Comparative analysis of WRF simulated surface winds with satellite observations over the Arabian Sea

Accurate estimation of meteorological parameters is crucial for the successful implementation of any operational oceanographic service. Surface wind information serves as a primary forcing in circulation models. Despite being an important aspect of climate studies, there remains a serious scarcity of extended in-situ observations, especially along the Indian coastline. Space-based observation of surface winds is limited by their spatial and temporal resolutions. In such a scenario simulation-based studies can be considered as the missing link between the satellite information and scarce in-situ observations. The present study attempts to understand the spatial variation of 10m winds simulated by the WRF (v-4.0) model against that Scatsat-1 scatterometer for the period of 20^th  -29^th June 2019. The entire study was conducted for the Arabian Sea region comprising the west coast of India. WRF-ARW was forced with two different initial conditions, NCEP-FNL  and GFS . NCEP-FNL initial conditions have a spatial resolution of 1° × 1° and a temporal resolution of 6 hours. GFS data have a spatial resolution of 0.25° × 0.25° and it is available at 3 hourly intervals. The model is simulated for the entire month of June 2019. Comparative analysis is carried out with 10m wind speed of Scatsat-1 level-4 analyzed winds from 20^th June to 29^th June as the satellite information was available only for that period. The analysis is carried out for the inner nested domain of 14km. The model simulated 10 m wind speed is spatially interpolated onto the size of the Scatsat-1 grid of 25km for spatial comparison. Spatial comparison 10 m wind speed of model simulation with NCEP-FNL initial conditions with that of Scatsat-1 revealed a high spatial correlation (0.6-0.7) between in the open ocean region but lower spatial correlation near the coastline. Similar analysis of model simulation initialized with GFS data showed a reasonably good spatial correlation in the open ocean but very low correlation along the nearshore region. This discrepancy can be attributed to the error in wind speed estimates of the satellite observations due to higher backscatter near the coastline. This indicates the inability of the model to represent the complicated topography of the study area coastline. The two different initial conditions reflected different patterns in spatial correlation due to the slight difference in the mode of generation of the two datasets. NCEP-FNL data is known to ingest 10% of more observational datasets than GFS, which might have reflected in the analysis. The entire analysis was conducted at a spatial resolution of 25km which again can be considered as a limiting factor. Hence it is expected that simulating the model at a high spatial resolution will resolve the complex topography of the nearshore region with improved accuracy.