Dynamics Beyond the NIR for the Vegetation Index: A Test Case of Various Sample Spectral Signature of Differing Leaf State of Health in the Shortwave Spectrum Bands

Bernardino Buenaobra1,2 , Mark Manhuyod4 , Annie Diola1,3 , Claurice Mangle1
Michelle Abarca4 , Benjamin Mabanta4 and Roland Otadoy1,2

1USC Phil-LiDAR Research Center, University of San Carlos
Nasipit Rd. Talamban, Cebu City Cebu Philippines 6000
Email: bjbuenaobra@usc.edu.ph, clauricemangle@gmail.com
2Department of Physics, School of Arts and Sciences, University of San Carlos
Nasipit Rd. Talamban, Cebu City, Cebu, Philippines 6000
Email: rolandotadoy2012@gmail.com
3Department of Biology, School of Arts and Sciences, University of San Carlos
Nasipit Rd. Talamban, Cebu City, Cebu, Philippines 6000
4Department of Geomatics and Analytics, Del Monte Philippines Inc.
M.Fortich,Bukidnon, Philippines 8705
Email: manhuyodjmv@delmonte-phil.com, abarcamb@delmonte-phil.com
mabantabz@delmonte-phil.com

 

KEYWORDS: shortwave, NIR, vegetation index, spectral signature

 

ABSTRACT: We report in this paper the results of spectroradiometric measurement in determination of the complete spectral signature of sample leafs with distinct and differing quality conditions by obtaining their response in the visible wavelengths (VIS), at near infrared (NIR) to the shortwave (SW) bands. Here we refer to the dynamics beyond the NIR as the extended range for the spectral response of the instrument from 1000nm to 2500nm. At these bands they indicate their relative reflectance and absorption values that translates to the quantification and indications of health of cell structure, water content, leaf biochemical, protein lignin, and celluse as well of the presence of nitrogen content. We obtained the numerical result by simple graphical analysis from the spectral plots at certain wavelengths for the vegetation index of interest.The reflectance values from 1510nm will yield the Normalized Difference Nitrogen Index (NDNI) where nitrogen concentration and overall biomass in foliage could be indicated. At 1754nm the Normalized Difference Lignin Index (NDLI) estimate the relative amounts of lignin contained in vegetation canopies. At 2000nm to 2200nm range the Cellulose Absorption Index (CAI) quantifies exposed surfaces that contain dried plant material and has absorption present that could indicate strong presence of cellulose. Canopy water content vegetation indices were also calculated from the response in the NIR with SW to provide an indication of the measured amount of water contained in the canopy. The Normalized Difference Water Index (NDWI) reflectance, which occurs at 857nm and 1241nm is known to be strongly related to the plant water content. This is interpreted to the effect that with higher water content often indicates healthier vegetation.The Moisture Stress Index (MSI) is sensitive to the effects of drought or catastrophic wetness increases in leaf water content at the absorption around 1599nm as well at 819nm and 1649nm that describes the Normalized Difference Infrared Index (NDII) values which uses a normalized difference formulation instead of a simplec ratio, with the index value increase with increasing water content. When compared with those spectral response in Landsat 8 OLI and MODIS spectrometer it did not yield numerical results since they did not yield the required frequencies at their spectral band specifications. These spectroscopy based measurements will serve for the purposes in the agricultural development efforts that could result in improved processes in agricultural crop management, plant canopy monitoring, and in the quantifying stressed vegetation .

 

1 INTRODUCTION

In the usual measurement of vegetation relative reflectance, the analyst concerns oneself of obtaining relative reflectance of the sample leaf or spectral signature of a canopy in the absorption the visible bands Blue, Green, Red bands are and where the invisible light are reflected in the near infrared NIR peak happen. This is the case for usual measurement of vegetation vigor for the Normalized Difference Vegetation Index (NDVI) that only requires the aforementioned bands. However, the complete picture of things for a spectral signature would extend beyond 1000nm to as far as 2500nm in those bands hence, an entirely new set of vegetation indices are defined. In this paper we performed laboratory hand measurements of sample leafs selected with a-priori health condition based on observation and field experience.

 

1.1 The complete vegetation spectra resulting from light interaction mechanism from VIS, NIR and the SWIR

In Figure 1 we show a practical plot from a healthy leaf obtained by hand held instrument (ASD Fieldspec c Spectroradiometer) done in house. We reference to this actual measurement to show the three important ranges of bands for spectral response of the plant leaf. A color bar for purposes of visualization of spectral colors in the visible range is shown as inset. In the first part the VIS (Visible range) (380nm-740nm) it is where the B-G-R spectral colors are present. In this range it is known that the leaf exhibit the absorption response in the Blue and Red Band and the reflection in the Green band hence also known as the chlorophyll bump in these area. At this shorter and higher frequencies (4-7.5 ×1014 Hz) wavelengths (750 – 400 nm) and at some higher quantum energy (1.65-3.1 eV) than that of NIR and SWIR (Hyperphysics 2016), the VIS section of the spectra is describing graphically the mechanism in the leaf pigments in their chlorophyll absorption. At the level of visible light interaction with matter. The primary mechanism for the absorption of visible light photons is the elevation of electrons to higher energy levels with many available states. The visible radiation is able to provide the photons absorbed by matter to an electron level transition alone and thus no ionization ever take place.These VIS bands when IR and SWIR bands are included comprises the complete bands which a full signature can be visualized (Shaw,G.A., Burke,H.K., 2003).

