Remote Sensing assessment of water stress effects on wheat

Remote Sensing assessment of water stress effects on wheat

R. K. Mahey, Rajwant Singh
S. S. Sidhu, R. S. Narang
Department of Agronomy Punjab Agricultural University
Ludhiana-141 004, India

Abstract
In order to monitor vegetation conditions, remote sensing techniques have been used successfully for several crop covers. The objectives of the study was to investigate the potential usefulness of spectral measurements to estimate leaf area index, biomass and detecting water stress in wheat (Triticum aestivum L.) Ground-based radiometric measurements were made on seven irrigation treatments throughout a complete crop cycle in order to monitor wheat growth and development under irrigated and stressed conditions in an experimental field at Punjab Agricultural University, Ludhiana during winter season of 1987-88 and 1988-89. Reflectance in the red (625 to 689 nm) and infrared (760 to 897 nm) bands was measured with hand held radiometer. Concomitant measurements of some of the agronomic variables were also made. Canopy air temperature difference(DT) was recorded at maximum crop cover stage. Spectral data have been correlate with plant height, leaf-area index, total wet/dry biomass, plant water content, grain yield, consumptive use of the crop and canopy air temperature difference. The results shows a significant correlation between spectral data derived from near infrared, red radiances and various agronomic variables. Infrared: red reflectance ratio (R) and normalized difference (ND) vegetation index were found highly and linearly correlated with yield establishing the potential of remote sensing for predicting grain yields. The correlation for R and ND was maximum during 75-104 and 76-102 days after sowing respectively during the two seasons. DT was also significantly related to yield as well as spectral parameters. The different temporal spectral response under no irrigation treatments also showed the usefulness of spectral measurements in detecting water stress effects on crop. So the results of the experiments show conclusively that a hand held radiometer can be used to collect spectral data which can supply information on whet growth, development and detecting water stress effects.

Introduction
Water is an important input for crop productivity which varies from place to place. Crops suffer from water deficit and its yields are reduced. Water availability will remain an important factor in years t come, thus an assessment of crop response to water availability under field conditions and knowledge of it becomes an essential requirements. From remote sensing devices operated from airborne system or satellites, it may be possible to make a quick assessment of vast areas. However, in the present development of remote sensing technology in India, an understanding of plant response to water deficit which can be recorded by remote sensing devices is a basic requirement in this direction hand held radio meters can be used to develop fundamental data on ration between radiometric data and crop growth parameters. Keeping in view the above considerations, the present field experiments were conducted on wheat.

Materials and Methods
Field experiments were carried out on wheat (Triticum aestivum L. var. WL 711) at Punjab Agricultural University, Ludhiana during Rabi seasons of 1987-88 and 1988-89. The field was irrigated and ploughed and wheat was sown and harvested in second week of November and April respectively in both the years. The experiment consisted of seven irrigation treatments, viz. no irrigation; one irrigation at crown root initiation (CRI); two irrigations at CRI + tillering (T)/Flowering (F); three irrigations at CRI + F + milking (M); four irrigations (as per recommended practice of 21, 62, 100, 125 days after sowing); five irrigations at CRI + T + booting + F + M; and irrigations based on cumulative pan evaporation of 75mm. In the last treatments total 3 irrigations were given which were equivalent to CRI + F + M stages.

The indigenously developed hand held radio meter was used to measure the radiance in situ from these plots on 8 measurement dates. The radiance was measured in red (625-689 nm with a peak at 665nm) and infrared (760-897nm with a peak at 830 nm) normal to the ground surface at a height of approximately 1.5 m above the crop canopy. Four to six spectral measurements per plot were averaged to account for the spatial variability of each plot. Immediately after each spectral measurement on a given plot, solar irradiance was measured fro a BaSO₄ panel. All these measurements were normalized with the irradiance obtained by the BaSO₄ panel. To describe growth patter radiance ratio ® of IR/Red ad normalized difference (ND) i.e. (IR-red/IR+red) were used on spectral parameters.

Results and Discussion

Spectral Response vis-à-vis Plant Growth Parameters
A simple linear (Model I) ad quadratic (Model II) regression analysis was used to related R and ND with growth and agronomic variables during 1987-88 and 1988-89. The relationship of parameters Viz, plant height, total fresh biomass, total dry biomass with R and ND was improved when model II was used (table 1). However, the plant height and the leaf-area index were linearly correlated with spectral indices (Fig. 1c). It was observed that there was general increase in R and ND with an increase in agronomic values during the vegetative growth but the trend was reserved in the later crop states. It appears that R and ND are function of the agronomic growth parameters.

Table 1: Correlation coefficients resulting from regression analysis during 1987-88.

Agronomic variable	Spectral parameter
	Infrared: red		ND
	Model I	Model II	Model I	Model II
Plant height	0.13	0.94	0.26	0.91
Leaf-area index	0.59	0.62	0.41	0.48
Plant water content	0.23	0.27	0.24	0.26
Total fresh biomass	0.14	0.63	0.28	0.73
Total dry matter	0.23	0.58	0.34	0.55

Crop development
The growth and development of crops can also be represented by the temporal variation of spectral parameters over the crop cycle. Radiance ratio and ND increase in the beginning with increasing green biomass, becomes maximum and then decreases due to senescence. The IR/red ratio and ND was always higher for an irrigated crop compared with unirrigated one (Fig. 1a&b). The difference in these spectral parameters for irrigated and water stressed whet were more during 75 to 102 days after sowing. Ti was found that at 85 to 91 DAS unrrigated and normal (4 irrigations) irrigated wheat differ significantly from one another with respect to radiance ratio and ND in both the years. Thus, spectral discriminability in irrigated and stressed plants is enhanced during this period. Similar result have been found by Kamat et. al. (1985).

