Published in Agron. J. 96:1282-1287 (2004).
© American Society of Agronomy
677 S. Segoe Rd., Madison, WI 53711 USA
Agroclimatology
Evaluation of Two Temperature Stress Indices to Estimate Grain Sorghum Yield and Evapotranspiration
Ayodele Ebenezer Ajayia,* and
Ayorinde A. Olufayob
a Dep. of Ecol. and Nat. Resour. Manage., Cent. for Dev. Res. (ZEF), Univ. of Bonn, Walter Flex Staße 3, 53113 Bonn, Germany
b Dep. of Agric. Eng., Federal Univ. of Technol., P.M.B. 704, Akure, Nigeria
* Corresponding author (ayodele_ajayi{at}yahoo.com)
Received for publication March 20, 2003.
 |
ABSTRACT
|
|---|
This study was performed to examine the relationship between canopy temperature (Tc)–based stress indices and grain yield and evapotranspiration (ET) of sorghum [Sorghum bicolor (L.) Moench]. The crop was subjected to 14 differentially irrigated treatments of which two were controls maintained at well-watered and dry conditions during three consecutive years. Soil water content and Tc were measured, and relationships between Tc–based stress indices [stress degree day (SDD), temperature stress day (TSD), and crop water stressed index (CWSI)] and yield as well as ET were examined. The Tc–air temperature (Ta) difference varied from –2 to +8°C in the stressed treatments and maintained a negative value for most of the time in the well-watered treatment. The relationship between Tc – Ta and vapor pressure deficit, commonly referred to as baseline in the determination of CWSI, was examined on function of wind speed and global solar radiation. Although observations showed that Tc can be influenced by climatic condition, this study confirmed that it can serve as a useful indicator of water stress in the case of sorghum. High correlation found between Tc–based stress indices TSD, SDD, and CWSI and ET as well as grain yield suggest the possibility of using these relationships for predictive purposes.
Abbreviations: CWSI, crop water stressed index • DAS, days after sowing • ET, evapotranspiration • SDD, stress degree day • Ta, air temperature • Tc, canopy temperature • TSD, temperature stress day
|
INTRODUCTION
|
|---|
WATER DEFICIT affects plant growth, metabolism, and yield (Lewis et al., 1974; Baldy et al., 1993). Accurate and timely determination of water deficit effect on yield reduction is of great importance. Many research objectives have therefore been focused on finding easy and efficient methods of predicting crop yield. Although different methods of yield prediction based on soil and plant measurements have been proposed, remote sensing of Tc provides an enormous advantage (Idso et al., 1980). The simplicity, rapidity, and the nondestructive nature of infrared thermometry measurement, and the fact that sampling is easy, made it applicable to disease and insect damage assessment (Nicolas et al., 1991), plant water stress assessment (Jackson et al., 1981), irrigation scheduling (Clawson and Blad, 1982), and yield prediction of water-stressed crops (Idso et al., 1977).
Two Tc–based stress indices, the SDD and TSD, are simple to estimate and have been shown to relate well to yield (Gardner et al., 1981a, 1981b; Diaz et al., 1983). The SDD concept was first developed by Idso et al. (1977) to estimate ET and yields for a variety of irrigated crops. It is defined as:
 | [1] |
where Tc and Ta are midday canopy and air temperatures (°C) on day i (Idso et al., 1977).
The TSD index is estimated as the difference between Tc (°C) and canopy temperature (°C) of a well-watered crop (Tcw):
 | [2] |
A well-watered plot of the same crop (i.e., at ET maximum) is thus required to estimate this index. It is important to note that these indices require few measurements unlike other temperature-based indices such as the CWSI, which may require solar radiation, vapor pressure, and wind measurements in addition to crop temperatures.
