Publication Date:
2018
Description:
〈p〉Publication date: December 2018〈/p〉
〈p〉〈b〉Source:〈/b〉 International Journal of Applied Earth Observation and Geoinformation, Volume 73〈/p〉
〈p〉Author(s): Rukeya Sawut, Nijat Kasim, Abdugheni Abliz, Li Hu, Ahunaji Yalkun, Balati Maihemuti, Shi Qingdong〈/p〉
〈h5〉Abstract〈/h5〉
〈div〉〈p〉Spectroscopy is regarded as a quick and nondestructive method to classify and quantitatively analyze many elements of the soil. Visible and Near-infrared reflectance spectroscopy offers a conductive tool for investigating soil heavy metal pollution. The main goal of this work is to obtain spectral optimized indices (RSI, NPDI and NDSI) related to soil heavy metal Arsenic (As), to estimate the As contents in soil based on geographically weighted regression model (GWR), and to investigate the plausibility of using these spectral optimized indices to map the distribution of heavy metal Arsenic in the soil of coal mining areas. The spectral optimized indices (RSI, NPDI and NDSI) derived from the original and transformed reflectance (the reciprocal (1/R), logarithm (lg〈sup〉R〈/sup〉), logarithm-reciprocal (1/lg〈sup〉R〈/sup〉) and root mean square method (〈math xmlns:mml="http://www.w3.org/1998/Math/MathML" altimg="si2.gif" overflow="scroll"〉〈msqrt〉〈mi〉R〈/mi〉〈mi〉 〈/mi〉〈/msqrt〉〈/math〉) were used to construct the GWR models. Then, the variables (RSIs, NPDIs and NDIs) were applied in estimating the Arsenic (As) concentration and in the mapping of the As distribution in this study region. The NPDIs calculated by the original and transformed reflectance (〈em〉R〈/em〉, 1/〈em〉R〈/em〉, lg〈sup〉R〈/sup〉, 1/lg〈sup〉R〈/sup〉, and 〈math xmlns:mml="http://www.w3.org/1998/Math/MathML" altimg="si2.gif" overflow="scroll"〉〈msqrt〉〈mi〉R〈/mi〉〈mi〉 〈/mi〉〈/msqrt〉〈/math〉) indicated higher correlation coefficient values than NDSI and RSI. The highest correlation coefficient and lowest 〈em〉p〈/em〉-values (〈em〉r〈/em〉≥0.73 and 〈em〉p〈/em〉=0.001) were found in thenear-infrared (NIR, 780–1100 nm) and shortwave infrared (SWIR, 1100–1935 nm). From the 4 prediction models (GWR) performances, it can be seen that Model-a (〈em〉R〈/em〉) showed superior performance to the other three models (Model-b (1/〈em〉R〈/em〉), Model-c (〈math xmlns:mml="http://www.w3.org/1998/Math/MathML" altimg="si2.gif" overflow="scroll"〉〈msqrt〉〈mi〉R〈/mi〉〈mi〉 〈/mi〉〈/msqrt〉〈/math〉) and Model-d (lg〈em〉〈sup〉R〈/sup〉〈/em〉)), and it has the highest validation coefficients (〈em〉R〈/em〉〈sup〉2〈/sup〉 = 0.831, RMSE =4.912 μg/g, RPD=2.321) and lowest AIC (Akaike Information Criterion) value (AIC=179.96). NPDI〈sub〉1417 nm, 1246 nm〈/sub〉 is more sensitive and potential hyperspectral index for As in the study area. Thus, the two band optimized index (NPDI〈sub〉1417 nm, 1246 nm〈/sub〉) might be recommended as an indicator for estimating soil As content. The hyperspectral optimized indices may help to quickly and accurately evaluate Arsenic contents in soil, and furthermore, the results provide theoretical and data support to access the distribution of heavy metal pollution in surface soil, promoting fast and efficient investigation of mining environment pollution and sustainable development of ecology.〈/p〉〈/div〉
Print ISSN:
0303-2434
Topics:
Geography
,
Geosciences
Permalink