利用指数函数、幂函数等4种模型对太湖2006--2007年50个站点250个样本的有色可溶性有机物(CDOM)光谱吸收系数进行拟合,发现其他3种模型都比传统的取自然对数进行线性回归的模型拟合效果好。统计检验显示,幂函数模型的光谱拟合效果最好,平均决定系数为0.9990,大于其他3种模型,而平均相对误差、均方根误差则小于其他3种模型。利用幂函数模型计算得到,在280~500nm光谱范围内,CDOM吸收光谱斜率均值为(6.66±0.40)nm^-1,变异系数6.08%,其频率基本上遵循正态分布。对于太湖CDOM吸收光谱斜率值,在空间分布上,梅梁湾显著小于大太湖开阔水域(P〈0.001);时间分布上,最大值出现在春季,最小值在秋季,夏季值小于冬季值。CDOM吸收光谱斜率与吸收系数呈显著负相关,相关性随波长增加而增强。
Spectral absorption of chromophoric dissolved organic matters (CDOM) in 250 samples collected from 50 sampling stations in Lake Taihu fi'om 2006 to 2007 were simulated using four different models in this present study. Four models included three exponential models using linear and nonlinear fitting, and a hyperbolic model. It was found that the other three models ( Model 2, 3 and 4) performed better than the conventional single exponential model ( Model 1 ). Statistics show that the power function model fitted the best, being 0. 999 0 in average determination coefficient, higher than the other three; but was lower in average relative error and root mean square error (RMSE) than the other three (ANOVA, P 〈 0. 001 ). Calculation, using the power function model (Model 4) , demonstrated within the spectra range 280-500 nm, the CDOM mean spectral slope(S) value was (6.66 ±0.40) nm^-1 with a variation coefficient being 6.08%. The fre- quency distribution of S value basically followed a normal distribution. Spatially, the S value was significantly higher in Meiliang Bay than in the open water area of Lake Taihu ( ANOVA,P 〈 0.001 ). Temporally, S value followed the order of spring 〉 winter 〉 summer 〉 autumn. The CDOM S value demonstrated a significant negative correlation with absorption coefficient, but a positive one with wavelength.