Abstract: Heavy metal pollution in farmland soils has become an environmental issue that cannot be overlooked in modern society. Traditional chemical detection methods are relatively outdated, and it is urgently needed to carry out dynamic, rapid and largescale monitoring of soil heavy metals. In this study, heavy metal pollutions in farmland soils were monitored by using spectral data obtained from portable matter spectrometer (ASD) and onboard hyperspectral imaging system (GaiaSky-mini) respectively. The results showed that GaiaSky-mini spectroscopy can discriminate the maximum and minimum value of As from partial least squares regression (PLSR) prediction model of As, Cd, Cr, and Pb, and roughly estimate the Cd content of samples, with residual predictive deviation (RPD) values being 1.13 and 1.50, respectively. The ASD spectral data can roughly estimate As content and quantitatively estimate Cd content, with RPD values of 1.45 and 1.95 for As and Cd, respectively. By comparing the regression results of two data sources, it is suggested that spectral data acquired from low-altitude unmanned aerial vehicle (UAV) can monitor heavy metals in soil with wider spectral bands and higher spectral resolution, whereas ASD spectral data has higher precision. It is important to apply low-altitude UAV hyperspectral data in further research to carry out rapid and large-scale monitoring of soil heavy metal content.

Key words: productivity., grassland, precipitation regime