A trend-preserving statistical downscaling framework and its application to China’s offshore wind field

Perfect prognosis (PP) statistical downscaling and model output statistics (MOS) bias adjustment methods are commonly utilized to generate regional to local climate change projections from coarse global model outputs. However, various sources of uncertainty associated with PP may produce implausible projections and alter the original global climate model signal under certain circumstances. The MOS approach is preferred only when the predictor and predictand resolution gap is small. This study combines PP and MOS to establish a trend-preserving statistical downscaling framework for gridded wind fields in China’s offshore regions. Our findings demonstrate that although different predictors resulted in similar outcomes for historical data, they led to substantial differences in future projections derived from PP methods. Using MOS in a postprocessing step for PP can drastically enhance the performance of the downscaling framework. The downscaled results indicate decreases in the offshore sea surface wind speed over the Bohai Sea and East China Sea and small increases over the South China Sea until the end of the century. The proposed framework can be applied to other regions.