[1]TIAN Anhong,FU Chengbiao,XIONG Heigang,et al.Hyperspectral Estimation of Na+ Ion in Saline Soils in Areas With Different Human Activities Using BPNN Model[J].Research of Soil and Water Conservation,2020,27(02):364-369.
Copy
Hyperspectral Estimation of Na+ Ion in Saline Soils in Areas With Different Human Activities Using BPNN Model
Research of Soil and Water Conservation[ISSN 1005-3409/CN 61-1272/P] Volume:
27
Number of periods:
2020 02
Page number:
364-369
Column:
Public date:
2020-03-30
- Title:
- Hyperspectral Estimation of Na+ Ion in Saline Soils in Areas With Different Human Activities Using BPNN Model
- CLC:
- S151.9
- DOI:
- -
- Abstract:
- Cation Na+ of soil salinity plays an important role in the formation of saline soil. The soils in no human activity(Zone A)and human activity(Zone B)of Xinjiang were studied. Field hyperspectral data and 0—20 cm soil samples were collected, and Na+ contents were tested. BP neural network(BPNN), partial least squares(PLSR)and stepwise multiple regression(SMLR)models were used to compare and analyze the hyperspectral estimation of Na+, and to explain the mechanism of Na+ estimation accuracy in areas with different human activities. The results show that the Na+ has the highest proportion and the largest mean value among the four cations(Ca2+, Mg2+, K+, Na+), the proportion of Na+ in Zone A and B is 48.429% and 62.274%, respectively, and the mean value is 1.590 and 2.148, respectively. First-order and second-order differential processing were been performed on the Original(R)and reciprocal(1/R)spectral transforms, and the band whose correlation coefficient was checked by 0.05 was extracted as the characteristic band. There are 24 models in Zone A and B for the three modeling methods, and the BPNN models with 1/R at the second-order differential are the best prediction models for Zone A and B, which meet the accuracy requirements when they are iterated by 19 times and 9 times, respectively. The relative analysis error RPD, the decision coefficient R2, and the root mean square error RMSE are 2.461 6, 0.860 9, and 0.350 1 in Zone A, and 2.169 8, 0.800 6, and 0.803 5 in Zone B, respectively. The prediction ability of BPNN for Na+ ions is very good, the prediction ability of PLSR is general, and the prediction ability of SMLR is very poor. These research results can provide a scientific basis for improving salinized soil in arid regions.
- References:
-
[1] 阿布都热合曼·哈力克.新疆岳普湖县土壤盐分空间变异及其分布特征研究[J].水土保持研究,2011,18(1):97-100.
[2] 王飞,杨胜天,丁建丽,等.环境敏感变量优选及机器学习算法预测绿洲土壤盐分[J].农业工程学报,2018,34(22):102-110.
[3] Wichelns D, Qadir M. Achieving sustainable irrigation requires effective management of salts, soil salinity, and shallow groundwater[J]. Agricultural Water Management, 2015,157:31-38.
[4] 林鹏达,佟志军,张继权,等.基于CWT的黑土有机质含量野外高光谱反演模型[J].水土保持研究,2018,25(2):46-52,57.
[5] 王明宽,莫宏伟,陈红艳.基于近地高光谱的土壤氯离子含量估测[J].水土保持通报,2017,37(6):214-219.
[6] 郑雯,明金,杨孟克,等.基于波段深度分析和BP神经网络的水稻色素含量高光谱估算[J].中国生态农业学报,2017,25(8):1224-1235.
[7] 龚巧灵,官冬杰.基于BP神经网络的三峡库区重庆段水资源安全评价[J].水土保持研究,2017,24(6):292-299.
[8] 刁万英,刘刚,胡克林.基于高光谱特征与人工神经网络模型对土壤含水量估算[J].光谱学与光谱分析,2017,37(3):841-846.
[9] 卢志宏,刘辛瑶,常书娟,等.基于BP神经网络的草原矿区表层土壤N/P高光谱反演模型[J].草业科学,2018,35(9):2127-2136.
[10] 郭云开,刘宁,刘磊,等.土壤Cu含量高光谱反演的BP神经网络模型[J].测绘科学,2018,43(1):135-139,152.
[11] 段鹏程,熊黑钢,李荣荣,等.不同干扰程度的盐渍土与其光谱反射特征定量分析[J].光谱学与光谱分析,2017,37(2):571-576.
[12] 乔娟峰,熊黑钢,王小平,等.基于最优模型的荒地土壤有机质含量空间反演[J].江苏农业学报,2018,34(1):68-75.
[13] Kuang B, Mouazen A M. Effect of spiking strategy and ratio on calibration of on-line visible and near infrared soil sensor for measurement in European farms[J]. Soil & Tillage Research, 2013,128:125-136.
- Similar References:
Memo
-
Last Update:
2020-02-25