The CWSI has been related to leaf water potential and yield of several crops and has also been used to schedule irrigations (Garrot et al., 1994; Sepaskhah and Kashefipour, 1994; Irmak et al., 2000). There are two well-known methods of calculating this index: an analytical approach proposed by Jackson et al. (1981) and an empirical approach described by Idso et al. (1981). Katerji et al. (1988) used the following equation to calculate CWSI:
 | [3] |
where Tc is the measured canopy temperature of a stressed plot, Tcmin is the measured canopy temperature of a fully irrigated plot, ET is evapotranspiration, and PET is potential evapotranspiration. Tcmax is a fictitious value estimated by supposing that the net radiation flux is suddenly converted to sensible heat flux (Katerji et al., 1988):
 | [4] |
where ra is the aerodynamic resistance (s m–1), Rn the net radiation (W m–2), 
the density of air (kg m–3), and Cp the specific heat of air at constant pressure (J kg–1 °C–1). In practice, Rn and Ta can be measured directly, and ra can be calculated from the values of wind speeds at 2 m and crop height using a formula proposed by Itier and Katerji (1983).
Gardner et al., (1981a), who correlated the yields of two hybrids of sorghum grown under several irrigation regimes with TSD, used a single planting date. Idso et al. (1980) showed that planting dates and climatic variability has an effect on the relationship between yield and 
SDD and therefore proposed the introduction of solar radiation. Diaz et al. (1983), who performed their experiments on wheat (Triticum aestivum L.), subsequently introduced incoming radiation at the vegetative stage to account for differences in planting dates. These studies were performed in the arid regions of USA under climatic conditions very different from that of France. Although infrared thermometry seems promising in southern France, particularly for irrigation scheduling of sorghum (Olufayo et al., 1993b), it would be necessary to establish its limits before going into practical application. This study was aimed at examining the relationship between Tc–based indices requiring only temperature measurements (i.e., TSD and SDD) and ET and grain yield. Several planting dates of grain sorghum were subjected to drought stress at various stages of growth to reveal any time variation or time trends in these relationships.
|
MATERIALS AND METHODS
|
|---|
Grain sorghum cultivar Argence was grown at the French Institute of Agricultural and Environmental Engineering Research (CEMAGREF) experimental station Lavalette in Montpellier (France) (43°40'N, 3°50'E). The experimental site covers about 2 ha, and soil was classified as a deep loamy clay Calciustolls. The crops were sown on 3 May 1990, 16 May 1991, and 7 May 1992 at 0.5-m spacing between rows. Fertilizers were applied before planting at the rate of 150 kg ha–1 N and 100 kg ha–1 P2O5. Plant populations varied between 200000 and 300000 plants per hectare.
There were 14 treatment plots altogether during the 3 yr of study (Table 1). They were maintained at different levels of irrigation based on crop developmental growth periods defined in Table 2 using the Vanderlip and Reeves (1972) phenological scale. Each treatment plot was about 24 by 48 m in size. These large plots were preferred to minimize experimental errors due to interference between neighboring microclimates. A previous study in this field showed that the soil is homogeneous (Olufayo, 1994). Replicated readings were taken in each plot. An average of 10 readings were recorded. Treatments were not replicated in space during each growing season, but some were replicated in time across the 3 yr. In particular, treatments T1, T3, and T11 were replicated across years.
Soil moisture content was monitored twice a week using a neutron probe at intervals of 0.10 m until about 3-m soil depth. An access tube was installed at the center of each experimental plot. A series of tensiometers at 0.10-, 0.20-, 0.30-, 0.40-, 0.60-, 0.90-, 1.00-, 1.20-, 1.40-, 1.60-, and 1.80-m soil depth were also installed at the center of each treatment plot. Evapotranspiration for each treatment was estimated using the water balance technique between soil surface and the plane of zero flux (Vachaud et al., 1978).
