Canopy Spectral Reflectance Feature and Leaf Water Potential of Sugarcane Inversion

Available online at www.sciencedirect.com

Physics Procedia 25 (2012) 595 – 600

2012 International Conference on Solid State Devices and Materials Science

Canopy Spectral Reflectance Feature and Leaf Water Potential of Sugarcane Inversion * Haibo Chen 1, Pei Wang1 and Jiuhao Li (corresponding author) 1 , Jingdong Zhang 2 and Luxiang Zhong 2 1

Key Laboratory of Key Technology on Agricultural Machine and Equipment Ministry of Education, South China Agricultural University, Guangzhou, China 2

College of Engineering, South China Agricultural University, Guangzhou, China

Abstract The canopy hyperspectral reflectance data and leaf water potential were recorded at tillering and elongation stage of sugarcane. Statistic analysis method was conducted on the correlation between canopy reflectance data and leaf water potential, combining the visible light band (460nm or 560nm) and the infrared light band (860nm or 960nm or 1200nm) reflectance into vegetation indices of RVI (ratio vegetation index) and NDVI (normalized difference vegetation index), the ratio of RVI and NDVI was linearly related to leaf water potential. Five single variables of linear function models against the leaf water potential were established, the results showed that the five types of function models had significant correlation if ¢=0.01, the two model established by using of the band 460nm and 1200nm, 560nm and 960nm were optimal and the correlation coefficient was 0.927.

2011 Published ofof [name © 2012 Published by by Elsevier Elsevier Ltd. B.V.Selection Selectionand/or and/orpeer-review peer-reviewunder underresponsibility responsibility Garryorganizer] Lee Keywords :Sugarcane, Canopy, Leaf water potential, Spectral feature, Inversion.

1. Introduction Most sugarcane fields in Leizhou Peninsula of Guangdong Province are dry sloping land. The irregular precipitation and the dry spell in spring and winter lead to the low and unstable yield of sugarcane. The per-mu yield of sugarcane is between 4 and 5 ton. Water short is the main factor restricts the development of sugarcane industry in this area [1]. The leaf water potential, which indicates the water status of plants, is influenced not only by soil conditions, but also by meteorological conditions. The leaf water potential is relatively stable early in the morning, so it is more proper to use this value to indicate the water short of a plant [2]. Zhang and Rana studied the quantitative and qualitative relationship of the *

This work is supported by the Science and Technology Plan of Guangdong Province of China Grant #2008B030303034 to J.H.Li

1875-3892 © 2012 Published by Elsevier B.V. Selection and/or peer-review under responsibility of Garry Lee doi:10.1016/j.phpro.2012.03.131

596

Haibo Chen et al. / Physics Procedia 25 (2012) 595 – 600

critical value of leaf water potential and soil water potential, soil moisture content respectively [3,4]. Hu studied the crop-water relations under drought and found that predawn leaf water potential (PLWP) was fallen significantly with the decrease of soil water content, the relationship between PLWP and soil water content could be expressed by a retardant equation [5]. Although leaf water potential essentially reflects the condition of crop water, it is destructive to the sample and without an equipment to monitor it automatically, so it is difficult to provide timely, accurate information of crop water. The crop hyperspectral reflectance is a method to monitor the crop water status for a large-scale field without contact and destructive, it is one hotpot of the application of remote sensing technology in agriculture. Penuelas found that the ratio of WI and NDVI (WI=R900/R970, NDVI= (R900-R680/ (R900+R680)) can predict the moisture content of leaves and plant or canopy’s water content accurately [6]. Mao analyzed the hyperspectral reflectance data of grapewater leaf under water stress condition and established the model of the dry basis moisture content of grape leaves [7]. This article studied the relationship between canopy reflectance characteristics and leaf water potential in sugarcane under different irrigations, which provide the theoretical support for completing the models of sugarcane water status based on the spectral data. 2.

