Digital Image Analysis and Spectral Reflectance to Determine Turfgrass Quality

Abstract

The limitations of the conventional visual rating system used to assess turfgrass quality include its subjective nature and the need for properly trained observers who can discern differences among treatments or turfgrass varieties. The objective of our study was to investigate if digital image analysis (DIA) and spectral reflectance [normalized difference vegetative index (NDVI)] can be used to evaluate turfgrass varieties. Trials were established at New Mexico State University and visual quality ratings, digital images, and NDVI were collected monthly on three warm‐season and three cool‐season variety trials and on one cool‐season and one warm‐season mixed species trial. Correlations among quality, NDVI, dark green color index (DGCI) and percent green cover (PCov) were computed. Multiple regression was used to determine if combining NDVI and DIA improved the association between visual turfgrass quality and other variables. Quality was most strongly associated with NDVI (R2 ranging from 0.37 to 0.65) for most datasets. Additionally, multiple linear regressions identified NDVI as the variable affecting a higher change in R2 when entered to the model than either DGCI or PCov. Visual quality had a weaker association with sampling date than did NDVI or DGCI, which indicates that NDVI may track quality changes more reliably over time. However, a stronger association between variety and visual quality than between variety and NDVI or DGCI indicates that a visual assessment detects varietal differences better. Therefore, it is questionable whether visual assessments can be replaced by NDVI or DIA to characterize the aesthetic appeal of turfgrasses accurately.