Canopy temperature readings were taken using a hand-held infrared thermometer (Tasco THI 300).1 The spectral band-pass of the instrument was 6 to 12 µm with a resolution of 0.1°C and a field of view of 10°. Other infrared thermometers used were the Raytek (model Raynger PM3), and the Everest 110 and 510B. Measurements were taken at an oblique angle to avoid sensing soil temperatures and also by viewing sunlit leaves. The data-averaging feature of the infrared thermometer was employed to reduce variability in Tc. Randomized readings from different areas within the effective area of a plot were considered as repetitions, and at least 10 readings were taken and an arithmetic average computed. Canopy temperatures were monitored on each bright, sunny day (which occurs more than 80% of time during summer in southern France) between 1130 and 1330 h solar time. The instruments' calibrations were checked in a laboratory before the start of the experiments, and systematic recalibrations were performed in the field during experiments (Olufayo et al., 1993a). Canopy temperature measurements began when leaf area index was about 2 (when the ground was well covered so as to avoid taking measurements of the soil's surface) and ended at about a leaf area index of 5 (at physiological maturity). A standard meteorological station (CIMEL 411) was situated 120 m from the experimental site. It furnished hourly and/or three hourly as well daily averages of Ta, relative humidity, wind speed, and solar radiation, which are necessary for the estimation of potential ET using the Penman formula (Penman, 1948). In addition, a portable electronic thermohygrometer (CORECI) was used to measure Ta and relative humidity at 10 cm above the crop canopy within the experimental plots. Sunshades were utilized to minimize direct solar incidence on the sensors.
Each plot was sprinkler-irrigated using a 12- by 12-m triangle arrangement at an average of once a week depending on the treatments and the soil moisture content readings. Irrigation water was applied at night when wind speed was low (<2 m s–1) to ensure even distribution. The required irrigation amounts per application were based on water balance model estimates of the prevailing root zone water depletion (see Table 1). Irrigation amounts allowed about a 13 mm of unfilled root zone storage capacity for rainfall that might occur soon after an irrigation.
Visual phenologic and crop height observations were made at weekly intervals to document growth stage development. Yields were standardized to 15% dry basis moisture. Panicles were harvested by hand and shelled mechanically.
|
RESULTS AND DISCUSSION
|
|---|
Climate
The growing season for sorghum extends from May to September. As shown in Fig. 1 for 1991, the climate is characterized by warm, mostly clear and dry sky conditions with relatively high evaporative demand (4 to 7 mm/d) during the early growth stages through dough stage. During hard dough through physiological maturity, it is characterized by cooler days and wet conditions with relatively low evaporative demand. However, substantial variation was observed from year to year. For example, total precipitation for the months of May to August was 100 mm in 1990, 130 mm in 1991, and 280 mm in 1992. The summer of 1992 was particularly wet, and the annual precipitation exceeded the 30-yr average value.
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 1. Mean air temperature (Ta), solar radiation (Sr), and average daily potential evapotranspiration (PET) at 10-d intervals during the growing season.
|
|
Seasonal Evolution of Canopy–Air Temperature Difference
The seasonal trends of Tc – Ta of stressed and well-watered treatment were different. The Tc – Ta varied from –2 to +8°C in the stressed treatment and maintained a negative value for most of the time in the well-watered treatment. The Tc – Ta fluctuated but increased progressively from 48 d after sowing (DAS) to 80 DAS in a stressed plot when there was no rainfall or irrigation. The fluctuations were due to daily variations of climatic parameters. The readings of Tc – Ta were affected by rainfall. It was observed that variations within the plot were greater during and after rainfall in Growth Period III (Gardner et al., 1981a).
A CWSI, which was defined by Jackson et al. (1981), requires the establishment of a "non-water-stressed baseline." The relationship between Tc – Ta and vapor pressure deficit requires Tc of a well-watered treatment. Two series of Ta and vapor pressure deficit data were used. One concerned measurements taken at 10 cm above the canopy, and the other came from the nearby meteorological station. In all cases, even though coefficient of determination (r2) was greater than 0.70, a considerable degree of scatter was observed.
The following is the regression equation obtained for a non-water-stressed baseline using all 3-yr data (1990–1992):
 | [5] |
where VPD is vapor pressure deficit. To explain the measure of scatter observed, the combined relationship was therefore examined as a function of global radiation and wind speed. Analysis showed that the dispersion was only slightly accounted for by solar radiation, probably because most of the data falls within a narrow range of high radiation values. It was noted that a higher correlation was obtained for low wind speeds. This confirmed the necessity to take into account other climatic variables such as solar radiation and particularly wind speed when using this approach (O'Toole and Hatfield, 1983). Details of this analysis were reported by Olufayo et al. (1993b). However, since 1992 was considered a wet year having the lowest climatic deficit (see Table 2) the following regression equation obtained for this year was therefore used in calculating this index:
 | [6] |
A typical seasonal trend of this index is shown in Fig. 2. The index increased progressively between 54 and 80 DAS, corresponding to a period of when there was no irrigation or rainfall. During this period, the percentage of extractable water within the root zone decreased from 35 to 5%. The daily fluctuations are partly due to the degree of scatter observed in the lower baseline.
