Effects of Crop Growth on Hydrological Processes in River Basins and on Regional Climate in China

Abstract

The regional climate model RegCM3 incorporating the crop model CERES, called the RegCM3_CERES model, was used to study the effects of crop growth and development on regional climate and hydrological processes over seven river basins in China. A 20-year numerical simulation showed that incorporating the crop growth and development processes improved the simulation of precipitation over the Haihe River Basin, Songhuajiang River Basin and Pearl River Basin. When compared with the RegCM3 control run, RegCM3_CERES reduced the negative biases of monthly mean temperature over most of the seven basins in summer, especially the Haihe River Basin and Huaihe River Basin. The simulated maximum monthly evapotranspiration for summer (JJA) was around 100 mm in the basins of the Yangtze, Haihe, Huaihe and Pearl Rivers. The seasonal and annual variations of water balance components (runoff, evapotranspiration and total precipitation) over all seven basins indicate that changes of evapotranspiration agree well with total precipitation. Compared to the RegCM3, RegCM3_CERES simulations indicate reduced local water recycling rate over most of the seven basins due to lower evapotranspiration and greater water flux into these basins and an increased precipitation in the Heihe River Basin and Yellow River Basin, but reduced precipitation in the other five basins. Furthermore, a lower summer leaf area index (1.20 m² m^–2 ), greater root soil moisture (0.01 m³ m^–3 ), lower latent heat flux (1.34 W m^–2 ), and greater sensible heat flux (2.04 W m^–2 ) are simulated for the Yangtze River Basin.

Citation

Qin, P.-H., F. Chen, and Z.-H. Xie, 2013: Effects of crop growth on hydrological processes in river basins and on regional climate in China. Adv. Clim. Change Res. , 4 (3), doi: 10.3724/SP.J.1248.2013.173.

Keywords

crop growth ; regional climate ; East Asian monsoon region ; hydrological process

1. Introduction

Wheat, rice, maize and other crops have different physical and physiological characteristics, and their growth and development processes affect the leaf area index, albedo, rainfall interception, runoff and soil moisture, and further affects climate through the exchange of water and energy between land surface and atmosphere. Such crops also significantly influence atmospheric radiation, precipitation and temperature pattern, soil moisture, and groundwater levels. This includes the agricultural system, and irrigation, harvesting and planting activities. Climate simulations must take reasonable account of all of these factors.

Regional climate modeling over the complex terrain of eastern China monsoon region has been carried out in the past 20 years. Such works include numerical simulation of anomalous monsoon summer rainfall [ Shi et al. , 2001 ], summer rainfall return test [ Liu et al. , 2005 ] and its spatial distribution and seasonal evolution [ Liu et al., 2011 , Xu et al., 2011  and Zeng et al., 2009 ], high-resolution simulation of regional climate [ Gao et al., 2011  and Shi et al., 2010 ], groundwater table dynamics, inter-basin water transfer processes coupled with the effects of regional climate transfer on climate impact research [ Chen and Xie, 2010  and Yuan et al., 2008 ], crop growth processes and the effects of land use on climate [ Gao et al., 2007 , Tsvetsinskaya et al., 2001a , Tsvetsinskaya et al., 2001b  and Yu and Xie, 2013 ]. In order to study how agricultural crop growth affects the regional climate, Chen and Xie, 2011a  and Chen and Xie, 2011b ] coupled the crop-growth model for maize, wheat and rice (CERES) with the regional climate model RegCM3 to produce the RegCM3_CERES model. The present study used RegCM3_CERES to simulate the regional climate for the East Asian region over a period of 20 years, investigating the effect of crop growth and its development processes on the hydrological processes and regional climate for seven major river basins in China.

2. Model characteristics

2.1. RegCM3

The International Centre for Theoretical Physics (ICTP) regional climate model RegCM3 [ Pal et al. , 2007 ] (http://users.ictp.it/~pubregcm/RegCM3/ ), is one of the most widely used meteorological models. It is based on the dynamical core of the Pennsylvania State University National Center for Atmospheric Research (PSU/NCAR) fifth-generation meso-scale meteorological model (MM5).

The key physical components of RegCM3 are the radiative transfer scheme of the third version of the U.S. National Center for Atmospheric Research Community Climate Model (NCAR/CCM3), the nonlocal planetary boundary layer Subgrid Explicit Moisture Scheme (SUBEX) for modeling large-scale precipitation, ocean heat flux parameterization, and a land surface model using the Biosphere-Atmosphere Transfer Scheme (BATS1e).

2.2. Crop model CERES 3.0 coupled with RegCM3

CERES 3.0 [ Jones and Kiniry, 1986  and Tsuji et al., 1998 ] is used to simulate the growth and development of crops such as wheat, rice, sorghum, cassava, soybeans, peanuts and potatoes. CERES 3.0 includes crop growth, water balance, crop roots, stems, leaves, spikes, grain growth calculations, and the nitrogen balance.

In order to study how crop growth and its development processes affect the regional climate, many studies have investigated the effect of coupling CERES with the growth processes of maize, wheat, and rice, using the land surface model BATS1e and RegCM3 [ Chen and Xie, 2011a  and Chen and Xie, 2011b ].

3. Experimental arrangements

In this study, RegCM3 was used for the control run (termed CTL model), and compared with the results for the coupled RegCM3 with CERES 3.0 (termed CSM model) to simulate the regional climate over eastern Asia, centered at 36°N, 102°E with 60 km horizontal resolution and 120×90 grid points. Both simulations used information from the European Centre for Medium-Range Weather Forecasts (ECMWF) 40-year re-analysis datasets (ERA-40) for setting initial and boundary conditions. For oceanic surface forcing, data from the National Oceanic and Atmospheric Administration (NOAA) Optimally Interpolated Sea Surface Temperature (OISST) [ Reynolds et al. , 2002 ] was used. The Grell scheme with Fritsch and Chappell closure [ Grell , 1993 ] was chosen as the convection scheme. In the CSM model, the effects of wheat, maize, and rice crops were studied, since they are widely grown in China; the proportional coverage of these crops was obtained from the databases of Global Land Cover Characteristics (GLCC), the Center for Sustainability and the Global Environment (SAGE) crop distribution, the Chinese Zoning Map of Cropping Systems (1989), and from the Multiple Crop Index map of China [ Yan et al. , 2005 ]. The dates of crop planting and harvesting were taken from the China Agricultural Phenology Atlas [ Zhang et al. , 1987 ]. The effects of crops outside of China were not considered; see Chen [2010] for more details.

The model simulations span 20 years, from January 1, 1982 to December 31, 2001. The first year was used for modeling the spin-up, and the results of the latter 19 years’ simulations were analyzed.

Figure 1. Study domain and distribution of the seven major river basins, YZ: Yangtze River Basin; HA: Haihe River Basin; HE: Heihe River Basin; HU: Huaihe River Basin; YL: Yellow River Basin; SH: Songhuajiang River Basin; PE: Pearl River Basin

4. Results

4.1. Precipitation and temperature

Figure 2 (a–g) show the observed and simulated monthly precipitation for the seven river basins. Overall, the coupled model produces significantly reduced biases of precipitation over the basins of the Haihe, Songhuajiang and Pearl Rivers compared to the control simulations, but shows increased summer precipitation over the Huaihe River Basin. Significant differences between precipitation simulations by the two models generally appear in summer, especially in the Huaihe River Basin. This is because the crop-growth simulation models primarily assess changes in leaf area index and stem area index when calculating the regional climate feedback, and summer is the main cropgrowing season.

Figure 2. Simulated and observed monthly precipitation (a–g) and monthly mean temperature (h–n) for the seven river basins (1983–2001 means)

To improve the analysis of crop growth impact on the simulated surface air temperatures (2 m above ground) over the study basins, the significant seasonal temperature variation was taken into account both from observations and simulations of monthly mean air temperatures for the whole 19-year period of 1983–2001 (Fig. 2 h–2 n). The model results show no significant improvement for the Yangtze River Basin, for which the CSM-simulated mean temperature shows a relatively large negative deviation in each year. For the Heihe River Basin, the monthly mean temperature shows an improvement in the negative deviation for many of the studied years. In the Haihe River Basin, and also for the summer months in the Huaihe River Basin, the simulation results show significant improvements.

4.2. Water balance of river basins

A general water balance model is described as change in water storage, ΔS  = P — E — R , where P  is precipitation, E is evapotranspiration and R is runoff, respectively. Figure 3 (a–g) show the inter-annual variations of the water balance components (total precipitation, runoff and evapotranspiration) over the seven river basins from 1983 to 2001. As shown in Figure 3 (a–g), compared with the amount of total precipitation and evapotranspiration, only a small amount of runoff was simulated as expected. Minor changes in runoff were found over the study period for most of the basins. Also, the total precipitation shows clear inter-annual changes, although no significant trends were found. Precipitation peaks appeared in 1998 for the Huaihe and Haihe River Basins, with corresponding varying degrees of increase in evapotranspiration and runoff.

Figure 3. Inter-annual (a–g) and seasonal (h–n) variations of water balance components (runoff, evapotranspiration, and total precipitation) of the seven river basins

Figure 3 (h–n) show the seasonal variations in the water balance components. Evapotranspiration in the summer months (JJA) for the Yangtze, Haihe, Huaihe and Pearl River Basins was basically about 100 mm per month. By comparison, the evapotranspiration rate for the Heihe River Basin was only 20 mm per month because of lower precipitation in that region. The comparison of 1983–2001 mean evapotranspiration rates corresponding to the maximum, the 25th percentile, the 75th percentile, and the minimum values indicates that more evapotranspiration took place in summer than in autumn (SON), especially over the Huaihe, Yellow and Songhuajiang River Basins. Minor inter-annual variations in the summer evapotranspiration rate were revealed for the Songhuajiang and Pearl River Basins.

The precipitation shows more significant interannual variation than either evapotranspiration or runoff in all seven river basins. The total precipitation (about 200 mm per month) in the summer months in the Pearl River Basin is twice as much as the evapotranspiration rate (about 100 mm per month); in autumn the evapotranspiration exceeds the total precipitation, and in winter (DJF) and spring (MAM) it approximates the total precipitation.

In the Yangtze River Basin the evapotranspiration rate in autumn is somewhat greater than the precipitation, but considerably less in spring and summer. However, in the Songhuajiang River Basin, precipitation exceeds evapotranspiration throughout the year. Maximum runoff occurs in summer in all basins except in the Songhuajiang River Basin, where the maximum runoff occurs in April, possibly due to permafrost and melting snow.

4.3. Effects of crop growth on vapor transfer and surface fluxes

In order to detect the role of large-scale water vapor transport on precipitation, the atmospheric moisture budgets were analyzed for the seven river basins. The corresponding balance equations of atmospheric moisture [ Schär et al. , 1999 ] and water cycle indices were defined as:

$\Delta W=Q_{in}-Q_{out}+\vert E-P{\mbox{,}}$

( 1)

$\beta =E/\left(Q_{in}+E\right){\mbox{,}}$

( 2)

$\chi =P/\left(Q_{in}+E\right){\mbox{;}}$

( 3)

where ΔW  (in mm d^–1 ) is the trend in the atmospheric water vapor content; β (in mm d^–1 ) is the local recycling rate; χ (in mm d^–1 ) is precipitation efficiency; Q_in  (in mm d^–1 ) and Q_out  (in mm d^–1 ) are the summer mean water fluxes entering and exiting each river basin; E  (in mm d^–1 ) is the mean evapotranspiration over each river basin; and P  (in mm d^–1 ) is the mean precipitation in each river basin.

