Published in Agron J 99:1338-1344 (2007)
DOI: 10.2134/agronj2007.0149
© 2007 American Society of Agronomy
677 S. Segoe Rd., Madison, WI 53711 USA
Modeling
Modeling the Effects of Water Temperature on Rice Growth and Yield under a Cool Climate
II. Model Application
Hiroyuki Shimonoa,*,
Toshihiro Hasegawab,
Tsuneo Kuwagatab and
Kazuto Iwamac
a Faculty of Agriculture, Iwate University, 3-18-8 Ueda, Iwate, 020-8550, Japan
b Department of Global Resources, National Institute for Agro-Environmental Sciences, 3-1-1 Kannondai, Tsukuba, 305-8604, Japan
c Graduate School of Agriculture, Hokkaido University, N9, W9, Kita-ku, Sapporo, 060-8589, Japan
* Corresponding author (shimn{at}iwate-u.ac.jp)
 |
ABSTRACT
|
|---|
Impact of water temperature (Tw) on rice growth and yield in Hokkaido, Japan, one of the coolest rice producing areas in the world, were quantitatively evaluated using the newly developed growth model that can simulate the effects of Tw independently from the effects of air temperature (Ta). Using this model, first we evaluated the benefits of having Tw warmer than Ta for rice yield. Without this difference, simulated rice yield was reduced by almost half. Second, the model also highlighted the causes of differences among sites in rice productivity; higher productivity was attributed to higher solar radiation (RD), higher wind speed (WS), and higher Tw. Finally, under a scenario of future global warming, the model estimated that a 3°C Ta increase above the current level increased Tw by 1°C and increased yield by 6%. However, adding the effects of changes in other factors (RD and WS) increased Tw by 0 to 2°C and changed yield by –30% to +41%. Our results demonstrate that Tw must be considered to understand growth and yield responses of rice to climate change, especially in cool regions.
Abbreviations: AH, atmospheric humidity • LAI, leaf area index • RD, solar radiation • RMSE, root mean square deviation • Ta, air temperature • Tw, water temperature
Received for publication April 24, 2007.
Modeling the Effects of Water Temperature on Rice Growth and Yield under a Cool Climate
II. Model Application
Hiroyuki Shimonoa,*,
Toshihiro Hasegawab,
Tsuneo Kuwagatab and
Kazuto Iwamac
a Faculty of Agriculture, Iwate University, 3-18-8 Ueda, Iwate, 020-8550, Japan
b Department of Global Resources, National Institute for Agro-Environmental Sciences, 3-1-1 Kannondai, Tsukuba, 305-8604, Japan
c Graduate School of Agriculture, Hokkaido University, N9, W9, Kita-ku, Sapporo, 060-8589, Japan
* Corresponding author (shimn{at}iwate-u.ac.jp)
Received for publication April 24, 2007.
Impact of water temperature (Tw) on rice growth and yield in Hokkaido, Japan, one of the coolest rice producing areas in the world, were quantitatively evaluated using the newly developed growth model that can simulate the effects of Tw independently from the effects of air temperature (Ta). Using this model, first we evaluated the benefits of having Tw warmer than Ta for rice yield. Without this difference, simulated rice yield was reduced by almost half. Second, the model also highlighted the causes of differences among sites in rice productivity; higher productivity was attributed to higher solar radiation (RD), higher wind speed (WS), and higher Tw. Finally, under a scenario of future global warming, the model estimated that a 3°C Ta increase above the current level increased Tw by 1°C and increased yield by 6%. However, adding the effects of changes in other factors (RD and WS) increased Tw by 0 to 2°C and changed yield by –30% to +41%. Our results demonstrate that Tw must be considered to understand growth and yield responses of rice to climate change, especially in cool regions.