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 2. Seasonal trends of canopy–air temperature difference (Tc – Ta) in stressed and well-watered treatments.
|
|
Yield Response Relationships
A t test analysis was performed to compare average grain yield from an entire plot with that obtained near the neutron access tube. Both paired and independent tests showed that the two sets of data were not significantly different at 
= 0.05. We therefore used only the grain yield data obtained near the neutron access tube in establishing yield vs. seasonal ET (emergence to physiological maturity) relationships. The regression equations obtained are shown in Table 3. An F test described by Scherrer (1984) for comparing several slopes and intercepts was used to investigate the possibility of time variation in the yield–ET regression equations. The slopes were equivalent, but the intercepts were found to be significantly different. However, on a relative basis, the F test showed that the slopes as well as the intercepts of regressions equations were equivalent at the 2.5% level (see Table 4). The single relationship obtained for the combined 1990 and 1991 data gives:
 | [7] |
whereY is yield, Ym is the yield at maximum ET, and Sy/x is the standard error of estimate y on x.
Average grain yield per plot was related to Tc–based indices. In calculating the summations of TSD and SDD, only positive values were considered (Jackson et al., 1977). The period of summation for each year covers 50 d and lies between the end of Growth Period II and IV (see Table 2). The resulting regression equations are presented in Table 3. Both TSD and SDD were well correlated with yield (r2 > 0.87). Grain yield decreased with increasing TSD. Similar relationships were obtained when relative yield was used (Table 4). Unlike the yield–ET relationship, there was no effect of planting date on the regression coefficients (Y–TSD relationship) as indicated by the F test analysis at the = 0.05 level. This implies that if yield under an unstressed condition is known, it is possible to estimate grain yield as a function of TSD values. This agrees with results reported by Gardner et al. (1981a) for two moisture-stressed sorghum hybrids. Although yield was also highly correlated with SDD, statistical results of F test on the effect of planting dates showed that the slopes of the two equations were similar and the intercepts were significantly different (Table 3). Contrary to the results obtained by Diaz et al. (1983), the introduction of solar radiation did not account for the effect of planting dates. However, when relative yield was examined, the statistical analysis showed that the two regression equations for the two planting dates were not significantly different at = 0.05 level (see Table 4). It therefore appears that in our environmental conditions, climatic factors other than solar radiation influence this relationship. When the three treatments in 1992 were included in the analysis, the combined data resulted in the following relationship:
 | [8] |
It is worth noting that the inclusion of the 1992 data did not affect the high correlation coefficient.
Relationship between Seasonal Evapotranspiration and Stress Indices
The results showed that ET was linearly and inversely related to both TSD (see Eq. [9] and [10]) and SDD. Many works have shown similar results for various crops in many locations (Idso et al., 1977; Walker and Hatfield, 1979; Diaz et al., 1983). A slightly better correlation was obtained with SDD (r2 = 0.95 to 0.98) than with TSD (r2 = 0.74 to 0.93) in this study. Statistical analysis showed that the ET–TSD relationship was influenced by planting date unlike the Y–TSD relationship. For the 1991 and 1990 combined data, the following linear equation was obtained:
 | [9] |
The correlation becomes very poor when 1992 data were included:
 | [10] |
Although the precise reason for this is not known, it appeared that the climatic diversity during experiments was partly responsible. As mentioned above, 1992 was described as being particularly wet. This would obviously modify the rate of water use by the crop. In addition, ET would be estimated less accurately in periods of excessive rainfall. On a relative basis and using all data (i.e., including 1992), the correlation was:
 | [11] |
where ETmax is seasonal ET in a wet treatment (i.e., T1) for each year. Similarly, the inclusion of 1992 data reduced the coefficient of determination (r2) to 0.48. For relative ET vs. TSD, the correlation coefficient was 0.94.
