Systematic errors in simulated Doppler wind lidar observations
G.D. Emmitt
2/17/2000
This memo addresses the issue of simulating systematic errors in the DWL observations. The
current plan for the NCEP OSSE effort is to assess the potential impact of DWL concepts on
global analyses and forecasts. The strategy is to define a range of global coverage and
observational accuracies that "bracket" conceivable and reasonable DWL
technologies/measurement combinations. A set of "bracketing" OSSEs has been developed
and endorsed by the oversight committee. The NCEP OSSE strategy is to proceed through
three tiers of impact assessment. Those three levels are:
-
Range of global coverages, clouds permitting
-
assumes near perfect LOS measurements where clouds permit
coverages will range from full tropospheric (0-20km)/2000 km swath to partial
profiles along a single ground track
-
Range of observational accuracies (Gaussian)
-
Gaussian errors are added to the near perfect LOS observations simulated for
the coverage OSSEs
errors will vary from <1m/s RMS to ~ 7-10m/s RMS
-
Range of systematic errors peculiar to DWL technologies
-
reprocessing of the simulated DWL for the coverage OSSEs is required to
simulate systematic errors
systematic errors are those which are characterized by correlated errors along a
LOS or set of LOSs
Types of systematic errors
For wind observations, we are most interested in the LOS velocity and its height
assignment. By systematic errors we mean the component of the total observational error that
is common (highly correlated) to many LOS wind estimates in space and/or time. The only
systematic errors that we are concerned with here are those that we know can occur but have
no way of knowing when they occur or how large they are in any given instance.
-
Systematic errors fall into four categories:
-
instrument/technology unique biases
platform caused biases
sampling biases that are atmosphere independent
measurement biases that are related to atmospheric conditions
Instrument unique
-
Here are a few examples of DWL technology unique systematic error sources
-
LO offset is incorrect. Can be caused by poor attitude or position (lat/lon)
information.(applies to coherent detection only)
-
Solar backscatter (applies to direct detection only)
-
Scanner position errors(angular knowledge errors)
-
Azimuthal dependent point knowledge errors
-
Temperature induced pointing errors
Platform induced
-
In most DWL space mission concepts, the attitude and position information are
provided by the platform. Some examples of platform induced errors are:
-
attitude offsets
-
positional errors (lat, lon, height)
-
jitter
Atmosphere independent sampling biases
-
These sources of systematic errors are related more to the scanning strategy than to the
instrument velocity detection scheme. Two examples are provided:
-
a non-scanning system that only measures the cross-track component of the
wind
-
a < 100% duty cycle instrument that is operated in a atmospheric situation de-
coupled manner
Atmospheric/earth surface situation biases
-
Being from optical sensors, there are several ways that DWL observations can be
biased in sampling. While sampling biases are not unique to future lidars in space, it is still
important to understand the nature of these opportunity-related biases. Some examples of
atmospheric sampling biases are:
-
LOS observations clouds permitting
-
LOS observations aerosols permitting
-
LOS propagation effects related to the refractive index
-
LOS observations in regions of very strong vertical motions (orographic,
thermal)
-
-
Surface motions which confound using surface returns to calibrate velocity
estimators
-
Ocean surface currents
-
Ocean waves
-
River currents
-
Blowing sand or snow
-
Swaying canopies
-
Proposed strategy for investigating the impacts of systematic errors
-
The options for including systematic errors in the simulated DWL data sets are
boundless. I recommend the following initial attempt to bound the issue.
-
Add a pointing knowledge offset to all observations that will produce a LOS velocity
estimator error of 1 m/s.
-
Add a pointing knowledge error only for those LOSs that do not contain a ground
return
-
Add a height assignment error to those LOSs not containing a ground return
-
Add an error that is a function of solar exposure