资源环境科技发展态势分析平台

Radiosonde data are valuable to the research community because they provide observations for multiple decades and can be used to validate model output. However radiosonde data often suffer from quality issues, which has undermined their credibility. Therefore, corrections for biases and changepoints are needed to remedy the situation. Homogenization of monthly radiosonde specific humidity (q) from the 1970s to the present has been performed on selected stations over the Southwest Pacific (SWP) at three pressure levels (i.e. 850, 700 and 500 hPa). A three-step procedure involving (1) adjustments for two sampling biases, (2) detecting secular changepoints (i.e. discontinuities) using both statistical techniques and metadata validation, and (3) an innovative break size estimate approach, has been implemented to achieve this aim. In the last step, a discontinuity-free pseudo-q is constructed from saturated specific humidity q(s) which itself is derived from an already homogenized temperature (T) time series. This pseudo-q serves as a reference that not only distinguishes artificial from natural changepoints but also helps to estimate the magnitudes of the discontinuity. On the decadal time scale, the adjusted q (q(adj)) exhibits spatially more consistent moistening at the 850 hPa level over most of the region and a contrast at 500 hPa between the moistening in the tropics and a drying in the subtropical South Pacific. Mean regional trend estimates are approximate to 1.8% (850 hPa), -0.2% (700 hPa) and 1.3% (500 hPa) per decade. A climatological comparison with the three latest reanalysis products-CFSR, ERA-Interim and MERRA, suggests the reanalyses have significant negative biases over Southeast Asia (SEA) at all three levels. Over Australia the biases are negative at 850 hPa while positive at 500 hPa. The reanalysis products tend to be more similar among themselves in the estimates of q, as compared to the radiosonde measurements of q and q(adj). The homogenized radiosonde q, when assimilated into reanalysis, is likely to lead to a more realistic model of the hydrological cycle.