The following linear relationships were obtained between the CWSI calculated by the three methods and (1 – ET/PET).
 | [12] |
 | [13] |
 | [14] |
where CWSI – k is the method by Katerji et al. (1988), CWSI – t is the analytical or theoretical approach (Jackson et al., 1981), and CWSI – e is the empirical method (Idso et al., 1981). Higher correlation was obtained using the CWSI calculated by the method described by Katerji et al. (1988), which did not produce negative values. This is partly due to the fact that the lower limit is a measured Tc of a full-irrigated plot whereas in the case of empirical and theoretical approach, both lower and upper limits of transpiration are fictitious (Jackson et al., 1988).
|
CONCLUSION
|
|---|
This study showed that Tc is a useful indicator of water status in the case of sorghum under the Mediterranean conditions prevailing at Montpellier, France.
Significant linear relationships were established between Tc–based indices and yield as well as ET. This is of great importance since it suggests the possibility of estimating yield and ET from Tc data. This would be of great benefit for establishing optimum irrigation strategies on a local scale. It is also important on a regional scale through the use of satellite data (e.g., NOAA-6, TISOS-N, HCMM, and METEOSAT) (Seguin and Itier, 1983).
Although CWSI (of Katerji et al., 1988) gave good results when related to (1 – ET/PET), it requires, like TSD, a plot to be maintained in well-watered condition. From our experience, however, it is difficult to ensure the same level of water regime in full-irrigated plots for the 3 yr in spite of the fact that water was constantly supplied weekly. Hence, SDD seem preferable since it does not incur additional difficulties, cost, and errors, which are related to the maintenance of a well-watered plot. In estimating SDD, however, either Ta above the canopy or from a nearby meteorological station could be used, but this must be specified if results from different locations are to be compared.
Although this experiment was performed in France, the experimental results obtained on sorghum would likely be useful in other parts of the world and especially in Africa, the crop's origin.
|
NOTES
|
|---|
1 Mention of a trade name does not imply endorsement of the product or company. 
|
REFERENCES
|
|---|
- Baldy, C., P. Ruelle, A. Fernandes, J.M. Konate, and A. Olufayo. 1993. Résistance à la sécheresse du sorgho-grain en climat méditerranéen et gestion optimale d'une quantité d'eau limitée. Sécheresse 2:85–93.
- Clawson, K.L., and B. Blad. 1982. Infrared thermometry for scheduling irrigation of corn. Agron. J. 74:311–316.[Abstract/Free Full Text]
- Diaz, R.A., A.D. Matthias, and R.J. Hanks. 1983. Evapotranspiration and yield estimation of spring wheat from canopy temperature. Agron. J. 75:805–810.[Abstract/Free Full Text]
- Gardner, B.L., B.L. Blad, D.P. Garrity, and D.G. Watts. 1981a. Relationships between crop temperature, grain yield, evapotranspiration and phenological development in two hybrids of moisture stressed sorghum. Irrig. Sci. 2:213–224.
- Gardner, B.L., B.L. Blad, R.E. Maurer, and D.G. Watts. 1981b. Relationships between crop temperature and the physiological and phenological development of differentially irrigated corn. Agron. J. 73:743–747.[Abstract/Free Full Text]
- Garrot, D.J., Jr., M.J. Ottman, D.D. Fangmeier, and S.H. Husman. 1994. Quantifying wheat water stress with the crop water stress index to schedule irrigations. Agron. J. 86:195–199.[Abstract/Free Full Text]
- Idso, S.B., R.D. Jackson, P.J. Pinter, Jr., R.J. Reginato, and J.L. Hatfield. 1981. Normalizing the stress-degree-day parameter for environmental variability. Agric. Meteorol. 24:45–55.
- Idso, S.B., R.D. Jackson, and R.J. Reginato. 1977. Remote sensing of crop yields. Science 196:19–25.[Abstract/Free Full Text]
- Idso, S.B., R.J. Reginato, J.L. Hatfield, G.K. Walker, R.D. Jackson, and P.J. Pinter, Jr. 1980. A generalization of the stress-degree day concept of yield prediction to accommodate a diversity of crops. Agric. Meteorol. 21:205–211.