Materials and Methods

A. Design of experiment This detection study was conducted in a large canopy in South China Agricultural University in Guangzhou city, China, at Longitude 113.3475 and Latitude 23.1574. Experiments were conducted during 30th Mar to 15th Aug 2009 using pot culture sugarcane, the new sugarcane 22 (T22) were planted in 28 opaque plastic containers of gray color with a length and width of 0.60m and a height of 0.4m on 30th Mar 2009 (seed spacing, 0.6m×0.3m, for producing high vegetation cover to minimize soil influences). The soil was a red loam taken from the xingfu farm in Leizhou City, Zhanjiang, Guangdong Province. Twenty eight containers were divided in 7 plots and randomly arranged in the canopy to minimize shading effects. At tillering and elongation period sugarcane were arranged four different irrigation levels as normal irrigation (CK), light drought (L), middle drought (M) and serious drought (S) respectively to be equivalent to 70-80%, 60-70%, 50-60% and 40-50% of field water-holding capacity, each treatment was repeated four times. Other managements are the same except irrigation. B. Control of irrigation The soil moisture content were measured using a TRIME-P3 (German) every other 3 days, on the side of the container two holes were drilled at the height of 10cm, 20cm and 30cm for TRIME-P3. Each layer was measured three times, the mean of three layers was the last soil moisture content of this container and then to irrigate and control the soil moisture content to the required levels. C. Canopy Spectrum measurement with a hyperspectral spectrometry Canopy Spectrum of each sugarcane samples were measured using a FieldSpec® 3 Spectroradiometer (Analytical Spectral Devices, Inc., Boulder, Colorado, USA) fitted with a Pro Lamp Assembly (A128932). This Spectroradiometer is specifically designed to acquire visible near-infrared (VNIR) and short-wave infrared (SWIR) spectra. It is a compact, field portable and precision instrument with a spectral range of 350-2500 nm and a rapid data collection time of 0.1 second per spectrum. The Spectroradiometer detects 350-2500 nm continuous bands with spectral resolution: 3nm@350-1000nm, 7nm@1000-2500nm; sampling interval: 1.4nm@350-1000nm, 2nm@1000-2500nm. The canopies of the sugarcane clumps center of each plot under different water were selected for spectral measurement from 11:00 to 14:00 in a sunny and unclouded day. The Pro Lamp Assembly and the Fiber Optic Input (the bare fiber has a 25 degree field-of-view) of the Spectroradiometer were mounted over the canopy of sugarcane. The distance between sensors and canopy surface is about 50 cm and thus that the scanned

Haibo Chen et al. / Physics Procedia 25 (2012) 595 – 600

area diameter by the Spectroradiometer is about 22.5 cm. At least five consistent measurements were taken from the measuring canopies and each measurement was the mean of three consecutive individual full-range spectral scans. D. Leaf water potential measurement with a potential system Leaf water potential of each sugarcane samples were measured using PSYPRO, a water potential system (Wescor USA) from 8:00 to 18:00. The middle of the first unfold leaf on the top of sugarcane was selected for leaf water potential measurement, at least three consistent measurements were taken from sugarcanes of each plot under different treatments, the last leaf water potential data of each treatment was the mean of three measurements. 3.

Results and Discussion

Leaf water potential(Mpa)

A. leaf water potential characteristic under different irrigations As can be seen from Fig.1 and Fig.2, the leaf water potential of sugarcane showed significant diurnal variation as atmospheric conditions changed. There was obvious variance under different irrigations, but the curve’s general trend was consistent. At tillage stage, the maximum leaf water potential was around 8:00 in the morning, as temperature raised and relative humidity decreased, leaf water potential decreased, it dropped to the lowest value at about 14:00 pm, then as the temperature and atmospheric evaporation gradually decreased, the sugarcane absorbed water from the root achieved a balance between leaf water potential and soil water potential, leaf water potential gradually recovered. At elongation stage the maximum leaf water potential was around 8:00 in the morning, the lowest appeared at about at 14:00-16:00pm. The leaf water potential curve at tillering stage was steep and elongation curve was relatively flat. Leaf water potential decreased with the decrease of soil water content, it showed CK> L> M> S, but different variance showed at different times. Leaf water potential at about 14:00 pm was choose to establish relationship with the spectral reflectance. Time

 











    &.

