On the basis of the EOF analysis of global geopotential height anomaly (GHA) field at 10 hPa level, the arctic oscillation (AO) and the means antarctic oscillation (AAO) can be detected more obviously at the upper level of atmosphere than the AO or the AAO in surface layer. Unlike the hemisphere pattern of the AO and the AAO in the surface lager given by previous authors, the AO or the AAO in the stratosphere has its global features. The zonal oscillations-the Southern Oscillation (SO) and the north oscillation (NO) in atmospheric surface layer become less clear in the upper air. The first mode (AO mode, abbreviated to AOM hereafter) and the second mode (AAO mode, abbreviated to AAOM hereafter) respectively have 41.47% and 27.04% of the total variance contribution. The cumulative variance contribution of the first two modes reaches 68.51%. These two modes are the main components for the interdecadal or decadal oscillation in the stratosphere. In addition, there still exist two kinds of oscillation patterns with less probability,namely, the symmetric pattern at mid-high latitudes in the Southern Hemisphere and the asymmetric pattern. Spectral analysis shows that the AOM and the AAOM all have a spectral peak for 22 a period, being consistent with the periodic variations of the solar magnetic field, and a peak for 11 a period, being consistent with the period of the numbers of sunspots. Step filter analysis shows that the influencing factor for the upper atmospheric oscillation is the solar activity. The fluctuation of the solar magnetic field is the more influencing factor than the variation of the sunspot number.

Obvious tendency and periodicity of the air temperature can be detected over the North Pole area. They are reflected as follows:a. the air temperature at the earth surface and in the middle layer of the stratosphere tends to be increased either in winter or in summer. The air temperature has increased 1.3℃ for about 50 years at a speed about 0.025℃/year in January, and 0.013℃/year in July. The air temperature in the middle layer of the stratosphere (10 hPa) in January has increased 10℃. The temperature rising speed in July is 0.14℃/year. Generally speaking, the temperature rising speed is quicker in winter than in summer and quicker in the upper layer than at the earth surface. b. The air temperature at the top layer of the troposphere (100 hPa) over the North Pole area tends to be increased either in winter or in summer. The air temperature in January has decreased 5.0℃ for about 50 years at a temperature decreasing speed about 0.094℃/year, and at a temperature decreasing speed about 0.032℃/year in July. The speed of the temperature decreasing is greater in winter than in summer. c. Periodicity. The air temperature respectively at different altitudes over the North Pole possesses interdecadal variation with a period of 22 years. In July the amplitude of the variation with a period of 22 years decreases rapidly from the high altitude to the low. This means that the 22-year's period is more obvious at the high altitude than at the low altitude. At the earth surface layer in North Pole there also is obvious decadal variation with a period of 11 years. The analysis indicates that the 22-years' period temperature variation is associated with the periodic variation of the solar magnetic field. The 11-year period temperature variation is corresponding to 11 year' period of the variation of the sunspot number.

Temperature and relative humidity profiles derived from two China-made global positioning system (GPS) radiosondes (GPS-TK and CF-06-A) during the east tropical Indian Ocean (ETIO) experiment were compared with Vaisala RS92-SGP to assess the performances of China-made radiosondes over the tropical ocean. The results show that there have relative large biases in temperature observations between the GPSTK and the RS92-SGP in the low troposphere, with a warmbias of greater than 2 K in the day and a cooling bias of 0.6 K at night. The temperature differences of the CF-06-A were small in the troposphere both in daytime and nighttime, and became large peak-to-peak fluctuations in the stratosphere. The intercomparison of the relative humidity showed that the CF-06-A had large random errors due to the limitation of sensors and the lack of correction scheme, and the GPS-TK had large systematic biases in the low troposphere which might be related to the temperature impact. GPS height measurements are clearly suitable for China-made radiosonde systems operation. At night, the CF-06-A and the GPS-TK could provide virtual potential temperature and atmospheric boundary layer height measurements of suitable quality for both weather and climate research. As a result of the intercomparison experiment, major errors in the Chinamade radiosonde systems were well indentified and subsequently rectified to ensure improving accuracy for historical and future radiosondes.