The Study Of Applying Hyper-spectral Remote Sensing Technology In Soil Moisture Monitoring

The development of precision agriculture needs accurate ground information. Remote sensing technologies, especially hyper-spectral remote sensing technology, which has very high and continuous spectrum, can be used to monitor the crop growth in large scale, in time and nondestructive, then to provide useful technology support to precision agriculture. Soil is a kind of basic natural resource for agriculture development, as an important component, soil moisture can affect on the growth and output of crops directly. The traditional measurement of soil moisture is inefficient, time-consuming and limited in scale, which makes it can not get soil moisture space distribution effectively to satisfy the development of precision agriculture. Optical remote sensing technologies (represented by TM, SPOT, etc.) are still not good for soil moisture monitoring in regional level or parcel level, because of its low spectral resolution. While with high spectral resolution, hyper-spectral remote sensing can be used to examine the slight variation of soil moisture promptly and accurately, and then contribute to precision agriculture.This research takes the inferior kind of black soil in Jilin Province as subject, using the hyper-spectral data obtained by field and laboratory experiment, applying a series of methods, to analyze the relationship between soil surface moisture and soil spectral reflectance, to inverse the real soil moisture, and to find the common water-sensitive wavelength of both field and laboratory experiment. The conclusion as follow:(1) After several transformations with laboratory spectral data, the sensitivity of spectral data has been approved after inversion calculation, especially after spectral derivative transformation and the first derivative of logarithm of reflectance.(2) Under the low field water holding capacity condition, the sensitive bands of the undisturbed soil spectrum focus on 1440-1460nm, 1800-1970nm and 2100-2160nm. While after the laboratory experiment, sensitive bands turned to 450nm, 1900nm and 2100nm. The correlation coefficient between laboratory spectral data and soil moisture can reach 0.89 at 2156nm, makes it the best one among the relevant results. And we also found that the laboratory soil spectrum is not merely obviously improved with the relevant relations between the water contents of soil, and the relevant intensity between one's own wave band obviously increase.(3) Normalized transformation can be good at eliminating the difference caused by different illumination terms. Mann-Whitney U test shows that the range of 651-820nm, 1490-1500nm and 2134-2300nm possess the characteristics of the same distribution, and the bands of 2100-2150nm is the common sensitive zone to soil moisture between the field and laboratory spectral data.