Indian Ocean AMVs: Moving to Meteosat-8 and assessing alternative options

On 1st February 2017, EUMETSAT approved the move from Meteosat-7 to Meteosat-8 as the primary Indian Ocean Data Coverage (IODC) service. Now in this new location, Meteosat-8 was the natural choice for the replacement of the Meteosat-7 Atmospheric Motion Vector (AMV) and Clear Sky Radiance (CSR) products in the ECMWF operational system. The switch means changing to a newer generation satellite leading to a significant increase in the amount of data available for the IODC in both products and, using Meteosat-10 as an example of the newer generation, better data quality was also anticipated. The first part of this report focuses on the analysis of Meteosat-8 AMVs, initially looking at first guess departure statistics that show the expected improvement over Meteosat-7, such as reducing negative speed biases at high levels in the extra-tropics. Data quality was, reassuringly, very similar between Meteosat-8 and Meteosat-10. Assimilation experiments tested the longer term impacts of the new dataset in the forecast system. These revealed continued benefit of the IODC with positive impacts on the vector wind field at high levels and small reductions in the standard deviations of the fit of conventional wind observations and humidity sensitive observations to the model backgrounds.

The second part of this report investigates potential options for the IODC beyond Meteosat-8. Here we inter-compare a selection of satellites - Meteosat-8, Indian National Satellite - 3D (INSAT-3D) and Feng-Yun-2E (FY-2E) - operated by different centres and consider their relative benefits or limitations. First guess departure statistics are used to understand the data quality and show that the variation in instruments and AMV derivation methods lead to relatively large differences in the values and spatial patterns of the statistics. INSAT-3D overall showed the best agreement with the first guess but this could be explained by a larger dependence of NWP information in the wind derivation. Despite the differences between the datasets, impacts on the forecast system were surprisingly very consistent. For completeness, the benefit of the All Sky Radiances (ASRs), available for Meteosat-8 only, was also considered with assimilation experiments showing positive impacts on the fit of independent humidity sensitive observations to the model background.

During the first experiments conducted for the switch from Meteosat-7 to Meteosat-8 an unexpected area of apparent degradation at 850hPa was found in the middle of the Indian Ocean. Part 3 of the report considers this feature in more detail. The area was also found to be affected by model bias leading to doubt that this was an entirely observation based issue. More thorough investigation was carried out with aid of the other IODC satellites which has highlighted some potentially suspicious behaviour in the low level winds where the wind speed of the AMVs varies little with height compared to the model. Sparse conventional observations coupled with evidence that it is also challenging area for model have made it difficult to make firm conclusions about the truth.