*1.2. ECMWF Reanalysis*

ECMWF is the European Centre for Medium-Range Weather Forecasts, this organization is advancing numerical weather prediction through global collaboration. The ECMWF reanalysis dataset was chosen for di fferent studies, because of its high temporal and spatial resolution [36]. ECMWF dataset covers a vast timespan starting from 1979 to the present time [37]. ECMWF Reanalysis datasets, including ERA5, ERA-interim, ERA-interim/land, CERA-SAT, CERA-20C, ERA-20CM, ERA-20C. Furthermore, ECMWF data gathers information from the global observation system comprehending di fferent kind of satellites, meteorological, buoys, weather stations and ships.

These limitations are mentioned above, that do not include RS satellite technology and reanalysis datasets [38], such as ECMWF, MERRA, NARR and ERA. A reanalysis dataset is the only source resolved long-term information of spatial wind information at wind turbine height. It provides data and potential assessment of this sector strongly related to the quality form of the meteorological situation [39]. Due to high costs and limitations of the measuring device, including coastal stations, buoys, ships, masts, the use of this kind of dataset can give possibility to understand wind speed with good quality in wind farm site assessment. This step is a very important criterion in possible

alternatives that can be provided by high-resolution reanalysis data. Furthermore, this kind of analysis can reduce the cost of on-site wind measurement [39]. Arun Kumar et al. [40], described the limits of offshore buoys, pinpointing that wind source assessment would be a challenging task. One of the best alternatives for inexpensive data and for filling the data gaps by providing a huge volume of data for extended periods is satellite RS. The assessment of wind source from single scatterometer could lead to inconsistency where there is a requirement of multiple satellite scatterometer. Therefore, four scatterometers viz. OSCAT, ASCAT-A, ASCAT-B and QuikSCAT with long-term in situ wind datasets over the North Indian Ocean were considered. It has been observed that QuikSCAT and OSCAT wind data have lesser bias with the range of 0.15 m/s (2.4%) to 0.83 m/s (15.1%) before adjustments. Linear regression was used for adjustment and the synergetic approach of linear regression adjustments and the combination of scatterometer data have resulted in smaller differences.