fig8

 

Near infrared (NIR) since it is adjacent to the visible spectrum that extend up the the low frequency at the Red band with frequencies .003-4 ×1014 Hz. For healthy vegetation and leaf, this region is where the highest reflectance occurs and where leaf or vegetation cell structure will be of interest as primary mechanism acts at this level. In contrast with VIS which absorbs Blue and Red radiation and partially reflects Green spectral colors, near infrared is at the level of light interaction with matter acts to set molecules into an overtone vibrations since having higher quantum energies (0.0012-1.65 eV) at wavelengths (0.8-2500nm). At these bands molecules can absorb certain resonant frequencies i.e. the frequency of absorbed radiation matches the transition energy of the group that vibrates. Infrared radiation can lead to molecular vibration that can cause heating and depth penetration, these overtone vibrations maybe different according to the molecule that absorbs it giving specific vibrational modes and degrees of freedom that also depend on molecules being linear and non-linear after (Stuart,B., 2004).

 

The range (0.8-2.5µm) with the longer wavelength what is called far infrared is shown in the Figure 1 as SWIR (Short Wave Infrared) which was adapted following remote sensing usage other than from chemistry; particularly those that would concern with Landsat 7 ETM+ and Landsat 8 OLI where the full bands have their discrete bands dedicated to IR and SWIR. Commercial spectrometers are commonly starting from 350nm, 400nm and ends at 1000nm (USB Ocean Optics US) while those that extends from it are called Spectroradiometers (ASD Fieldspec US) instruments that covers everything from 350nm to as far as 3500nm.The mechanism for SWIR vibrational modes are the same as those already described for NIR. In the SWIR the response of the vegetation are descriptive of water content, the leaf biochemical, protein lignin and cellulose. In the following pages we detail the calculations for a select vegetation indices for our purpose.

 

2 MEASUREMENT SET-UP AND WORKFLOW

The measurement of the complete spectra of experimental sample specimens begin with identification of various stages and condition of health state of the plant leafs. The classified leafs are then labeled and marked and sequenced for for hand held test. The instrument (ASD FieldSpec c Spectrometer) is calibrated using the white sulfur standard for calibration and background subtraction. The leafs then are measured and the measurements are logged on a computer. They are processed by an spreadsheet program and plotted. The plots are used to calculate by graphical manner to obtain the the vegetation index of interest. The results provides a baseline for establishing the signatures of afflicted plants, they are recorded and archived until the next opportunity to get newer samples. Optional but not required is the comparison of the results when the spectra is plotted over the Landsat 7 ETM+ or in Landsat 8 OLI and determine the relative spectral response from their shortwave bands. The basic set-up is shown in Figure 2 below:

fig9

 

3 RESULTS AND DISCUSSION

We present the following results from the calculation of selected IR vegetation indices that are beyond the 1000nm band. The definition and formulas are given in Table 1 as well as the indication of possible range and typical values to expect for green vegetation. A representative graphical solution is provided mainly for the sample that are normal and then followed with those samples that indicates a known deviation manifested by their leaf features these specimen have been coded in Table 3.

3.1 Selected IR Vegetation Indices Definitions and Formulas

 

The definitions and formulas of the selected vegetation indices were all obtained obtained and condensed from a white paper after (Harris Geospatial 2016) and is depicted in the Table 1 below:

fig10

 

3.2 Summary of results of SWIR Vegetation Indices from selected samples

In order to facilitate the determination of these values in a long list of 350nm – 2500nm and their one to one correspondence with measured reflectance; we take note the fact that one sample measurement translates to 2150 rows by 2 columns for each spectral signature. By this column vector arrangement the data is search by use of ”VLOOKUP” function by the spreadsheet program e.g. =VLOOKUP(L2,$A$2:$B$2152,2,FALSE) where the $A$2 is the column of frequencies and $B$2152 is the column of corresponding relative reflectance and the L2 contains the cell for which what wavelength to look for in the lookup table.The basic spreadsheet structure is show in Table 2 below:
fig11

In Table 3 below we summarize the numerical results obtained from the spectral signature data for the specimen as tagged and identified by physical observation.

fig12

 

3.3 Characteristic spectral signatures results from known physical effects

We show the spectral plots of selected type of leaf characteristic for each of the identified samples with a-priori knowledge of health and vigor as they tagged and coded. The following Figures 3,4,5,6,7 that follows below are representative spectra of the characteristic effects.
fig13

fig14

3.4 Results from comparison with Landsat 8 OLI SWIR1 and SWR2 bands

We show the complete Landsat 8 OLI Bands in Figure 8 and while in Figure 9 we only show the infrared bands including the Cirrus band. We overlay the spectral signature of a healthy leaf on the IR bands and obtain those frequencies that would be usefull for the calculation of the vegetation indices as have been done in the previous section. We summarize these results in Table 4.
fig15
fig16

