Abstract: Rapid, high-precision and low-cost monitoring of vegetation status in small watersheds is basis forevaluating the efficacy of returning farmland to grassland and ecological environment construction. In order to enrichthe diversity of vegetation index, this paper aims to propose a visible light vegetation index suitable for monitoringthe vegetation status of small watersheds on the Loess Plateau. Based on the orthophoto images of small watershedsobtained by the UAV remote sensing technology, a simplified visible light vegetation index (SVVI) was developed fortwo small typical watersheds on the Loess Plateau. Then, the threshold was determined by the sample statistics, and theextraction effects of eight common visible vegetation indexes were compared with the supervised classification resultsof support vector machine (SVM). Finally, the accuracy and applicability of the SVVI were tested by the confusionmatrix. The results showed: 1) The SVVI can effectively suppress the information of non-vegetation features. Whenextracting areas with rich types of features and relatively low vegetation coverage, the extraction accuracy of SVVIreached as high as 96%. 2) In the verification area with relatively single features and high vegetation coverage, theextraction accuracy of SVVI was still more than 90%, indicating that SVVI had a good applicability. Compared withthe supervised classification results, the vegetation coverage calculation based on the SVVI can effectively retain thevegetation information and realize high-precision monitoring of vegetation status in small watershed of Loess Plateau.