...
| Expand | ||||||||
|---|---|---|---|---|---|---|---|---|
| ||||||||
|
...
The validation strategy is described in [D1] section 2.4. We use the bias, i.e. mean difference between Cloud_cci and the reference data, as the metric for accuracy. The bias corrected root mean squared error (bc-RMSE) is used to express the precision of CDR compared to a reference data record. The stability is the variation of the bias over a multi-annual time period.
Bias (accuracy): | Mean difference between Cloud_cci and reference data |
bc-RMSE (precision): | Bias corrected root mean squared error to express the precision of Cloud_cci compared to a reference data record |
Stability: | The variation of the bias over a multi-annual time period. |
TCDR evaluation is divided into: (i) validation against high quality and satellite-based reference observations (CALIOP, DARDAR and AMSR-E) and (ii) an intercomparison with well-established, satellite-based cloud datasets of similar kind (MODIS).
...
The validation results for TCDR are provided in [D1] section 7, together with recommendation for use. Here is an extract from table 7-2 in [D1]:
Comments from [D1] | |||
Cloud fractional cover | Accuracy |
| Level-2 validation against CALIOP |
|
| Values taken from Table 4-2 and Table 4-13 (L3C comparisons to MODIS C6.1) | |
Cloud top height/ pressure | Accuracy | 0.12km (liquid cloud) | Level-2 validation against CALIOP |
Stability (per decade) |
| Values taken from Table 4-4 and Table 4-15 (L3C comparisons to MODIS C6.1) | |
Liquid cloud optical depth | Accuracy | n/v | No validation possible due to a lack of reliable reference data. through LWP and IWP validation |
Stability (per decade) |
| Values taken from Table 4-5 and Table 4-16 divided by mean MODIS C6.1 Terra COTliq (13) (L3C comparisons to MODIS C6.1) | |
|
|
| No validation possible due to a lack of reliable reference data. through LWP and IWP validation |
|
| Values taken from Table 4-6 and Table 4-17 divided by mean MODIS C6.1 Terra COTice (10) (L3C comparisons to MODIS C6.1) | |
|
|
| Level-2 validation against AMSR-E (Figure 3-1) |
|
| Values taken from Table 4-9 and Table 4-20 divided by mean MODIS C6 LWP (123g/m²) (L3C comparisons to MODIS C6) | |
Ice water path | Accuracy | -39.9% | Level-2 validation against DARDAR |
|
| Values taken from Table 4-10 and Table 4-21 divided by mean MODIS C6 IWP (208g/m²) (L3C comparisons to MODIS C6) | |
|
|
| No validation possible due to a lack of reliable reference data. through LWP and IWP validation |
|
| Values taken from Table 4-7 and Table 4-18 (L3C comparisons to MODIS C6) | |
|
|
| No validation possible due to a lack of reliable reference data. through LWP and IWP validation |
|
| Values taken from Table 4-8 and Table 4-19 (L3C comparisons to MODIS C6) |
Intercomparison (using the monthly mean data from January 2017 to December 2019) of ICDR products with MODIS present biases consistent with values found through the TCDR MODIS comparison (D1, section 4.1) for cloud fractional cover (- 5 %), cloud top pressure (-15 hPa) and liquid water path (-15.0 g/m2), but higher values for ice water path, with global average bias of -31 g/m3 for SLSTR-A and -46 g/m3 for SLSTR-B (TCDR IWP bias with MODIS was -29 g/m3).
...