Abbreviations: AH, atmospheric humidity • LAI, leaf area index • RD, solar radiation • RMSE, root mean square deviation • Ta, air temperature • Tw, water temperature
|
INTRODUCTION
|
|---|
RICE YIELDS IN HOKKAIDO, the northern-most island of Japan (Fig. 1
), vary largely among years and locations. Various factors are involved in the temporal and regional variations in grain yields, some parts of which can be due to Tw, in addition to Ta. It is known that Tw rather than Ta has strong impact on rice growth and yield, especially in a cool climate (Enomoto, 1936; Sakai, 1949; Satake et al., 1988), but there is no modeling study to analyze its role on rice yield. In the previous study, we developed a model that covered major growth processes affected by Tw, and can take into account the effects of Tw independently from Ta (Shimono et al., 2007). Using this model, we evaluated the impact of Tw on rice yield.
View larger version (49K):
[in this window]
[in a new window]
|
Fig. 1. Map of the sources of the data used for model development and analysis, and average grain yield of rice on Hokkaido, Japan, from 1985 to 1999. Data were obtained from the Annual Report of Crop Statistics for Hokkaido (municipalities version), which was provided by the Ministry of Agriculture, Forest and Fisheries of Japan.
|
|
One of the major problems in application of this model to wider areas is the limited availability of Tw data compared with Ta. Recently, Kuwagata and Hamasaki (2001) developed a Tw estimation model at the National Agricultural Research Center for Tohoku Region (Morioka, Japan, 39°42' N, 141°10' E), which predicted Tw from five climatic factors [Ta, RD, WS, and atmospheric humidity (AH)] and leaf area index (LAI) using the Penman's method. Because their model includes most of the factors affecting Tw, it may be used to estimate Tw in other areas, which may be combined with the growth model to evaluate rice growth and yield.
The objectives of this study were first to test the growth model combined with the Tw estimation model using the data collected at four locations in Hokkaido; where rice is widely cultivated (Fig. 1). Second, beneficial effects of warmer Tw than Ta on rice production were quantified at these locations. Third, factors that influence temporal and regional variations in growth and yield were analyzed by the model. Finally, the effect of future global warming on rice yield in Hokkaido was assessed with the model.
|
COMBINING THE GROWTH MODEL WITH A Tw PREDICTION MODEL
|
|---|
Tw Prediction Model
The model developed by Kuwagata and Hamasaki (2001) estimates daily average Tw from four climatic variables (daily average Ta [°C], WS [m s–1], AH [g kg–1], and daily RD [MJ m–2 d–1]) and LAI (m2 m–2). In their model, Tw is calculated at two steps. First, Tw on a bare flooded field (Tw0) is estimated using the Penman's method which was based on the heat balance. Then, the difference between Tw0 and Tw under canopy cover (
Tw) is taken into account. The relation between Tw and LAI is affected by WS and RD.
Database
Climatic data from 1990 to 1999 for the four study areas are summarized in Table 1. Because RD data were not available for Iwamizawa, data for Sapporo, the nearest weather station that measured RD (about 40 km away), were used. Similarly, RD and AH for Pippu and Ohno were obtained from Annual Reports of the Japan Meteorological Agency in Asahikawa and Hakodate, respectively.
Tw data for testing the Tw estimation model were from Hokkaido University for Sapporo and from the Hokkaido Prefectural Agricultural Experiment Stations for Pippu (Kamikawa Agricultural Experiment Station), Iwamizawa (Central Agricultural Experiment Station), and Ohno (Dohnan Agricultural Experiment Station). The Tw data in 1997 and 1998 were used for Sapporo, Iwamizawa, and Ohno, and those in 1998 and 1999 were used for Pippu. Since Tw in Ohno was measured only at 0900 h (Tw9am), the daily average Tw was estimated from Tw9am based on the relation between Tw9am and daily Tw at Hokkaido University (Tw = 1.08 Tw9am). The Tw was measured 5 cm below the water surface. The water level during the reproductive growth period (from PI to heading) was kept at 
20 cm during the cool summer year of 1993 at all locations, and 10 cm for other years.
