Inversion of soil salinity in China's Yellow River Delta using unmanned aerial vehicle multispectral technique.

Autor: Zhang Z; Institute of Land Reclamation and Ecological Restoration, China University of Mining and Technology (Beijing), Beijing, 100083, China., Niu B; College of Resources and Environment, Shandong Agricultural University, Tai'an 271018, China., Li X; College of Resources and Environment, Shandong Agricultural University, Tai'an 271018, China., Kang X; College of Geographical Sciences and Surveying Engineering, China University of Mining and Technology (Beijing), Beijing, 100083, China., Wan H; Center of Xintai Mineral Industry Development, Tai'an 271200, China., Shi X; Wells Fargo, San Francisco, CA, 94105, USA., Li Q; Mesofilter Inc, San Jose, 95131, USA., Xue Y; College of Information Science and Engineering, Shandong Agricultural University, Tai'an 271018, China., Hu X; College of Information Science and Engineering, Shandong Agricultural University, Tai'an 271018, China. huxiaozhy@163.com.
Jazyk: angličtina
Zdroj: Environmental monitoring and assessment [Environ Monit Assess] 2022 Dec 29; Vol. 195 (1), pp. 245. Date of Electronic Publication: 2022 Dec 29.
DOI: 10.1007/s10661-022-10831-0
Abstrakt: The rapid and accurate acquisition of soil property information, especially the soil salinity (SS), is required for saline soil management in the Yellow River Delta (YRD). In this study, Lijin County and Kenli District were selected as study area. Unmanned aerial vehicle (UAV) multispectral data and soil sample data were acquired from March 25 to 28, 2019. Pearson correlation and gray correlation analyses were first used to screen sensitive spectral bands/indices, which were used for model parameters construction. Three machine learning and one statistical analysis methods were used to construct the SS inversion models. The results found that the gray correlation coefficient value were greater than the Pearson coefficient value for all bands and indices. Based on the gray correlation coefficient, nine sensitive bands and indices were selected to construct 18 model parameters. By comparing the 4 models, it was concluded that the BPNN model had the highest inversion accuracy, and its calibration coefficient of determination (R 2 ) and root mean square error (RMSE) were 0.769 and 1.342, respectively. The validation R 2 and RMSE were 0.774 and 1.975, respectively, and the relative prediction deviation (RPD) was 2.963. The SS estimation results based on BPNN model were consistent with those of the field investigation. Rapid and accurate inversion of SS based on UAV multispectral technique was achieved in this study, which provides technical support for regional management.
(© 2022. The Author(s), under exclusive licence to Springer Nature Switzerland AG.)
Databáze: MEDLINE
Externí odkaz: