【摘要】 为了监测全球海洋上空的大气可降水量，已发射上天多种星载微波波谱仪和辐射计，相应地发展建立了多种反演算式。我们利用逐步回归分析，研究比较了反演晴天大气可降水量（Ｐｗ）的多种通道组合以及算式中不同的亮温（ＴＢ）函数形式的效果。主要结果有：（１）对应于高中低Ｐｗ值有不同的最佳通道组合；（２）常选的水汽吸收中心线２２．２３５ＧＨｚ不太适合于反演高中且变化范围大的Ｐｗ值；（３）有必要建立分气候区域和分季节的反演算式；（４）在反演算式中采用１ｎ（Ｔ０－ＴＢ）一般比线性形式有更好的回归和反演效果，但在有２２．２３５ＧＨｚ通道时则不然；（５）反演中低或变化范围大的Ｐｗ时，２２．２３５ＧＨｚ亮温的平方项能显著改善回归和反演结果。更多还原

【Abstract】 For remote sensing of atmospheric precipitable water (P W) over the oceans, a number of retrieval regression algorithms have been developed with different microwave channel combinations and different function forms of brightness temperature (T B). The objective of this work is to compare and examine the effects of different channel combinations and different function forms of T B with three simulated T B data sample sets, corresponding to high, medium, and low mean values of P W. The successive regression analysis indicates that: (1) the optimum dual or triple channel combination is different for different variation range of P W; (2) the channel 22.235 GHz is not suitable for measuring high total water vapor amount with large variation range, so the triple channel combination (19, 22, 37 GHz) used by SSM/I is not very ideal for the global retrieval of P W; (3) it is necessary to establish the climate zonally or seasonally divided retrieval algorithm because it has smaller regression residual error; (4) the function form ln(280-T B) generally gives better regression and retrieval results than the linear form of T B, but it is not true for 22.235 GHz; and (5) the use of the term T 2 B 22 can improve the regression as well as retrieval results in the case of low and medium values of P W.更多还原