For testing the growth model combined with the Tw prediction model, we collected field measurements for the grain yield, spikelet fertility, and heading date of ‘Kirara 397’ in Pippu, Iwamizawa, and Ohno (Fig. 1) that were in the Annual Reports of the Performance Tests for Recommendable Varieties conducted at Hokkaido Prefectural Agricultural Experiment Stations (Kamikawa, Central, and Dohnan Agricultural Experiment Stations) based on data collected from 1990 to 1999. In addition, data from experiments conducted in 1992 and 1993 (Ichikawa et al., 1995) and in 1995 and 1998 at Hokkaido University (Shimono et al., 2002 and 1995, unpublished data), which were not used for model development, were provided for testing the model. Data for Pippu in 1994 were not available.
Crop management practices at the four locations were similar and resemble those commonly used on Hokkaido at present. In summary, germinated seeds of rice were sown in plug trays in late April (three seeds per cell) and seedlings were raised under a polytunnel for about 1 mo. They were transplanted into a paddy field in late May. At Sapporo, the date of transplanting ranged from 23 to 28 May. Planting density was mostly 25 hills m–2, with the following exceptions: 23.3 hills m–2 in 1995 and 1998 at Sapporo, 30 hills m–2 in 1990, and 20 hills m–2 in 1997 and 1998 at Iwamizawa, and 20 hills m–2 in 1990, 1991, 1992, and 1993 at Ohno. Each hill had three plants. Fertilizers were applied and incorporated in the soil before flooding (N = 8.0 – 10.0 g m–2, P = 4.2 – 4.9 g m–2, and K = 5.7 – 8.3 g m–2).
All model parameters used in this part of the study are those given at Sapporo for the growth model and the Tw estimation model. No tuning of parameters was conducted. The model validation was conducted using the actual time of transplanting. Planting density and water depth were set to the reported values at each location.
Model Validation
The growth model combined with the Tw estimation model simulated Tw well (Fig. 2
), with a root mean square deviation (RMSD) of 1.2°C over a range of 14 to 26°C (r = 0.920, P < 0.001). It also simulated well heading date, with an RMSD of 4.4 d over a range of 66 to 91 d (r = 0.845, P < 0.001); spikelet fertility, with an RMSD of 7.8% over a range of 3 to 97% (r = 0.904, P < 0.001); and grain yield, with an RMSD of 78.2 g m–2 over a range of 19 to 677 g m–2 (r = 0.832, P < 0.001). Measured grain yields differed greatly among the four locations, being highest at Pippu (an average of 586 g m–2 for all years), followed by Iwamizawa (495 g m–2) and Sapporo (484 g m–2), and lowest at Ohno (441 g m–2). Estimated grain yields were the highest at Pippu (609 g m–2), followed by Iwamizawa (498 g m–2) and Sapporo (520 g m–2), and were lowest at Ohno (476 g m–2). The present model had an error of about 10% in yield prediction, but overall, successfully simulated regional and temporal differences in heading date, spikelet fertility, and grain yield.
View larger version (34K):
[in this window]
[in a new window]
|
Fig. 2. Relationship between measured and estimated water temperature (Tw, expressed as average for 5 d), heading date (HD, expressed as days after transplanting, DAT), spikelet fertility, and grain yield of rice grown at four locations (Hokkaido, Japan). Lines represent y = x, and the 10% intervals on either side of that line. *** P < 0.001.
|
|
|
MODEL APPLICATION
|
|---|
Water Blanket Effect on Growth and Yield on Hokkaido
Under cool climates, Tw is higher than Ta and can serve as a "water blanket" that protects the shoot base and young developing panicles from low temperatures. In the present study, we quantified this effect on growth and yield on Hokkaido by means of simulation trials conducted under two scenarios. In Scenario 1, Tw was estimated using a Tw estimation model based on heat balance (Tw > Ta), whereas in Scenario 2, Tw was assumed to be equal to Ta (i.e., no water blanket effect, Tw = Ta).
The grain yield under Scenario 2 was substantially lower than that under Scenario 1 at Sapporo (Fig. 3a
). The average yield decreased from 581 to 286 g m–2 when Tw was set equal to Ta, and the coefficient of variation increased from 16 to 59%. This difference can be mostly attributed to the differences in spikelet fertility and harvest index (Fig. 3b, 3c) that result from the differences in the two scenarios. Furthermore, a 26% reduction in total dry weight (from 1343 g m–2 under Scenario 2 to 990 g m–2 under Scenario 1) also had a detrimental effect on grain yield. In addition, a substantial delay in heading date of as much as 22 d (the average value for the 10-yr period of data collection) was estimated under Scenario 2, and this could increase the risk of encountering unsuitably cool temperatures during the grain-filling period. The difference between estimated Tw and observed Ta from transplanting to heading was estimated to be 3.8°C (10-yr average).
View larger version (19K):
[in this window]
[in a new window]
|
Fig. 3. Simulated effects of assuming that Tw was warmer than Ta (the so-called "water blanket" effect)(Tw = Ta) on grain yield, harvest index, and total dry weight of the rice plant at Sapporo from 1990 to 1999. Scenario 1, normal condition (Tw > Ta); Scenario 2, Tw = Ta. Simulation conditions: transplanting on 20 May; planting density of 25 hills m–2; and a water depth during reproductive growth of 10 cm.
|
|
Even in a high-yielding area such as Pippu, the grain yields simulated under Scenario 2 (Tw = Ta) were severely reduced (by 71%) compared with the simulated results under scenario 1 (Tw > Ta), and exhibited increasing variation among years (Table 2). At Iwamizawa and Ohno, grain yields under Scenario 2 were about 50% lower than under Scenario 1. These results indicate that about half of the grain yield on Hokkaido can be attributed to the water blanket effect, and that practical rice cultivation in areas with a cool climate may be impossible without having flood water that remains warmer than the air. The results of our simulation agree well with the observation that upland (nonflooded) rice is not cultivated on Hokkaido, and that Aomori Prefecture (20 km south of Hokkaido) is the northern limit for cultivation of upland rice in Japan.
View this table:
[in this window]
[in a new window]
|
Table 2. Simulated total dry weight (TDW), harvest index (HI), and grain yield (GY) of cultivar ‘Kirara 397’ at Sapporo, Pippu, Iwamizawa, and Ohno under normal conditions [water temperature (Tw) > air temperature (Ta)] based on climate data from 1990 to 1999, and under scenarios in which Tw = Ta and Ta is 3°C above the normal condition (Ta + 3°C).
|
|
Factors Affecting Regional Differences in Growth and Yield on Hokkaido under Current Climates
The observed yield levels on Hokkaido differed greatly among sites (Fig. 1); for example, the yield at Pippu is higher than that at Sapporo, Iwamizawa, and Ohno. As shown in Table 1, this difference cannot be explained based on Ta per se during the growing season from June to September, which ranged from 17.8 to 18.8°C, with the lowest Ta at Pippu. Various other climatic factors such as RD and WS affect Tw, and can thus affect growth and yield. To determine the factors responsible for the higher yields at Pippu, we conducted a simulation analysis. Three scenarios were tested: (i) case Ta, in which the Ta value at Pippu was used for all locations; (ii) case RD, in which the RD at Pippu was used for all locations; and (iii) case WS, in which the WS at Pippu was used for all locations. The results were presented as the difference from the base level, which was the simulated level under normal environmental conditions at each location. Note that the model simulation of yield differences under normal conditions (Tw > Ta) showed that the yield was higher at Pippu than at any other location (by 4–23%), resulting from both a higher total dry weight and a higher harvest index (Table 2).
Figure 4
shows that changing RD to the level at Pippu increased grain yield (by 5–12%) at all three locations compared with the grain yield under the normal environmental conditions for those areas. This indicates that a high level of RD is one major reason for the high yield observed at Pippu. The effects of Ta and WS varied greatly among locations and parameters. At Iwamizawa, reducing WS to the level at Pippu increased grain yield by 8% due to increased LAI and dry weight, but case Ta only increased grain yield by 2%. At Ohno, the effect of reducing WS was smaller than that at Iwamizawa was small (4%), but case Ta apparently raised grain yield by 8%. A lower Ta during early growth at Ohno than at Pippu (Table 1) decreased LAI, and higher Ta during later growth stages (Table 1) shortened the growth duration, leading to lower total dry matter production and yield (Fig. 4).
