中文摘要：

选取新疆奇台县的134个土壤样本,利用土壤反射率对数的一阶导数光谱分别对4种小波函数进行多层离散分解,采用PLSR方法分别建立了土壤速效钾含量的反演模型,并对其精度值进行检验。结果表明：小波分解获得的各层低频系数以1~3层较高,而其余各层则较低。所有函数分解的6层中,均以第2层低频系数建模的精度最高,随着分解层数（〉2层）的增加,其精度值和显著性明显降低。相同尺度下,采用4种小波函数的低频系数构建的反演模型的精度差异较小,而Bior1.3为最优函数;基于Bior 1.3分解的ca2低频系数建模的R2达0.964,RMSE仅为8.19 mg·kg-1,且为极显著水平,为最佳反演模型,经样本检验后发现,此模型可用以快速、准确估算土壤高光谱速效钾含量。

英文摘要：

The available components of soil organic matter content is an important factor for spectral characteris- tics of soil, and it can provide important information for soil digital management and precise fertilization if avail- able components of soil can be estimated accurately using hyperspectral technology. Although the traditional chemical method had a high precision, there were mainly shortcomings, such as high cost, time-consuming, so it was unable to meet the needs of modem precision agriculture fertilization technology. In order to predict the avail- able kalium content of soil more quickly and accurately, and improve the precision and practicability of the soil available kalium estimation model by removing the noise of soil hyperspectral reflectance, this paper studied the inversion relationship between soil spectrum and soil available kalium content used wavelet analysis and based on hyperspectral technology. With 134 soil samples selected at Qitai County in Xinjiang, the first derivative spec- trum of the soil sample logarithmic reflectance was decomposed to many layers by using 4 wavelet functions re- spectively, and PLSR was used to establish the prediction models respectively and test precision values. Through comparison analysis, the optimal wavelet decomposing resolution for extracting the characteristic spectrum of soil organic matter was ascertained, and the best forecasting model was established. The results show that： 1-3 layers low-frequency coefficients of wavelet decomposition were better, while the rest were worse. In 6 layers of all function decomposition, the highest accuracy of inversion models construct by low-frequency coefficients were all ca2, with increased the decomposition layers, the precision and significance decreased significantly. In the same scale, there was little accuracy difference between inversion models constructed by 4 wavelet functions low-frequency coefficients, while Biorl.3 was optimal. The best inversion model was ca2 that built by Bior 1.3, with R~ and RMSE were