Anomalous Variations of NDVI for a Non-Vegetated Urban Industrial Area of Gumi, Korea

Abstract

NDVI is often used to investigate how vegetation contributes to the surface urban heat islands (SUHI) effect.Generally, NDVIs obtained for partially vegetated or built-up areas have been found to be less than 0.4. However, anomalous NDVIs that are inconsistent with established concepts and relationships are frequently obtained for non-vegetated urban land cover. It is possible that these anomalous NDVIs have been distorted by the surface colors and patterns of rooftops. Therefore, we obtained NDVIs for an urban industrial area of Gumi, South Korea, and investigated how these NDVIs were affected by such surface colors and patterns. By assessing 148 factory rooftops, obtained NDVIs indeed showed a tendency to be affected by the surface colors and patterns of rooftops. A color classification of white, blue, purple, and red revealed that blue-and purple-colored rooftops resulted in higher NDVIs than other rooftops. Moreover, these rooftops were sometimes misidentified as forest and vegetation. The color tone, affected by brightness intensity, also contributed to the NDVIs obtained for blue-and purple-colored rooftops. Extreme cases showed that NDVIs obtained for rooftop surfaces could attain values indicative of dense vegetation (i.e., NDVIs ≥ 0.6), when blue-colored rooftops were combined with complex surface patterns that generated more shadow. From these results, we concluded that the established relationship between LST and NDVI is likely to be invalid for non-vegetated urban industrial areas, and that NDVIs obtained for such areas should be used with caution, particularly in studies of the SUHI effect.

Variabilidade e tendências de temperatura no Sudeste do Brasil: uma análise em escalas local e regional

“…The cloud cover response acts by reducing temperature as cloud cover increases, due to the increased albedo cooling effect, and it is statistically significant in GAM only for Tmax a rough estimate, we suppose a NDVI of 0.4 as a threshold in which below this value the vegetation cover is too low, with a predominance of built-up areas, and 0.65 as a threshold which above this value there is a high vegetation cover (Lambin and Ehrlich, 1996;Bhang, 2014), even though there are no exact values of NDVI that separate areas with low or high vegetation cover, especially when considering urban areas. In our sample data, we have 54 % of stations with NDVI < 0.4 and none of them with NDVI > 0.65.…”