 

We see that with definition given in Table 1 when we consider the IR Bands in Landsat 8 OLI will not be able to yield results using the formulas since the frequencies of interest either lie outside their spectral response or either only one of the two needed wavelengths are available for calculation. These are specially having no response at 1510nm and 1680nm for NDNI, at 2000nm and 2200nm for CAI, 819nm for MSI and NDII, 1241nm for NDWI, 1680nm and 1764nm for NDLI. We conclude that on the extended range beyond NIR the vegetation indices cannot be compared with Landsat 8 OLI bands. However, at the VIS range up to 800nm becomes possible since at these bands both the ASD FieldSpec c and Landsat 8 OLI have spectral response present as shown by (Woranut ,2011) using Moderate Resolution Imaging Spectroradiometer (MODIS) bands.

 

3.5 Results from comparison with MODIS Land, Cloud and Aerosols Band

We compare also the results from the ASD Spectroradiometer c with the Moderate Resolution Imaging Spectrometer after (MODIS Specification, 2016) from a subset of its 36 bands particularly those first seven bands herein tabulated below in Table 5.
fig17

 

fig18

When we consider MODIS spectroradiometer bands for the calculation the IR vegetation indices we arrive at the same problem in obtaining the required frequencies in band in MODIS as for NDNI, CAI, MSI, DNII, NDWI and NDLI hence, even at a smulated response using these bands we cannot have a correlation at specific bands as with the field instrument.

 

4 CONCLUSIONS AND RECOMMENDATIONS

The results from the hand held spectroradiometer instrument measurement operating from 350-2500nm for known samples have shown values as expected in the typical green vegetation figures. It is also true as well as for those expected range for those otherwise. However, the lack of the proper spectral response from both Landsat 8 OLI and with MODIS have not provided a validation and comparative analysis of results between the instrument and the simulated response using their IR available bands. It maybe possible to make adjustments to the definition for the IR vegetation indices at those specific frequencies for which they were defined so that the Shortwave bands in Landat 8 OLI can useful and well as the MODIS of the many groups of 36 bands available. An experimental verification maybe needed to observe at a molecule level if the energies of a new selection of the wavelength will have quantum energies that will result to vibration of atoms to warrant it to be give reflectance at those select frequencies.At the end of this experimental work we have also produced the spectral signatures of known states of leaf health.

 

Acknowledgments

We would like to thank firstly the PCIEERD from the national Department of Science and Technology (DOST) for their support for the Philippine LiDAR Resource Mapping project specially their encouragement to report and publish results as a research component to the national project. To our generous corporate co-authors and company the Del Monte Company Philippines Inc. from the pineapple plantation in Manolo Fortich in Bukidnon, Cagayan de Oro province for sponsoring the first author in several plantation visits and flight missions. We thank the Department of Geomatics and Analytics with their senior manager Mr. Benjamin Manbanta with his lead and innovations for leading the UAV team and specially providing the authors to entrust us the qualification of the ASD Fieldspec c spectrometer instrument. The project leader and co-author Dr. Roland Otadoy, the Chiefs, Engr.Aure Oraya of LiDAR1 and Renante Violanda of LIDAR2 components for the time and resources in the development of this paper at the USC Phil-LiDAR Research Center. Dr. Danilo Dy from the Department of Biology for the discussion on plant biology. We thank Mr. Kremlin D. Reyes Account Executive for Scientific Instrumentation Technology of the DKSH Philippines, Inc. for allowing us the trial use for evaluation purposes of their spectrometer instrument for this study

References

Harris Geospatal, 2016. Vegetation Analysis: Using Vegetation Indices in ENVI., Retrieved September 14, 2016,
from http://www.harrisgeospatial.com/Learn/WhitepapersDetail/TabId/802/ArtMID/2627/ArticleID/
13742/ Vegetation-Analysis-Using-Vegetation-Indices-in-ENVI.aspx

Hyperphysics , 2016. The interaction of radiation with matter., Retrieved September 13, 2016, from
http://hyperphysics.phy-astr.gsu.edu/hbase/mod3.html

MODIS Specifications, 2016. MODIS Moderate Resolution Imaging Spectradiometer.,
Retrieved September 19, 2016, from: https://modis.gsfc.nasa.gov/about/specifications.php

Shaw,G.A., Burke,H.K., 2003. Spectral Imaging for Remote Sensing. 14(1), Lincoln Laboratory Journal, Massachusetts Institute of Technology.
pp.3-28

Stuart, B., 2004. Infrared Spectroscopy: Fundamentals and Applications. John Wiley and Sons, West Sussex,
pp. 11-16.

Woranut, C., et al., 2011. Comparison of Surface Reflectance between Field Spectroradiometer and Multi
Satellite Image for Crop Monitoring
Proceedings of the Asian Conference on Remote Sensing (ACRS) 2011.,Retrieved September 18, 2016 from
http://a-a-r-s.org/acrs/index.php/acrs/acrs-overview/proceedings-1?view=publication&task=show&id=963