/

0

6

Fig. 1 Sugarcane leaf water potential under different irrigations˄tillering stage). Time

 /HDIZ DWHUSRWHQWLDO0 SD











    

&.

/

0

6

Fig. 2 Sugarcane leaf water potential under different irrigations(elongation stage).



597

598

Haibo Chen et al. / Physics Procedia 25 (2012) 595 – 600

B. Spectral reflectance characteristic under different irrigations As shown in Fig.3 and Fig.4, the sugarcane canopy spectral reflectance was low in the visible light band (400-700nm), there was a clear reflection peaks between 550nm and 560nm, known as the "green peak", it’s reflectivity was generally about 10%-20%, this spectral feature was mainly affected by various pigments of sugarcane leaves (chlorophyll, lutein, carotene, anthocyanin, et al.) in which the concentration of chlorophyll played a leading role. The reflectivity and transmittance were very low because of these pigments’ strong absorption. The infrared light band (760-1200nm) reflectance was much higher than the visible light band’s, it had the wavy shape, a high reflectivity and a low absorption rate, the spectral reflectance intensity of the platform depends on the internal structure of sugarcane’s leaves. In the 1300-2500nm band, as the influence of atmospheric moisture, the spectral reflectance of sugarcane was instability, particularly there were water vapor absorption bands near 1400nm and 1900nm, and it significantly affected the absorption of atmospheric water vapor. Different irrigations remarkable affected sugarcane's growth, the canopy hyperspectral reflectance also had the corresponding changes. As can be seen from Fig.3 and Fig.4, there were obvious variance under different irrigations, the general trend of spectral curve was consistent, the same peaks and valleys, However, the band reflectance values were different, the canopy reflectance of most band was CK
Fig.3 sugarcane canopy spectral reflectance under different irrigations(tillering stage).

Fig.4 sugarcane canopy spectral reflectance under different irrigations(elongation stage).

599

Haibo Chen et al. / Physics Procedia 25 (2012) 595 – 600

C. inversion models of leaf water potential of sugarcane The vegetation index is widely uses for qualitative and the quantitative evaluation vegetation's cover and the growth condition parameter. The relation between leaf water potential and the vegetation index was analyzed, combining the visible light band (460nm or 560nm) and the infrared light band (860nm or 960nm or 1200nm) reflectance into vegetation indices of RVI (ratio vegetation index) and NDVI (normalized difference vegetation index), the ratio of RVI and NDVI was linearly related to leaf water potential. Five single variables of linear function models against the leaf water potential were established as shown in Table I: TABLE I Various inversion models of leaf water potential of sugarcane

Selected band

independent variable x

simulation equation

correlation coefficient

prediction

r˄n=32˅

r˄n=20˅

F

460nm,860nm

RRVI/RNDVI

Y=-0.125X-0.455

0.692

0.772

13.8

460nm,960nm

RRVI/RNDVI

Y=-0.247X+0.135

0.683

0.723

13.1

460nm,1200nm

RRVI/RNDVI

Y=-0.141X-0.084

0.718

0.927

15.9

560nm,960nm

RRVI/RNDVI

Y=-0.186X+0.204

0.754

0.927

19.8

560nm,1200nm

RRVI/RNDVI

Y=-0.823X+3.732

0.744

0.887

18.6

D. Accuracy test of prediction model In order to test five inversion models, 20 samples were used, the results were showed in Table I. As can be seen from Table I, the correlation coefficient of forecasting result was between 0.772 to 0.927, the mean value was 0.847, and it showed that the estimate result of the above 5 regression equation were quite ideal. The two models established by using of the band 460nm and 1200nm, 560nm and 960nm were optimal and the correlation coefficient was 0.927. Therefore, when using the canopy hyperspectral reflectance to monitor leaf water potential, these two combination model were optimal. 4.

