Data Intercomparison Among Reanalyses And Observations In Perspective Of Multi-Scale Variations Of Summer Climate In East China

International negotiations on climate change, the study of the responses to global warming in summer climate variation all need high quantities, high resolution, continuous and representational dataset. The two datasets, ERA40and NCEP/NCAR reanalyses datasets, which are used widely, performed inconsistent in climate variations in east China for the differences in model, data sources, quality control plan and assimilation scheme. Therefore, the work which used observations to verify the quantity of the reanalyses is of important theoretical and practical significance for the reanalysis datasets be used in climate diagnosis and model evaluation.Using surface air temperature, pressure, geopotential height and wind speed from radiosonde observations over China, as well as specific humidity at all the isobaric level in the troposphere calculated from the in-situ observed temperature and depression of the dew point, which are then extracted as elements of surface air temperature, sea level pressure (surface pressure), tropospheric temperature, geopotential height,wind speed and specific humidity over East China in summer (JJA), the comparison is performed of these observed quantities with those from the ERA40and NCEP/NCAR reanalyses on the long-term and regional averages, long-term trends, variance and periods. Three datasets of observations, the ERA40and NCEP/NCAR reanalyses are employed in this paper, hereafter named as the O, E and N respectively. Results are as follows.(1) Both ET and NT can reflect the spatial and temporal variations of China temperature, the former is more accurate.ERA40temperature (ET) corresponds closely to the observed temperature (OT) for the values and the anomalies, but NCEP/NCAR temperature (NT) is lower than the OT. ET and NT all can reflect annual variations of OT’s. The ET is more closely to the OT than NT for the long-term trends and interdecadal variations, especially in East China. The significant shift of China temperature around1976has been shown by the three datasets, while the ET is more consistent with the OT.(2) All the datasets show increasing pressure over East China. The ERA40sea level pressure (ES) has similar values with the observed pressure (OP), while the NCEP/NCAR surface air pressure (NP) and sea level air pressure (NS) are much higher than the counterparts of the ERA40and the observations, and much lower than the observations before the year of1965.Although showing significant uncertainty in the study of long-term trends of the OP, the two reanalysis datasets have ability to represent the inter-annual variability of the OP, with better performance in the eastern part (low latitudes) than in the western part (high latitudes) of China, while the representation of ES is superior to the NP and NS and marked by similar periodic variations with the OP.(3) Temperature of the two reanalyses can mimic the horizontal distribution and vertical variation characteristics of the OT in the troposphere, while the description of temperature change becomes better with height increasing. The ET (ET and NT) can reproduce the distribution characteristics of long-term trends of the OT in the lower (middle and upper) troposphere. Also, the ET and NT have ability to represent the temperature characteristics on the aspect of spatial distributions and inter-annual variations. For the reproduction of periodic feature, the ET shows better performance in the lower lever, while the two have comparable (different) performance in the middle (upper) level. Compared with the OT, the ET and NT are slightly lower in value, but with similar anomaly values.(4) EH and NH, height of the two reanalysis datasets, can represent the distribution characteristics of the observed height field (OH), which is marked by decrease of geopotential height from south to north in the troposphere. However, the values of the EH and NH are lower than the OH. Over East China, the two reanalysis datasets can better represent the long-term trends of the OH in the upper level than in the lower level, while the EH and NH have corresponding performance in the upper level and the EH are better in the lower level. The EH and NH can manifest the OH for its inter-annual variability which is marked by more significant variations in North China (upper troposphere) than in South China (lower troposphere), as well as its spatial distributions and periodic variations in the middle and lower troposphere, but with different performance for the temporal variations.(5) EV and NV, wind speed of the two reanalysis datasets, have ability to represent the distributions of observed wind speed, with better performance in the upper level than in the lower level, and the lower-level wind speed is represented more close to the observations in East China than in West China. Both the reanalysis datasets can represent the trend features of the observed wind speed that is decreasing after the turning point around the year1990, but overestimate the trends. The EV and NV have similar performance in representing the decadal variations and long-term trends of the observed wind speed (OV). Also, the EV and NV can reproduce the distributions of inter-annual variations of the OV, but overestimate the range of the inter-annual variations, with equal performance in the upper level and EV prior to NV in the lower level. Generally, wind speed of the two reanalysis datasets slightly underestimates the observations in value, but have anomalies in proportion to the observations.(6) EQ and NQ, specific humidity of the reanalysis datasets, are able to represent the distribution characteristics of observed specific humidity (OQ), and have equal performance of the observed distributions in the middle and lower troposphere. Both the EQ and NQ are weak (good) at representing the long-term trends (inter-annual distributions), in which the EQ has better performance than the NQ. Also, both have some capability to represent the periodic variability of the OQ in the middle and upper troposphere. The anomaly differences between EQ and OQ are smaller than those between NQ and OQ.The PSFC phenomenon can be observed in temperature, pressure/geopotential height, wind speed fields in mid-lower troposphere.