Abstract: In the present study, 6 kinds of vegetation indexes (VEG, CIVE, EXG, EXGR, NGRDI, COM) were selected to estimate vegetation coverage of grassland based on RGB channel information, and their effects were compared with self-developed APP. Meanwhile, the influence on estimation error was discussed from 3 aspects including the green degree, weather conditions and vegetation coverage. Besides, the stability of each method was estimated from the perspective of threshold variation with illumination. It was shown that the estimation accuracy of 6 methods was higher than 90%. COM method reached the maximum accuracy of 95.41%, while, NGRDI method exhibited the lowest accuracy of 92.87%. The estimation error of each method for dark green grass was smaller than that of yellow green grass, and the estimation error in cloudy condition (cloudiness≥70%) was smaller than that in sunny condition (cloudiness≤10%). With the increase of vegetation coverage, the estimation error of VEG, CIVE, EXG, COM increased, while, no regular changes on estimation error of EXGR, NGRDI method were found. All 6 methods showed no obvious changes (threshold fluctuation≤0.02) with the changes of solar zenith angle and light intensity in a day. Under different weather conditions, the thresholds of VEG, CIVE methods had no obvious change (threshold fluctuation≤0.01), while, the thresholds of the remaining methods changed significantly (threshold fluctuation≥0.03). In short, all the 6 methods could satisfy the application in cellphone platform. As there was no need to consider the impact of illumination changes, VEG and CIVE possessed better versatility compared with the other methods.

Key words: vegetation coverage, digital photos, grassland, vegetation index