HIGH-RESOLUTION ASCAT SCATTEROMETER WINDS NEAR THE COAST

KNMI is involved in the Level 2 scatterometer wind processing of the EUMETSAT Ocean and Sea Ice Satellite Application Facility (OSI SAF), the EUMETSAT Numerical Weather Prediction SAF and the EUMETSAT Advanced Retransmission Service (EARS). In the NWP SAF, scatterometer wind processing portable software packages are being made available freely. In the OSI SAF, currently four different wind products are available: OSI SAF SeaWinds 100-km product (operational status), OSI SAF SeaWinds 25-km product (under operational review), OSI SAF ASCAT 25-km product (pre-operational), and EARS ERS-2 25-km regional product (demonstration) that may be viewed at www.knmi.nl/scatterometer. The latter product is available within one hour and may be used for weather nowcasting. In such application, as well as other applications, winds near the coast and at high resolution are required. Therefore, KNMI attempts the development of ASCAT scatterometer wind products at higher resolution and nearer to the coast, which effort will be presented at the conference.

EUMETSAT develops an ASCAT radar backscatter product (L1) on a 25-km swath grid, a 12.5-km grid, and at full measurement resolution. The two former products are achieved by applying spatial averaging kernels to the latter product, i.e., respectively of 50-km and 25-km resolution. These averaging kernels are used to suppress noise in the measurements, but, on the other hand, prevent wind retrieval in coastal regions due to their spatial extent. Currently, the swath grids are applied rigorously and no ASCAT scatterometer winds are available in the first 80 km off the coast. Applying box averaging rather than kernel averaging results in scatterometer winds at closer proximity to the coastline. By replacing the Hamming filter kernels (i.e., cosine weighting function) with a simple box (i.e., constant weighting function over a limited radial distance), we can produce 25-km sampled winds, which, in contrast to the 25-km sampled (nominal-resolution) level 2, will provide sea-surface wind information up to 25 km off the coastline, at 35- to 40-km resolution. Moreover, these boxes may be optimally chosen such that they are located close to the coast, but still not contaminated by land. In particular, by using a non-regular swath grid near the coast, we can maximize the number of wind observations at shorter distance off the coast. Later on, these different swath gridding and spatial averaging strategies will be applied to the 12.5-km product and processing may be sustained at even shorter distances to the coast. We expect high resolution ASCAT coastal winds may be produced down to 15 km off the coast.

Different spatial averaging strategies may allow more noise in the L1 data and thus in the L2 retrieved winds. To suppress this random noise KNMI has developed spatial filtering techniques, which maintain small-scale meteorologically-relevant spatially-coherent structures in the resulting scatterometer wind fields. This filter, by the so-called Multiple Solution Scheme (MSS) and 2-Dimensional Variational Ambiguity Removal (2D-VAR), will be illustrated. The MSS collects additional information from the scatterometer wind inversion step, i.e., information on the probability of all possible winds, as retrieved from the input local backscatter measurements. This wind vector probability distribution at the swath grid is subsequently used as input to the 2DVAR, which provides a meteorologically balanced and spatially coherent wind field.