View larger version (38K):
[in this window]
[in a new window]
|
Fig. 4. Simulated differences in grain yield and total dry weight at maturity of rice based on the assumption that the conditions at Pippu (Table 1) were used for the other three locations (Sapporo, Iwamizawa, and Ohno). Values represent the percentage change in each of these parameters under different scenarios compared with the results under normal environmental conditions for each location. (1) Case Ta: the air temperature at Pippu was used for all the other locations. (2) Case RD: the solar radiation at Pippu was used for all the other locations. (3) Case WS: the wind speed at Pippu was used for all the other locations. Simulation conditions: transplanting on 20 May; planting density of 25 hills m–2; water depth during the period of reproductive growth, 10 cm.
|
|
Previous studies showed that Ta during the whole crop cycle (May–September) and during the reproductive growth period (July–August) was the variable most frequently used to explain the regional differences in growth and yield (Hoshino and Okabe, 1960; Fujiwara et al., 1966; Hanyu, 1971). However, in the present study, RD and WS generally had a greater influence than Ta on the differences in grain yields between our four study locations. We showed that 35 to 76% of these differences could be attributed to differences in RD and that 4 to 52% of the differences could be attributed to WS. These results arise from the fact that RD affects not only canopy photosynthesis and dry matter production but also affects Tw; similarly, WS can influence grain yield by affecting Tw. A significant portion of the regional differences could therefore be associated with Tw, in addition to the direct effects of RD.
Effect of Future Global Warming on Growth and Yield on Hokkaido
By the end of this century, Ta is expected to increase by 3°C in average as a result of the predicted increases in concentrations of atmospheric CO2 and other greenhouse-effect gases (IPCC, 2002). However, several climatic factors associated with these global changes, such as RD and WS, may also change, although the magnitude and direction of the changes are difficult to predict using present climate models. To understand the risks posed by these uncertain changes, we simulated the effects of combined changes in RD (± 30%) and WS (± 30%) under global warming (a 3°C increase in Ta, referred to as the Ta + 3°C case) on Tw before the heading stage and thus, on grain yield.
Our model predicted that case Ta + 3°C would increase Tw before the heading stage by 1.0°C and increase yield by 6% at Sapporo in comparison with the results using mean weather data from 1990 to 1999 (Table 3). With an additional 30% radiation, Tw would increase by 2.0°C and yield would increase by 41%, whereas with 30% less radiation, Tw would not change and grain yield would decrease by 30%. Interestingly, WS also affected predicted future values, although the magnitude was smaller than that for RD. For example, with a 30% increase in WS, Tw would increase only by 0.5°C and yield would increase by 3%, but with a 30% decrease in WS, Tw would increase by 1.7°C and yield would increase by 9%. Thus, the projected change in Tw and grain yield with a 3°C increase in Ta varied greatly (from 0 to 2°C and from –30% to +41%, respectively) depending on the other climatic conditions.
View this table:
[in this window]
[in a new window]
|
Table 3. Effects of elevated air temperature (Ta increased by 3°C), accompanied by arbitrary changes in solar radiation (RD) and wind speed (WS), on simulated water temperature (Tw) before heading and on grain yield (GY) of cultivar ‘Kirara 397’ at Sapporo based on mean climatic conditions from 1990 to 1999.
|
|
Previous attempts to model the effect of global warming on rice yield was conducted using a Ta–based model (Rosenzweig and Parry, 1994; Horie et al., 1995; Kropff et al., 1995; Parry et al., 2004), but the present Tw–based model clearly demonstrated that the relationship between Ta and Tw could be affected by factors such as RD and WS, and that yield could be affected by changes in these parameters even for an identical Ta increase. On the basis of our results, predictions of yield responses to global warming, especially in regions that currently have cool climates, should account for the effects of Tw.
|
CONCLUSION
|
|---|
The present growth model based on Tw offers several advantages over previous models based solely on Ta for the prediction of rice growth and yield on Hokkaido. Our model analysis showed that Tw, which is higher than Ta under normal conditions, is a critical factor both for growing rice in the cool climate of Hokkaido and for analyzing temporal and regional variations in rice productivity under current and future climates.
