Fuzzy Logic-Based Visualization and Interpretation of NDVI Maps
Andrei Racoviţeanu, Artur Kazak, Nicolae Secrieru, Mihai Ivanovici · 2024
Normalized Difference Vegetation Index (NDVI) is widely-used for the monitoring of agriculture crops. Its interpretation is based on thresholds and the colormaps used for the visualization of NDVI maps are discrete. We propose a fuzzy logic approach for both the visualization and the interpretation of the NDVI maps. The rationale is that the NDVI values of agricultural crops exhibit a certain spread, while the interpretation based on thresholds is prone to a certain degree of incertitude. We define several membership functions for modeling three sub intervals of the NDVI values in the [0, 1] range, useful for the interpretation of the vegetation status for agricultural crops. The three sub intervals are corresponding to three categories of vegetation status: dried-out vegetation, moderately healthy and strongly developed. For a Sentinel-2 time series, we compare the visualization using the crisp logic, the Sentinel-2 color palette, and the color palettes resulting from the proposed membership functions implementing the fuzzy logic. We show that our colormap has a smoother color gradient, which can facilitate a more visually-appealing and interpretable transition of colors. We also show an example of inference for the interpretation of the NDVI values for an operational usage of the proposed fuzzy logic approach.