Revised normalized difference nitrogen index (NDNI) for estimating canopy nitrogen concentration in wetlands

Abstract

Nitrogen is an important element for plant growth, photosynthesis and productivity. It has been presented to have a close correlation with wetland purification and estuarine eutrophication. The quantification of nitrogen concentration in wetland plants from hyperspectral data can provide valuable information to help develop approaches to research wetland ecosystem functioning and biophysical processes. This paper developed a revised nitrogen index (revised NDNI) on the basis of NDNI and SAVI to estimate canopy nitrogen concentration of wetland reed from ground-measured and spaceborne hyperspectral data while reducing wetland background effects. The SAVI was revised through replacing the red band by the 1510nm band and a soil self-adjustable factor to increase vegetation sensitivity and diminish background noise. The revised NDNI integrated advantages of NDNI that was sensitive to canopy nitrogen concentration and revised SAVI that was used as the denominator in revised NDNI to reduce contributions of wetland background reflectance. The results showed that the revised NDNI was more robust than NDNI for nitrogen predictions of reeds, with a relatively higher accuracy (R2=0.84, 0.88, 0.78). It was more insensitive to the variations of LAI than the compared indices, with a lower correlation coefficient (R2=0.27). These results demonstrated that the revised NDNI was both sensitive to canopy nitrogen status of reeds and resistant to the background effects. A validation was performed using Hyperion image over two experimental fields in the study area. The results of canopy nitrogen concentration estimations exhibited a good correlation with ground-based measurements (R2=0.79). The revised NDNI was proved to be more suitable for canopy nitrogen concentration prediction of aquatic vegetation (such as Phragmites australis) in the context of wetlands.