A comparison of the predictive potential of various vegetation indices for leaf chlorophyll content

Leaf chlorophyll is an important indicator of nutritional stress, photosynthetic capacity, and growth status of plants. Therefore, it is important to accurately estimate leaf chlorophyll content over a range of spatial and temporal scales. However, traditional methods for chlorophyll measurement mainly rely on chemical analysis, which is not universally applicable to dynamic monitoring of plants in large areas because it is time consuming and requires destructive measurements. Hyperspectral remote sensing has enormous potential for accurate retrieval of plant biochemical parameters. Therefore, it has become the most popular means to retrieve chlorophyll content, by establishing empirical relationships between different vegetation indices and chlorophyll content. In recent years, many vegetation indices have been developed for the inversion of chlorophyll content. Only the vegetation indices that are less affected by the external environment are used, in order to establish a strong and robust model. Therefore, it is very important to assess the anti-disturbance ability of different vegetation indices. In this paper, a new method is proposed to quantitatively assess the anti-disturbance ability of vegetation indices. The anti-disturbance ability of several vegetation indices that are commonly used for chlorophyll estimation was evaluated using this method. We concluded that the anti-disturbance ability of VogD, TCARI/OSAVI, and Datt99 is stronger than the other vegetation indices tested. However, if a vegetation index is not sensitive to chlorophyll content, predicting chlorophyll content holds no value even though it has a strong anti-disturbance ability. Therefore, we used the slope of the best fitting function between the vegetation index and chlorophyll content (defined as d(VI)/d(Chlorophyll content)) to measure the sensitivities of different indices to chlorophyll content. Finally, we found that TCARI/OSAVI was one of the best vegetation indices to estimate chlorophyll content at the leaf level. However, we have only considered three factors to evaluate the anti-disturbance ability of different vegetation indices, which is still far from enough because the canopy reflectance is affected by many factors. Therefore, we should account for more factors in the future.