|
ACKNOWLEDGMENTS
|
|---|
We thank Y. Numao and H. Tanno in the Hokkaido Prefectural Kamikawa Agricultural Experimental Station, and K. Tanaka in the Hokkaido Prefectural Central Agricultural Experimental Station for allowing us to use data of water temperature and grain yield at Pippu, Iwamizawa, and Ohno in Hokkaido. We also thank M. Okada for his valuable comments for model analysis. This study is a part of the doctor thesis of Hokkaido University. A grant-in-aid for scientific research (B) from the Japan Society for the Promotion of Science (Project No. 11460006) supported this study in part.
|
REFERENCES
|
|---|
- Enomoto, N. 1936. Effects of cool irrigation on rice growth and yield. (In Japanese.) Agriculture 662:10–20.
- Fujiwara, T., H. Abe, and T. Ishiguro. 1966. Evaluation of the risk for cool weather damage at the different regions in Hokkaido, Japan. (In Japanese.) Hokunou 33:1–13.
- Hanyu, J. 1971. The stabilities of temperature and rice yield during the cultivation period in Hokkaido. Part 1. The critical conditions of temperature for the safe cultivation of paddy rice. (In Japanese with English abstract.) Res. Bull. Hokkaido Natl. Agric. Exp. Stn. 98:1–11.
- Horie, T., H. Nakagawa, M. Ohnishi, and J. Nakano. 1995. Rice production in Japan under current and future climates. p. 143–164. In R.B. Matthews et al. (ed.) Modeling the impact of climate change on rice production in Asia. CAB International, Oxon, UK.
- Hoshino, T., and S. Okabe. 1960. The increase in stability of crop production in Hokkaido. (In Japanese with English abstract.) Res. Bull. Hokkaido Natl. Agric. Exp. Stn. 75:1–6.
- Ichikawa, S., T. Tsunoda, T. Hashimoto, N. Moki, Y. Wakazawa, M. Urayama, S. Konno, and M. Maekawa. 1995. Agronomic performances of a rice variety, Kirara 397 in different planting types under cool summer in 1993. (In Japanese with English abstract.) Res. Bull. Univ. Farm Hokkaido Univ. 29:23–30.
- IPCC. 2002. Climate change and biodiversity. IPCC Tech. Paper V. Intergovernmental Panel on Climate Change, Geneva.
- Kropff, M.J., R.B. Matthews, H.H. van Laar, and H.F.M. ten Berge. 1995. The rice model ORYZA1 and its testing. p. 27–50. In R.B. Matthews et al. (ed.) Modeling the impact of climate change on rice production in Asia. CAB International, Oxon, UK.
- Kuwagata, T., and T. Hamasaki. 2001. Simulation model for predicting water temperature on the paddy field. (In Japanese.) New Technol. Agric. 14:196–199.
- Parry, M.L., C. Rosenzweig, A. Iglesias, M. Livermore, and G. Fischer. 2004. Effects of climate change on global food production under SRES emissions and socio-economic scenarios. Global Environ. Change 14:53–67.[CrossRef]
- Rosenzweig, S., and M.L. Parry. 1994. Potential impact of climate change on world food supply. Nature 367:133–138.
- Sakai, K. 1949. Effects of deep-flood irrigation on grain yield of rice plants in a cool summer year. (In Japanese.) Agric. Hortic. 24:405–408.
- Satake, T., S.Y. Lee, S. Koike, and K. Kariya. 1988. Male sterility caused by cooling treatment at the young microspore stage in rice plants. XXVIII. Prevention of cool injury with the newly devised water management practices—Effects of the temperature and depth of water before the critical stage. Jpn. J. Crop Sci. 57:234–241.
- Shimono, H., T. Hasegawa, and K. Iwama. 2007. Modeling the effects of water temperature on rice growth and yield under a cool climate: I. Model development. Agron. J. 99:1327–1337 (this issue).[Abstract/Free Full Text]
TOP
ABSTRACT
INTRODUCTION
