated from background profile(from The Met
Office Model).
- Corrected-Background: As for
the above except that the observed brightness temperatures have
been bias corrected. See reference Eyre, A bias correction scheme.
for simulated TOVS brightness temperatures, ECMWF Tech Memo
No. 186, October 1992.
- Retrieved-Background: The
background profile and (bias-corrected) observed brightness
temperatures are input to a 1D-Var (one-dimensional variational
analysis) scheme which returns a statistically-optimal profile
of temperature and humidity. Brightness temperatures are calculated
from this profile and here are compared against the profiles
calculated from the model background.
|
| |
|
|
| |
- Gross: the only quality control
on these obs has been to require that the observation brightness
temperatures lie between 90 and 340 K.
- Strict: The quality control here applies all cuts outlined
in global
rejection plots and is closest to what is assimilated into
the Met Offices numerical weather prediction scheme.
|
| |
|
|
| |
| See the combined observations
and forward model error variance (sometimes referred to as O+F)
and the background error mapped into radiances space |
