Comparing Common Hyperspectral Vegetation Indices for Their Ability to Estimate Seasonal Nitrogen and Other Variables in Winter Wheat Across a Growing Season

Abstract

Abstract Field spectroscopy was used to assess the nitrogen status and monitor crop growth and production of wheat (Triticum aestivum L, cv. Chara) under a range of nitrogen fertilizer treatments and two irrigation levels (rainfed and irrigated) throughout the growing season. The association between a range of commonly used vegetation indices for the detection of green biomass, nitrogen or chlorophyll concentration, and measured crop characteristics (nitrogen concentration, total nitrogen accumulated, dry weight, wet weight, relative foliar cover and moisture content) was determined for six moments in the growing season. At about 95–144 days after sowing, a strong correlation (r² between 0.7 and 0.99, p<0.01) existed between vegetation indices and wheat biomass, relative foliar cover and total nitrogen load per square meter. However this association did not hold when multiple sample dates were combined. Changes in total biomass and/or foliar cover strongly affected the association between vegetation measures and the vegetation indices, making between-date comparisons difficult.

Appendix: Summary of the Indices Used in this Study

Indices Typically Used for Biomass (Green Leaf Area)

NDVI (Normalized difference vegetation index). Commonly used to assess the amount of green biomass, uses the difference between reflectance around 800 nm, which is influenced by cell-wall scattering, and reflectance around 670 nm, which is influenced by chlorophyll absorption.

OSAVI (Optimized Soil Adjusted Vegetation Index (Haboudane et al., 2002)). This index is based on the Soil Adjusted Vegetation Index (Huete, 1988). Effectively chlorophyll absorption around 670 nm is used as an index, which is stabilized for soil background using reflectance in the NIR (800 nm).

MSAVI (Modified Soil Adjusted Vegetation Index (Qi et al., 1994)). This is another index based on SAVI. MSAVI minimizes the effect of soil-background on vegetation signals using a soil-background adjustment function, which has been empirically derived from NDVI and WDVI. Effectively chlorophyll absorption around 670 nm is used as an index, which is stabilized for soil background using linear and quadratic reflectance in the NIR (800 nm).

RDVI (Re-normalized Difference Vegetation Index (Haboudane et al., 2004)). Similar to the NDVI, this index uses the ratio between the sum and the difference of a band in a chlorophyll absorption feature (670 nm) and a band in cell-wall scattering region (800 nm) to predict biomass. To reduce saturation effects at higher biomass levels, the square root is taken of the nominator.

Indices Typically Used for Chlorophyll/Foliar N

LAIVI (Leaf Area Insensitive Vegetation Index (Haboudane et al., 2002)) (TCARI/OSAVI). In order to reduce the effects of non-closed canopies and soil-reflectance on TCARI, TCARI is divided by OSAVI, which results in LAIVI.

MCARI (Modified Chlorophyll Absorption index (Daughtry et al., 1992)). A measure of the depth of chlorophyll absorption at 670 nm relative to the reflectance at 550 and 700 nm,

mSR ₇₀₅ . (Modified simple ratio index around 705 nm (Sims and Gamon, 2002)). Based on a simple ratio index. Reflectance is normalized by subtracting a stable reflectance (e.g., reflectance at a saturated absorption feature 445 nm) from a reference index (750 nm) and dividing it by the difference between the stable reflectance and an index wavelength (705 nm).

mND ₇₀₅ (Modified Normalized Difference Vegetation Index (Sims and Gamon, 2002)). Based on normalised ratio indices such as NDVI. The index is based on the difference between two bands, and normalized for shifts in overall intensity by dividing by the sum of the index and reference bands. To further reduce the effect of surface reflectance, the reflectance of a saturated absorption feature is subtracted from the nominator.

NRI _693,1559 (Normalized ratio index at 1559 (Ferwerda et al., 2005)). A variation on the NDVI, this index uses the less-saturated part of chlorophyll absorption at 693 nm as an index, and a foliar water absorption feature around 1559 nm as reference.

NR _693,1770 (Normalized ratio index at 1770 (Ferwerda et al., 2005)). A variation on the NDVI, this index uses the less-saturated part of chlorophyll absorption at 693 nm as an index, and a carbon-hydrogen absorption feature as reference.

PRE (Position of the red edge (Gong et al., 2002)). The wavelength where the first derivative reaches a maximum, within the spectral region of the red edge 670–740 nm.

PYE (Position of the yellow edge (Gong et al., 2002)). The wavelength where the first derivative reaches a maximum, within the spectral region of the yellow edge (550–582 nm).

PIRE (Position of the infrared edge (Gong et al., 2002)). The wavelength where the first derivative reaches a maximum, within the spectral region of the infrared edge (1300–1460 nm).

TCARI (Transformed Chlorophyll Absorption Index (Haboudane et al., 2002)). A transformed variant of the chlorophyll index MCARI, which is more sensitive to low chlorophyll values and more resistant to vegetation non-photosynthetic materials.

TVI (Triangular Vegetation Index (Broge and Leblanc, 2000)). Promoted as a general vegetation index, it uses an absorption ‘triangle’ to quantify the depth of the chlorophyll absorption feature between 550 and 750 nm, centred at 670 nm.