Canopy Temperature vis-à-vis Consumptive use and grain yield
Canopy air temperature differences (D)T were proposed by Wiegand and Namken (1966) to be indicative of water stress. It is know that whenever a crop has sufficient moisture it will transpire freely and its temperature will be lower than that of the ambient air due to reduced transpiration rate. The canopy temperature in the morning and afternoon hours was 0.8-3°C and 3-4°C higher under unirrigated plots as compared to irrigated plots (table 2). This shows that canopy-air temperature can also indicates the water stress in crops.

Table 2: Canopy temperature (°C) and canopy air temperature differences (DT°C) acquired during 987-88 wheat growing season.

Days after sowing	Time of day	Canopy temperature		DT
Days after sowing	Time of day	Urrigated	Irrigated	Unirrigated	Irrigated
85	1030	1.8	18.0	2.5	4.6
	1420	23.2	19.9	3.0	6.2
102	1120	22.3	19.3	1.1	3.4
	1420	23.6	20.0	2.6	5.3
124	1200	21.7	21.5	2.9	2.7
	1430	22.4	22.0	6.0	6.1
131	1030	26.7	23.6	2.5	2.1
	1430	29.8	25.3	3.1	3.2

The linear regression analysis at 85, 102, 125 and 131 days after sowing wheat during 1987-88 showed that consumptive use of water by crop and grain yield were linearly correlated with DT especially at 85 DAS. The highest r value of 0.85 between consumptive used and DT was obtained at 85 DAS. Similarly, at 85 DAS the r values of 0.81 and 0.71 were obtained between yield and DT measured 1030 and 1420 hours. This shows that canopy air temperature differences measured during maximum crop cover stage may help to predict water stress and grain yield.

Wheat Yield Estimates
The leaf area index is highly correlated to the spectral parameters especially during maximum crop growth stage (table 3). The consumptive use (CU) of water is also related to LAI during this period, which in turn is related to spectral indices. The IR/red and ND were integrated over different periods during the two growing seasons to study the feasibility of remote-sensing application of the leaf area duration concept. The integrated values were then correlated with the yield (table 4). The regression relation between the grain yield and the integrated values of vegetation indices over the three periods show that the period 75-104 days (the plateau of the growth the curve), corresponding to maximum presence of green crop canopy, sowed the highest value of correlation coefficient with yield (Fig. 1d). Similar results have been found.

Table 3. Linear regression (les square) correlation (r) between different parameters.

Dependent V/S Independent Parameters	Days after sowing
	1987-88			1988-89
	75	85	104	76	91	102
R V.S LAI ND V/S LAI CU V/S LAI CU V/S R CU V/S ND Yield V/S LAI Yield V/S R Yield V/S ND	0.96 0.96 0.85 0.91 0.74 0.77 0.89 0.46	0.91 0.92 0.74 0.63 0.60 0.88 0.73 0.80	0.84 0.85 0.91 0.73 0.60 0.91 0.76 0.84	0.54 0.78 0.84 0.27 0.24 0.95 0.80 0.33	0.92 0.97 0.89 0.96 0.73 0.91 0.94 0.50	0.59 0.82 0.92 0.67 0.21 0.89 0.74 0.47

by tucker et. al (9180). The high correlation between the spectral parameters determined at any time during the maximum crop coverage and grain yield clearly establish the potential and possibilities of remote sensing in predicting grain yields.

Table 4. Regression derived estimates for 3 time segments using spectral indices to predict the grain yield.

Time period (days)	Spectral parameters
	ND			Infrared : red
	Intercept	Slope	r	Intercept	Slope	r
1987-88
52-75	84.55	-75.84	-0.40	-65.91	24.47	0.75
75-104	-120.07	229.94	0.99	-43.46	13.75	0.93
104-124	-25.85	138.97	0.81	-1.89	9.56	0.37
1988-89
60-76	110-55	-102.36	-40	-36.20	15.53	0.98
76-102	45-41	130.31	0.78	2.88	6.54
102-117	25-73	38.08	0.52	7.33	7.33	0.74

Conclusions

The agronomic variables are related to spectral parameters. Hence, spectral data can provide information about plant growth and development.
Spectral indices can be used for detecting water stress in wheat.
Canopy-air temperature difference can provide information regarding water stress in crop.
Spectral parameters are highly correlated to physiological variables canopy-air temperature differences, consumptive use of water by crop and final economic biomass production.

References

Kamat, D.S., Gopalan, A.K.S., Shashikumar, M.N. 1985 Assessment of water stress effects on crops, int., J. Remote Sensing, 6:577.
Tukcer, C.J., Holben, B.N., Elgin, J.H., and McMurtey, J.E. 1980, relationship of spectral data to grain yield variations. Photogram, Engng. Remote sensing, 46;657.
Wiegand, C.L. and Namken, L.N., 1966 Influence of plant moisture and air temperature on cotton leaf temperatures. Agron. J., 58:582.