- Irmak, S., D.Z. Hamman, and R. Bastug. 2000. Determination of crop water stress index for irrigation timing and yield estimation of corn. Agron. J. 92:1221–1227.[Abstract/Free Full Text]
- Itier, B., and N. Katerji. 1983. Une expression simplifiée de la résistance aérodynamique d'un couvert végétal. Acta Oecol., Oecol. Gen. 3:215–227.
- Jackson, R.D., S.B. Idso, R.J. Reginato, and P.J. Pinter, Jr. 1981. Canopy temperature as a crop water stress indicator. Water Resour. Res. 17:1133–1138.
- Jackson, R.D., W.P. Kustas, and B.J. Choudhury. 1988. A reexamination of the crop water stress index. Irrig. Sci. 9:309–317.
- Jackson, R.D., R.J. Reginato, and S.B. Idso. 1977. Wheat canopy temperature: A practical tool for evaluating water requirements. Water Resour. Res. 13:651–656.
- Katerji, N., B. Itier, and I. Ferreira. 1988. Etude de quelques critères indicateurs de l'état hydrique d'une culture de tomate en région semi-aride. Agronomie (Paris) 8:425–433.
- Lewis, R.B., E.A. Hiler, and W.R. Jordan. 1974. Susceptibility of grain sorghum to water deficit at three growth stages. Agron. J. 66:589–591.[Abstract/Free Full Text]
- Nicolas, H., R. Rivoal, J. Duchesne, and Z. Lili. 1991. Detection of Heterodera avenae infestations on winter wheat by radiothermometry. Rev. Nématol. 14:285–290.
- Olufayo, A.A. 1994. La caractérisation des effets du stress hydrique du sorgho grain à l'aide des indicateurs bioclimatique. Ph.D. thesis. Ecole National Supérieure Agronomique de Rennes, Rennes, France.
- Olufayo, A., C. Baldy, and P. Ruelle. 1993a. Effect of ambient conditions on calibration of hand held infrared radiothermometers. Agronomie (Paris) 13:835–843.
- Olufayo, A., C. Baldy, P. Ruelle, and J. Konate. 1993b. Diurnal course of canopy temperature and leaf water potential of sorghum (Sorghum bicolor (L.) Moench) under a Mediterranean climate. Agric. For. Meteorol. 64:223–236.
- O'Toole, J.C., and J.L. Hatfield. 1983. Effet of wind on the crop water stress index derived by infrared thermometry. Agron. J. 75:811–817.[Abstract/Free Full Text]
- Penman, H.L. 1948. Natural evaporation from open water, bare soil and grass. Proc. R. Soc. London, Ser. A 193:120–146.
- Scherrer, B. 1984. Biostatistics. Gaëtan Morin Editeur, Chicoutim, QC, Canada.
- Seguin, B., and B. Itier. 1983. Using midday surface temperature to estimate daily evaporation from satellite thermal IR data. Int. J. Remote Sens. 4:371–383.
- Sepaskhah, A.R., and S.M. Kashefipour. 1994. Relationship between leaf water potential, CWSI, yield and fruit quality of sweet lime under drip irrigation. Agric. Water Manage. 25:13–22.
- Vachaud, G., C. Dancette, S. Sonko, and J.L. Thony. 1978. Méthode de caractérisation hydrodynamique "in situ" d'un sol non saturé: Application à deux types de sols au Sénégal en vue de la détermination des termes du bilan hydrique. Ann. Agron. 29:1–36.
- Vanderlip, R.L., and H.E. Reeves. 1972. Growth stages of sorghum (Sorghum bicolor (L.) Moench.). Agron. J. 64:13–16.[Abstract/Free Full Text]
- Walker, G.K., and J.L. Hatfield. 1979. Test of the stress-degree-day concept using multiple planting dates of red kidney beans. Agron. J. 71:967–971.[Abstract/Free Full Text]
TOP
ABSTRACT
INTRODUCTION
