基于多源数据的临近空间下边界大气变化特征对比OA

Based on ERA5 and CRA atmospheric reanalysis datasets for 2009,together with data from the NR-LMSISE-00 empirical atmospheric model,this study employs empirical orthogonal function(EOF)analysis to investigate the spatiotemporal characteristics of atmospheric temperature(T20)and density(D20)at the lower boundary of near-space.For T20,both the reanalysis datasets and the empirical model exhibit an annual mean dis-tribution characterized by relatively low temperatures near the equator and higher temperatures at mid-latitudes,which is associated with the dynamical transport of the Brewer-Dobson circulation.However,substantial differ-ences are observed at high latitudes.The reanalysis datasets show relatively high temperatures over the southern mid-to high latitudes and the lowest temperatures over the Arctic,while NRLMSISE-00 produces a relatively warm Arctic and cold Antarctic.This discrepancy may be attributable to the model's limited ability to represent high-latitude temperature variability during periods of low solar activity.The EOF1 spatial pattern derived from CRA data is characterized by higher values in the south and lower values in the north,while NRLMSISE-00 ex-hibits the opposite pattern.The PC1 time series from both datasets show pronounced seasonal variability;however,the empirical model fails to capture short-term disturbances,such as the sudden stratospheric warming event and the easterly phase of the quasi-biennial oscillation in 2009.This limitation is likely related to the model's insuffi-cient sensitivity to anomalous dynamical processes in the lower stratosphere.The EOF2 patterns from both datasets consistently exhibit relatively low values at the equator and higher values at mid-latitudes,while their PC2 time series display a bimodal structure.However,NRLMSISE-00 does not reproduce the low-value center over the Arctic evident in the CRA data,and its fluctuation amplitude is substantially weaker.For D20,both datasets show annual mean distributions characterized by higher densities near the equator and lower densities over Antarctica,broadly opposite to the corresponding temperature distributions.However,substantial differences occur over the Arctic:NRLMSISE-00 produces relatively low densities,in contrast to the higher values in the reanalysis datasets.This discrepancy may be related to limitations in the high-latitude representation of the MSIS framework and differences in the underlying mechanisms used to construct atmospheric density.The EOF1 pattern derived from CRA is characterized by lower values in the south and higher values in the north,whereas NRLMSISE-00 exhib-its high values near the equator and lower values toward both poles.The corresponding PC1 tine series shows pro-nounced seasonal variability in CRA data but predominantly periodic fluctuations in the model results.For EOF2,the CRA pattern is broadly consistent with the annual mean distribution,while NRLMSISE-00 exhibits a hemi-spherically symmetric pattern with an approximately reversed spatial structure.This discrepancy may be associated with the use of hemispheric-specific datasets and parameterization schemes in the empirical model.The PC2 time series derived from CRA exhibits a bimodal seasonal structure,whereas that from NRLMSISE-00 displays a uni-modal structure,with its peak occurring earlier than that in the CRA data.This temporal discrepancy may reflect the model's limited responsiveness to atmospheric dynamical variability and its reduced ability to capture transient processes.Overall,the results reveal substantial differences between meteorological reanalysis datasets and the NRLMSISE-00 empirical model in representing the principal atmospheric parameters at the lower boundary of near-space.These findings improve our understanding of the complex spatiotemporal variability of temperature and density in this region provide useful guidance for improving the representation of key physical processes at high latitudes and the model's response to short-term atmospheric disturbances.