This study compares the performance of different models used to assess the local wind near-surface conditions at an agricultural site in Scania, southern Sweden. The methods are: (a) the WASP model (Wind Analysis and Application Program), (b) separate linear regressions of the two wind vector components, (c) a regression model based on vector correlation, and (d) linear regression of scalar wind. Each method was tested with three different data sets over nine months: wind measurements from the nearby Sturup airport SYNOP station, 10 m surface wind and surface geostrophic wind produced by the operational Swedish Mesoscale Analysis system (Mesan). The wind climate estimations were compared with observed winds at the field site, with respect to mean wind speed, wind direction, wind speed frequency distribution and the relative frequency of winds above 6 m s−1. All models performed reasonably well with data from Sturup and Mesan surface wind, but gave less reliable results with the Mesan geostrophic data. The estimated frequency of winds above 6 m s−1 was in general lower than the observed frequency. Overall, best results were obtained with WASP in combination with measurements from Sturup.