Conclusion Based on the above results and discussion, it can be concluded that:

1. Leaf water potential decreased with the decrease of soil water content, It showed CK> L> M> S, the leaf water potential curve at tillering stage was steep and elongation curve was relatively flat. Leaf water potential at about 14:00pm was choose to establish relationship with the spectral reflectance. 2. The relation between leaf water potential and the vegetation index was analyzed, combining the visible light band (460nm, 560nm) and the infrared light band (860nm,960 nm,1200 nm) reflectance into vegetation indices of RVI (ratio vegetation index) and NDVI (normalized difference vegetation index), the ratio of RVI and NDVI was linearly related leaf water potential, five single variables of linear function models against the leaf water potential were established, the results showed that the five types of function models had significant correlation if Į=0.01, the model established by using of the band 560nm and 960nm was optimal and the correlation coefficient was 0.754. 3. Five inversion models were tested, the two models established by using of the band 460nm and 1200nm, 560nm and 960nm were optimal and the correlation coefficient was 0.927. It showed there was a good correlation between canopy reflectance and leaf water potential in sugarcane under different soil water.

600

Haibo Chen et al. / Physics Procedia 25 (2012) 595 – 600

4. Further studies are needed to identify bands and relative parameters that are most useful to monitor leaf water potential in sugarcane under field condition. References [1]

L.J.Zhang, Q.Mu, Y.Q.Lin, H.Zhang and R.K.Chen. Studies on the Relationship of Chlorophyll Fluorescence Characters

and Drought Tolerance in Seedling of Sugarcane Under Water Stress, Scientia Agriultura Sinica, vol.37, no.11, pp. 1718-1721, 2004. [2]

M.Mastrorilli, N.Katerji and G.Rana. Productivity and water use efficiency of sweet sorghum as affected by soil water

deficit occurring at different vegetative growth stages. European Journal of Agronomy, vol.11, pp. 207-215, 1999. [3]

Y.P.Zhang, W.Q.He and J.Han. Suitable Irrigation Indexes in Different Growing Stage of Corn.Irrigation and

Drainage,vol.20,no.4,pp. 18-20,2001. [4]

G.Rana, N.Katerji and M.Mastrorilli. Environmental and soil-plant parameters for modeling actual crop

evapotranspiration under water stress conditions. Ecological Modelling, vol.101, pp. 363-371, 1997. [5] J.C.Hu, W.X.Cao,D.Jiang and W.H.Luo. Quantification of Water Stress Factor for Crop Growth Simulationĉ.Effects of Drought and Water-logging Stress on Photosynthesis, Transpiration and Dry Matter.Acta agronomica sinica, vol.30, no. 4, pp. 315320, 2004. [6]

J.Penuelas, J.Pinol, R.Ogaya and I.Filella. Estimation of plant water concentration by the reflectance water index

WI(R900/R970).Intern-ational Journal of Remote Sensing, vol.18, pp. 2869-2872, 1997. [7]

H.P.Mao, X.D.Zhang, X.Li and Y.Zhang. Model establishment for grape leaves dry-basis moisture content based on

spectral signature.Journal of Jiangsu University (Natural Science Edition), vol.29, no.5, pp. 369-372, 2008. [8]

Y.C.Tian, C.Yong, W.X.Cao, D.Jiang and Y.Zhu. Relationship Between Canopy Reflectance and Leaf Water Potential in

Rice Under Different Soil Water and Nitrogen Conditions. Journal of Soil and Water Conservation, vol.17, no.3, pp. 178-183, 2003. [9]

P.Ceccato, S.Flasse and J.M.Gregoire. Designing a spectral index to estimate vegetation water content from remote

sensing data. Remote Sensing of Environment, no.82, pp. 198-207, 2002. [10]

A.S.Daniel and A.G.John. Estimation of vegetation water content and photosynthetic tissue area from spectral reflectance:

a comparison of indices based on liquid water and chlorophyll absorption features. Remote Sensing of Environment, vol.5798, pp.112, 2002.