Abstract：In order to rapid determine soil properties through hyperspectral analysis, 154 soil samples were collected in farmland in Xinzheng of Henan province. The raw hyperspectral reflectances of samples were measured by the standard procedure with a spectrometer of ASD Field Spec3 equipped with a high intensity contact probe under the laboratory conditions. The raw spectral reflectances of 400 ~ 2 400 nm were smoothed by Savitzky-Golay and then were transformed to one spectral indices, i. e., first order differential reflectance ( FD) . The correlation analysis was done between first order differential reflectance and soil organic matter, available N, available P, available K, PH, Fe, Cr, Cd, Zn, Cu and Pb by Pearson correlation analysis ( P = 0.01) . The best spectral characteristic shared bands were selected by the fuzzy clustering maximum tree method.Based on the panel data model, hyperspectral composite inversion model of farmland soil properties was built in Xinzheng. The results show that: the panel date model is significant and the goodness of fit is high ( R2 v=0. 9776, DW = 2. 03, F = 110. 97) . The precision test indicates that model has general ability of quantificational prediction of AN, AK and Cd with RPD valuing ranges 1. 4 ~ 1. 8. The model has quantitative prediction ability of Cr with RPD values ranging 1. 8 ~ 2. 0 as well as good ability of quantificational prediction of PH, SOM, Fe and Zn with RPD values ranging 2 ~ 2. 5. AP, Cu and Pb can be performed and predicted in the model with excellent ability quantificational prediction, which RPD values is greater than 2. 5. Therefore, the panel date model can be hybrid inversion various heavy metals and the calculation is simple and fast. It can be used to rapidly monitor soil heavy metals of hyperspectral reflectance in Xinzheng.

Keyword：hyperspectral;the maximum tree method for fuzzy clustering;panel data model;soil properties;hybrid inversion;xinzheng city;