Assessing newly developed and published vegetation indices for estimating rice leaf nitrogen concentration with ground- and space-based hyperspectral reflectance

Non-destructive and quick assessment of leaf nitrogen (N) status is important for dynamic management of nitrogen nutrition and productivity forecast in crop production. This research was undertaken to make a systematic analysis on the quantitative relationship of leaf nitrogen concentrations (LNCs) to different hyperspectral vegetation indices with multiple field experiments under varied nitrogen rates and varied types in rice (Oryza sativa L.). The results showed that some published indices had good relations with LNC such as two-band indices, R750/R710 (ZM), Gitelson and Merzlyak index two (GM-2), R735/R720 (RI-1dB), R738/R720 (RI-2dB) and the normalized difference red edge index (NDRE), three-band indices, adjusted normalized index 705 (mND705), physiological reflectance index c (PRIc), terrestrial chlorophyll index (MTCI), and red edge position derived with four point linear interpolation (REP_LI). Three-band indices performed better than two-band indices, with MTCI exhibiting the best logarithmic relation to LNC in rice. Then, hyper-spectral vegetation indices computed with random two bands (λ1 and λ2) from 400 to 2500 nm range were related to LNC of rice. The results indicated that two-band indices combined with bands of 550–600 nm and 500–550 nm in green region had good relationships with LNC, and simple ratio index SR(533,565) performed the best in all two-band indices, similar to the published three-band indices (mND705, PRIc and MTCI). New three-band indices R434/(R496 + R401) and R705/(R717 + R491) were proposed for prediction of LNC with improved ability over the SR(533,565) and published spectral indices. Moreover, R705/(R717 + R491) performed well in all the data from ground spectra, modeled AVIRIS and Hyperion spectra, and acquired Hyperion image hyperspectra. The R434/(R496 + R401) also exhibited well in both ground and modeled AVIRIS and Hyperion image spectra, but could not be tested with the acquired Hyperion image because of the absence in radiometric calibration of the bands less than 416 nm. Overall, the newly developed three-band spectral index R705/(R717 + R491) should be a good indicator of LNC at ground and space scales in rice. Yet, these new indices still need to be tested with more remote sensors based on ground, airborne and spaceborne, and verified widely in other ecological locations involving different cultivars and production systems.

Research highlights▶ A systematic analysis was made on the quantitative relationship of leaf nitrogen concentrations (LNCs) to different hyper-spectral vegetation indices (newly developed and published) with multiple field experiments under varied nitrogen rates and varied types in rice. ▶ The new simple ratio index SR(533,565) was developed for estimation of LNC in rice, which performed the best in all two-band indices, similar to the published three-band indices (mND705, PRIc and MTCI). ▶ The newly developed three-band index R705/(R717 + R491) was proposed for prediction of LNC with improved ability over all two-band, three-band and published spectral indices. ▶ All newly developed and published indices were tested with ground-based hyperspectra data of different bandwidths, modeled airborne and space-borne spectra (AVIRIS and Hyperion), and acquired space-borne Hyperion image hyperspectral data, and R705/(R717 + R491) is recommended for reliably estimating rice LNC at ground and space scales.