Table 1 shows the simulated atmospheric water balance components and water cycle indices for summer in all seven river basins. In these computations, a rectangular domain of approximately the same area as the basin, including the river, is used for the water vapor flux into each river basin. Taking the crop growth process into account, more water vapor influx and efflux of the Yangtze, Heihe and Pearl River Basins were found. The influx increased by 1.34 mm and the efflux increased by 1.66 mm for the Pearl River Basin, while for the Huaihe and Yellow River Basins the water fluxes are decreasing. The water vapor convergence (MC  = Q_in  — Q_out ) from CSM run was 0.32 mm less than that from CTL run in the Pearl River Basin,so was in the Songhuajiang River Basin, while more amount were found for the other basins. The local water recycling rates showed lower level in all basins except in the Heihe River Basin due to the less evapotranspiration and more water vapor influx. In addition, the coupled model simulated higher precipitation efficiency in the Heihe and Yellow River Basins and lower precipitation efficiency in the other basins, including a higher amount of 1.87% in the Pearl River Basin.

Table 1. Atmospheric water balance components and water cycle indices for summer months in the seven river basins (CTL: RegCM3 control; CSM: coupled RegCM3_CERES)

Basin	Model	P  (mm d–1 )	E  (mm d–1 )	Qin  (mm d–1 )	Qout  (mm d–1 )	MC  (mm d–1 )	ß  (%)	χ  (%)
YZ	CTL	4.46	3.31	7.41	5.86	1.55	30.88	41.60
	CSM	4.46	3.27	7.56	5.97	1.59	30.19	41.18
HA	CTL	4.24	3.25	15.10	14.28	0.82	17.71	23.11
	CSM	4.14	2.86	15.21	14.15	1.06	15.83	22.91
HE	CTL	0.90	0.61	6.27	5.96	0.31	8.87	13.08
	CSM	0.93	0.63	6.37	6.04	0.33	9.00	13.29
HU	CTL	3.52	3.02	9.10	8.68	0.42	24.91	29.04
	CSM	3.10	2.57	8.74	8.25	0.49	22.72	27.41
YL	CTL	2.83	2.27	9.96	9.60	0.36	18.56	23.14
	CSM	2.77	2.17	9.76	9.37	0.39	18.19	23.22
SH	CTL	4.66	3.19	13.93	12.36	1.57	18.63	27.22
	CSM	4.51	3.06	14.24	12.70	1.54	17.69	26.07
PE	CTL	7.74	4.14	25.19	21.16	4.03	14.12	26.39
	CSM	7.50	4.06	26.53	22.82	3.71	13.27	24.52

The Yangtze River Basin was chosen as a case study to show the effects of crop growth and development on land-atmosphere interaction during the summer months. Figure 4 shows the mean differences in summer between the simulations of the CSM and CTL models of the leaf area index (LAI), soil moisture, latent heat flux, sensible heat flux, precipitation, and temperature. When the crop-growth process was taken into account by the regional climate model (CSM), the simulated summer LAI for the middle and lower reaches of the Yangtze River Basin is lower compared to the results from the control run (CTL). This finding is mainly due to the effects of the primary crops, which are maize in the middle reaches and rice in the lower reaches of the basin.

Figure 4. Summer mean differences (CSM minus CTL) for the Yangtze River Basin for (a) LAI, (b) root layer soil moisture, (c) latent heat flux, (d) sensible heat flux, (e) precipitation, and (f) temperature

The LAI of rice, which has just been planted in summer, is smaller than the LAI of maize; thus the LAI in the middle reaches of the Yangtze River Basin increases significantly with the growth of maize. However, in the CTL model the LAI is calculated from the deep-soil temperature, with the result that it simulates a larger LAI in summer when soil temperatures can be expected to be higher. By comparison, the LAI over the whole Yangtze River Basin simulated by the CSM model is 1.20 m² m^–2 lower than that by the CTL model (Table 2 ). The LAI reduces the transpiration from the vegetation and increases the root-zone soil moisture by 0.01 m³ m^–3  (Fig. 4 b; Table 2 ), which leads to a lower amount by 1.34 W m^–2 in the latent heat flux and to a higher amount by 2.04 W m^–2 in sensible heat flux. The total precipitation simulated by the CSM model is also lower in the Yangtze River Basin, which is consistent with lower values in simulated LAI. Slight differences were found in the simulated total precipitation in the upper reaches of the Yangtze River Basin, consistent with the minor differences in LAI.

