| Soil is the very important part of the Earth surface. It is closely related to human health and living environment. The evaluation of the heavy metal pollution of soil is traditionally based on the field data collection and chemical analysis for the samples. There are a few researches on the basis of color index and digital number value of remote sensing images. This study based on the principle of color index and digital number value (DN value) of remote sensing images on spectral reflectance. Using the Landsat 8 images on May 3rd of 2014 as the remote sensing and measurement of the collected soil samples near the steel factory in the same time, this paper analyzed the relationships between color index and heavy metal content and DN value as well. Subsequently, a feasible and rapid evaluation method for evaluation of heavy metal contamination of the soils, which uses remote sensing DN value and was corrobolated by the color index of soils. The main conclusion are as followings.1. The content of the Cr, Cu, Mn, Pb and Zn are various on the sampled section, which is depending on its soil type. In the horizontal direction the contents of the heveay metals are varied with the distance to the pollution source and the prevailing wind direction. According to the evaluation using the Nemerow comprehensive contamination index, in the downwind area the sampled sites,3～9km far from the steel plant is attributed to the moderate heavy metal pollution, and others,9～12km far from the steel plant are light to moderate pollution. In upwind the sampled sites,0～3km far from the steel plant are uncontaminated to light pollution. The analysis of correlation of the elements shows that Fe has a good correlation with Cu and Zn, with the coefficient between 0.5 to 0.9, which indicates that they originated from the same pollution source. Moreover, the element of Cr probably originated from the different pollution source.2. The variation of the color index and heavy metal content in the vertical direction on the section are opposite to the horizontal direction. The coefficient between a* is significantly negative correlated with Cr (the correlation coefficient is=-0.665), and has negative correlations with Zn, Pb and Mn (the correlation coefficient are between 0.335～0.537),however a* has positive correlation with Fe (the correlation coefficient is 0.107),it may be related to the reson that Fe is mainly existing in the form of magnetite and have little hematite; a* have no correlation with Cu(the the correlation coefficient is 0.013); Meawhile, L* has significantly negtive correlateds with Fe, Cu (the correlation coefficient are 0.887 and 0.925);and with Mn is significant negtive correlated (the correlation coefficient is R2=-0.665), and has low correlation with Pb and Zn (the correlation coefficient are 0.21 and 0.30), and have no correlation with Cr(the correlation coefficient is 0.045); b* also has a significantly negative correlation with Mn (the correlation coefficient is 0.828), and with Zn, Fe and Cu are significant correlated (the correlation coefficient are between 0.629～0.665), b* have low negative correlation with Cr, Pb (the correlation coefficient are 0.493 and 0.202).3. DN value has a good relationship with color index, a* has a high correlation with visible band (the correlation coefficient are between 0.399～0.431), and has low correlation with other bands; b* has the highest correlation with near infrared band (the correlation coefficient is 0.385); L* has significantly negative correlation with short-wave band (B7) and.the correlation coefficient is 0.71, L* has significantly negative correlation with short-wave band (B6) and.the correlation coefficient is 0.650.4. The visible band has a high correlation with Cr (the correlation coefficient is between 0.275-0.302) and almost has no correlation with other bands; The Near-infrared band have a highest correlation with Fe (the correlation coefficient is 0.537); The short-wave band (B6,B7) have significantly correlation with Fe, Cu (the correlation coefficient are on the above of 0.6).5. The results based on model of single-band DN value have 53.8% consistent with the result of Newerow single factor evaluation, the result of 23.1% are different within one grade; the result of 15.4% are different within two grade; The results based on model of multi-band DNvalue have 38.5% consistent with the result of Newerow single factor evaluation, the result of 46.2% are different within one grade; the result of 15.4% are different within two grade; Evaluation model based on a single band have higher credibility than the evaluation model based on multi-band, which may be due to the information redundancy between some bands. So, it is effective to use the DN value of remote sensing images to estimate heavy metal pollution, and it is a method that esitimate soil heavy metal pollution simplely, fastly and real-time. |
PDF Full Text Request