Effects of color temperature of visible light camera on vegetation index calculation in unmanned aerial vehicle vegetation remote sensing.

Vegetation indices based on the spectral information of the RGB three-bands have been widely used in the field of unmanned aerial vehicles (UAVs) vegetation remote sensing. Affected by the automatic color temperature adjustment of cameras, the phenomena of color cast often occur when UAVs acquire RGB images. However, its impacts on the calculations of vegetation indices remain largely unknown. We analyzed the changing trends of 13 vegetation indices for non-vegetation objects, coniferous trees, broad-leaved trees and herbs under the color temperature gradient from 5000 K to 8000 K. The results showed that changes in color temperature significantly affected the calculation results of visible light vegetation indices. With increasing color temperature, the 13 vegetation indices exhibited five changing trends, namely rising, falling, unimodal, stable, and fluctuating. The trends of the normalized green-blue difference index, excess green index, the normalized green-red difference index, modified normalized green-red difference index, visible atmospherically resistant index, blue-green ratio vegetation index, blue-red ratio vegetation index, improved dual greenness index and greenness percentage index with the change in color temperature were affected by neither the object types nor the species. The trends of the triangular greenness index, green leaf index and red-green-blue vegetation index with change in color temperature were affected by both the object types and species identity. The trend of the normalized blue index with change in color temperature was affected by object types but not by species identity. The inter-specific differences within the same vegetation index could be amplified with the increase in color temperature. Weather conditions during image acquisition were a crucial factor influencing data quality. When utilizing vegetation indices calculated from UAV RGB imagery to interpret ecological patterns, strict quality control at the data collection process was necessary. Otherwise, the interpretability of vegetation indices would be weakened due to the impact of color temperature.