80
Chapter 10
of CMW product accuracy is therefore not straightforward. Second,
radiosonde observations are sparse in both space and time. They occur at
best twice a day, mainly over Europe. Large areas such as the Atlantic
Ocean or the African continent are covered only with very sparse
radiosonde observations. Consequently, this method cannot be used to
detect the presence of systematic spatial biases in the CMW product.
2. Product monitoring against assimilated data. A second type of
validation mechanism is performed by ECMWF, comparing the bias and
standard deviation between the CMW product and the forecast winds.
These comparisons provide meaningful information on the spatial
consistency between these two fields. Once again however, the two
quantities do no represent exactly the same natural process and cannot
be used to estimate the product absolute accuracy.
3. Impact studies. The strongest validation test is performed by ECMWF.
Numerical weather forecasts are generated with and without the
assimilation of the CMW product. The validation is based on the impact
of the CMW on the forecast quality. This test verifies whether the
product accuracy matches the user’s requirement. When the user’s
requirements are met, the product assimilation has a positive impact on
the forecast quality.
Quality control and product validation are two important processing steps
of operational meteorological product extraction. Quality control is applied
everywhere and all the time, as the products are generated. The quality
estimation associated with the products ensures an optimal interpretation and
usage of the delivered information. Validation is an off-line activity, which
provides an indirect estimation of product accuracy. It is applied in a limited
number of cases and permits to monitor the effect of retrieval algorithm
modification and input data quality on the product reliability. If necessary,
validation results can be used to adjust quality control mechanisms.
It has been seen that quality control is of primary importance to increase
the intrinsic value of a product, providing a way to improve its
interpretation. Hence, the definition of new products as well as the
development of advanced retrieval algorithms should account for the
possibility to assign consistent quality scores. These issues will be addressed
in the next section in the context of the exploration of the MSG potential for
land surface monitoring.
Chapter 10
of CMW product accuracy is therefore not straightforward. Second,
radiosonde observations are sparse in both space and time. They occur at
best twice a day, mainly over Europe. Large areas such as the Atlantic
Ocean or the African continent are covered only with very sparse
radiosonde observations. Consequently, this method cannot be used to
detect the presence of systematic spatial biases in the CMW product.
2. Product monitoring against assimilated data. A second type of
validation mechanism is performed by ECMWF, comparing the bias and
standard deviation between the CMW product and the forecast winds.
These comparisons provide meaningful information on the spatial
consistency between these two fields. Once again however, the two
quantities do no represent exactly the same natural process and cannot
be used to estimate the product absolute accuracy.
3. Impact studies. The strongest validation test is performed by ECMWF.
Numerical weather forecasts are generated with and without the
assimilation of the CMW product. The validation is based on the impact
of the CMW on the forecast quality. This test verifies whether the
product accuracy matches the user’s requirement. When the user’s
requirements are met, the product assimilation has a positive impact on
the forecast quality.
Quality control and product validation are two important processing steps
of operational meteorological product extraction. Quality control is applied
everywhere and all the time, as the products are generated. The quality
estimation associated with the products ensures an optimal interpretation and
usage of the delivered information. Validation is an off-line activity, which
provides an indirect estimation of product accuracy. It is applied in a limited
number of cases and permits to monitor the effect of retrieval algorithm
modification and input data quality on the product reliability. If necessary,
validation results can be used to adjust quality control mechanisms.
It has been seen that quality control is of primary importance to increase
the intrinsic value of a product, providing a way to improve its
interpretation. Hence, the definition of new products as well as the
development of advanced retrieval algorithms should account for the
possibility to assign consistent quality scores. These issues will be addressed
in the next section in the context of the exploration of the MSG potential for
land surface monitoring.