Monitoring The Soil Nitrogen And Soil Moisture In Winter Wheat Fied Based On The Hyperspectral Technology

The research study on soil hyperaspectral and soil nitrogen, moisture content of the correlation relevant, and according to the R2extract the characteristic bands, and established the best soil nitrogen, moisture content monitoring model. In order to apply hyperaspectral monitor crop growth provide some guidance.These test data were collected including soil hyperaspectral, soil nitrogen, and moisture content at different growth stage at Shanxi Agricultural University in2011-2012, then excluded the unusual spectrum, smoothed and made spectral transformation. Through correlation analysis and stepwise regression to establish the best monitoring model of nitrogen and water. And the same time test the model of validation and accuracy. The conclusions as follow:1) After11kinds of transformations, the information of spectral becomes more rich and correlation greatly improved, and to extract the characteristic bands easily to established the monitoring models. The results showed that the soil nitrogen content with (1/R)" had the best correlation and moisture content with R" had the best correlation.The reflectivity and negatively correlated with nitrogen content, with the increase of nitrogen content, reflectivity begin to reduce. Different stages nitrogen content of soil has different spectral curve but tends to be parallel. The curve between350nm to800nm was growing fast, then became flattens. Soil spectral reflectance was not the same in different growth period, at filling stage was the lowest, heading stage was highest, then mature and jointing period, Among the Log(1/R)’was the best spectral parameters to prediction soil nitrogen content all growth period. R’was the best spectral parameters to prediction the nitrogen content of heading and mature period, Log(R)’was the best spectral parameters to prediction the nitrogen content of filling stage.3) Reflectivity and water relationship was relatively complex, in this paper, we studied for the low water content of soil. The result showed that with the increase of water content, the reflectivity begun to decrease, and the correlation of water content with reflectivity became remarkable. Reflectance after transformation and the correlation between water content increase. Through to the determination of soil water content and spectral reflectance in wheat, and carried on the correlation analysis and stepwise regression, the monitoring model was established. Best monitoring model of moisture in the whole stages for reflectivity logarithm as parameters monitoring model was established. Various stages of the best models were respectively from bottom reflectivity of first order differential, second order differential, first order differential and logarithm of second order differential for the parameter monitoring model was established. Verified model had good stability and high prediction accuracy.4) Severely affected by the moisture of the spectrum, when the soil containing water, the spectral characteristics of the soil reaction is mainly the basic characteristics of the water, in the case of containing water was therefore difficult to monitor other substances in the soil. Water has a certain influence on the spectral reflectance of the soil, showed the wet earth albedo was smaller than the reflectivity of dry soil, and wave also does not have the dry soil. Water for using spectral parameter to establish models for predicting the whole growth period and each growth period of nitrogen was larger, the influence of monitoring in the whole stages model estimation precision was reduced by14.74%, the influence of the prediction precision of the model for each period is different, had the greatest influence to the heading stage, accuracy was reduced by19.68%.