|
A crucial issue of climate reanalysis is the impact that
discontinuities in the input observations may have
on the time series of the assimilation system.
Observing system changes can lead to spurious
trends or even discontinuities in the assimilation output.
As part of the MERRA testbed, we examine the impact of changes in the
observing system on a short retrospective analysis.
Data impact studies will be one of NASA contributions
to the US Climate Change Science Program
(CCSP).
|
SSM/I Impact Test
In this section we examine the impact of removing
SSM/I observations
on the GEOS4 assimilation system.
At present we use the GEOS4-CERES experiment as a control analysis
and placeholder.
The GEOS4-CERES reprocessing was performed to support the
CERES instrument team research efforts; it is not publicly available.
We will move experimentation and validation to GEOS5
when the system reaches the appropriate level of maturity.
Click on any image to expand it in a separate window.
Taylor plot of SSM/I Data Impact
In the Taylor plot above we compare how closely two different model runs
match to verification data from Global Precipitation Climatology Project
( GPCP ).
The two runs are GEOS-4 assimilations, with (dot) and without data from the
Special Sensor Microwave/Imager ( SSM/I ).
A standard Taylor diagram displays three statistics:
- the correlation coefficient between the observed and simulated field (angle),
- the normalized root-mean-square (RMS) difference between the two fields (distance from the 1:1 point), and
- the normalized standard deviation of the fields (radius).
Note that the 1:1 point, in the middle of the x-axis, denotes absolute agreement.
In addition, this Taylor plot is color-coded by month and uses a central dot
to show the shift in data points with (dot) and without SSM/I data.
Back to top of page
EOS Observations and Climate Analysis
NASA EOS provides many new or enhanced data types over the time period
from 1998 to the present.
Introducing new data into a long reanalysis may have unexpected
or even spurious results in evaluation of the climate,
climate variability or trends.
While we hope to take full advantage of the new observations,
the impact of these data need to be assessed scientifically
in a stepwise manner.
The baseline MERRA analysis will incorporate only minimal EOS observations.
This baseline will provide the starting point for new branches of the data
assimilation that include new data types from the EOS platforms.
For example, the Moderate Resolution Imaging Spectroradiometer
( MODIS )
instrument will produce aerosol, cloud and surface temperature data, not
currently used in the data assimilation system.
When the methods for assimilating these data are ready, we can produce
long reprocessing runs including the data, with a well-documented
control system. This can then be used to assess the impact of the data
on the climate time series.
This aspect of the project will likely not begin until a significant portion
of the baseline analysis is complete and evaluated (perhaps mid to late 2007).
Further, it will also depend on the availability of computing resources
at that time, as well as the maturity of the individual reserarch projects
developing the new methods.
Back to top of page
Date Last Modified: 04/23/08
|
