The Analysis Of The Differences Between Reanalysis Data And Cooperative Observation Data

For its unique advantages, reanalysis data make up the shortage of observed data. But reanalysis data are due to data assimilation of the simulated results and observed data. There might be three aspects of errors in reanalysis data:numerical model, the design of assimilation and observation methods. So it is very necessary to evaluate the quality and adaptability of reanalysis data, In this paper, by using the observed data and NCEP reanalysis data in the same period of12Cooperative observation stations of different geographical areas and different underlying surface characteristics in Arid and semi-arid area, the differences between these two kinds of data were systematically discussed from several angles:short-term spatial and temporal variation characteristics, long-term temporal features, surface energy characteristics and combined with WRF model simulation results. The main results are as follows:1. Short-term analysis:The daily maximum of sensible heat flux NCEP reanalysis data was greater than that of cooperative observation data. The variation trend of daily maximum of NCEP data and cooperative observation data were relatively consistent in all the stations except stations in the central part of northwest region. Considering statistical correlation and error absolute value distribution, sensible heat flux NCEP reanalysis data has the best quality in the eastern part of northwest region. The absolute values of the errors of these two kinds of data were mainly concentrated in the range of0-50W/m2. In the eastern part of northwest region the quality of sensible heat flux reanalysis data were the best.The daily maximum of latent heat flux NCEP reanalysis data were greater than that of cooperative observation data in all stations except the stations in eastern part of northwest region. The absolute values of the errors of these two kinds of data were mainly concentrated in the range of0-50W/m2.In the northeast and north China, the quality of latent heat flux reanalysis were better.2. Long-term analysis:The sensible heat flux and the latent heat flux reanalysis data were bigger than their observed data. The sensible and latent heat flux of NCEP data and observed data were relatively consistent in year, season, daily, ten-day average, three-day average and month average variation in SACOL station. Sensible and latent heat flux NCEP data can well reflect the variation characteristics of surface flux in SACOL station. With time scales ranging from6-hour average to month average, the error of NCEP reanalysis data and the observed data decreases, the credibility of NCEP reanalysis data gradually improved.6-hour average contrast:The trends of Sensible heat and latent heat flux reanalysis data and observed data were relatively consistent in each time level. In the afternoon and evening, these two kinds of data were all positive, and reanalysis data were greater than the observed data. While in the morning and night they were negative, and the absolute value of reanalysis data were greater than that of the observed data. These two kinds of data of sensible heat flux had maximum discrete degree in the evening and the minimum at night. These two kinds of data of latent heat flux had maximum discrete degree in the evening and the minimum in the morning.3. The analysis of surface energy characteristic:the change range of sensible heat reanalysis data were larger than that of observed data. Ranging from-150to450W/m~2.The range of observed data was smaller. In the western farmland and alpine meadow underlying surface, the observed data were mainly concentrated in the range of-80-120W/m2. In the Desert grassland underlying surface area, the range is larger, it is from-160to250W/m2. The observed data showed the Microclimate characteristics. It depends on the actual underlying surface characteristics of the station. The reanalysis data depends on the geographical location of station.The distributions of sensible heat and latent heat flux reanalysis data and observed data had opposite characteristics. The changing range of net radiation flux reanalysis data range is greater than that of the measured data.Sensible heat flux fields:The monthly average of sensible heat flux field NCEP data and observed data were consistent. They both have a strong sensible heat center in the northwest, the value were120w/m2and65w/m2, respectively.Latent heat flux fields:The latent heat flux fields of NCEP data and observed data were obviously different. The observed data have a positive center in the102°E, The NCEP data have no center. This is because in the central part of northwest, the local microclimate characteristics make the latent heat flux observed data were greater than NCEP data, which shows in the flux field was a positive latent heat center.4. In the eastern part of northwest region the sensible heat flux observed data were greater than or the same as the simulated value, in the rest stations the simulated values were greater than the observed data; The simulation of latent heat flux was opposite. The simulation effect in the peak area was worse than in other time.