HISTORICAL ATMOSPHERIC ANALYSIS BY WEATHER CATEGORY ASSIMILATION USING GAUSSIAN TRANSFORMATION

Abstract

 Before instrumental meteorological data are available, old diaries are an important source of daily atmospheric conditions. Data assimilation is a powerful technique for achieving reconstructions because it optimally combines observations with climate models. Although descriptive weather records can be approximately converted into cloud cover, it is difficult to assimilate cloud cover because of the non-Gaussianity of its error distribution. The objective is to reconstruct daily historical weather over Japan by cloud cover data assimilation using a Gaussian transformation method. This study conducts an experiment in the winter of 1995 using visually observed weather conditions recorded by Japan Meteorological Agency (JMA). Gaussian transformation shows great potential to improve cloud cover assimilation in actual reconstructions, especially for humidity fields in the lower troposphere and temperature fields in the middle and upper troposphere. This study demonstrates the possibility of reconstructing high-temporal-resolution atmospheric conditions before the 20th century with diaries data.