The cost function of the data fusion process and its application
<p>When the complete data fusion method is used to fuse inconsistent measurements, it is necessary to add to the measurement covariance matrix of each fusing profile a covariance matrix that takes into account the inconsistencies. A realistic estimate of these inconsistency covariance matrices...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2019-05-01
|
| Series: | Atmospheric Measurement Techniques |
| Online Access: | https://www.atmos-meas-tech.net/12/2967/2019/amt-12-2967-2019.pdf |
| Summary: | <p>When the complete data fusion method is used to fuse
inconsistent measurements, it is necessary to add to the measurement
covariance matrix of each fusing profile a covariance matrix that takes into
account the inconsistencies. A realistic estimate of these inconsistency
covariance matrices is required for effectual fused products. We evaluate
the possibility of assisting the estimate of the inconsistency covariance
matrices using the value of the cost function minimized in the complete data
fusion. The analytical expressions of expected value and variance of the
cost function are derived. Modelling the inconsistency covariance matrix
with one parameter, we determine the value of the parameter that makes the
reduced cost function equal to its expected value and use the variance to
assign an error to this determination. The quality of the inconsistency
covariance matrix determined in this way is tested for simulated
measurements of ozone profiles obtained in the thermal infrared in the
framework of the Sentinel-4 mission of the Copernicus programme. As
expected, the method requires sufficient statistics and poor results are
obtained when a small number of profiles are being fused together, but very
good results are obtained when the fusion involves a large number of
profiles.</p> |
|---|---|
| ISSN: | 1867-1381 1867-8548 |