Table 2. Summer mean differences (CSM minus CTL) for LAI, root layer soil moisture (RSW), latent heat flux (LE), sensible heat flux (SENA), precipitation, and temperature over all the seven river basins

Basin	LAI (m2 m–2 )	RSW (m3 m–3 )	LE (W m–2 )	SENA (W m–2 )	Precipitation (mm d–1 )	Temperature (°C)
YZ	–1.20	0.01	–1.34	2.04	0.00	0.29
HA	–3.45	0.03	–11.23	7.26	–0.10	0.79
HE	0.00	0.00	0.52	–0.69	0.03	0.05
HU	–4.56	0.04	–13.16	9.80	–0.42	0.59
YL	–1.30	0.01	–3.06	1.69	–0.06	0.31
SH	–2.30	0.01	–3.79	7.12	–0.15	0.62
PE	–1.64	0.02	–2.43	4.19	–0.24	0.40

Compared with the control run, temperature in the Yangtze River Basin simulated by the CSM model is higher by around 0.29°C, especially in the Sichuan Basin (more than 1.0°C). The spatial patterns of the simulated temperature differences are consistent with those of the sensible heat flux. This also underlines the effectiveness of the simulated results, since any higher amounts in temperature should lead to a higher sensible heat flux. Additionally, the summer LAI simulated by the CTL model (calculated through the deep soil temperature, as mentioned above) reaches 5.98 m² m^–2 in the Huaihe River Basin, which is unrealistically high. When the progress of crop growth was accounted for using the CSM model, the overestimation of the LAI fell by 4.56 m² m^–2 and thus the simulated latent heat flux is substantially 13.16 W m^–2  lower and the sensible heat flux is 9.80 W m^–2 higher in the Huaihe River Basin.

5. Conclusions and discussion

In this study, the regional climate model RegCM3 coupled with the crop model CERES was used to study the effects of crop growth and its development on regional climate patterns and hydrological processes over seven river basins in China. A 20-year numerical simulation showed that incorporating the crop growth and its development processes significantly improved the simulated precipitation amounts in the Haihe, Songhuajiang and Pearl River Basins. The impact of crop growth and its development on regional climate was more obvious in summer, because the simulated LAI mainly increases during summer.

The coupled model simulates the water balance components well for the seven river basins. Minor differences between the simulated and observed runoffs were found for most of the basins over the period 1983–2001. The summer evapotranspiration in the Yangtze, Haihe, Huaihe and Pearl River Basins was about 100 mm per month. The effect of crop growth on water vapor transfer was also analyzed in this study. When the crop growth process was considered, both the water vapor fluxes into and out of the Yangtze, Heihe, Songhuajiang and Pearl River Basins were higher, but lower in the Huaihe and Yellow River Basins, which was related to precipitation amount. Futhermore, the Yangtze River Basin was chosen as a case to show the effect of crop process on land-atmosphere interaction. It was found that crop growth caused lower LAI, lower latent heat flux, and higher sensible heat flux and led to lower precipitation and higher temperature.

Many uncertainties remain in detecting the effects of crop growth on regional climate pattern using the coupled regional climate model, although different factors influencing hydrological processes and regional climate patterns are combined. Difficulties prevail due to the fact that the vegetation cannot be described in adequate detail in regional climate models. These aspects will be the subjects of further work.

Acknowledgements

The authors would like to thank the anonymous reviewers for their comments and suggestions on this paper. This work was supported by the National Basic Research Program of China (Nos. 2010CB428403 and 2010CB951001), and the National Natural Science Foundation of China (No. 91125016).

References

1. Chen, 2010 F. Chen; Investigating the effects of crop growth and development and interbasin water transfer on regional climate, Institute of Atmospheric Physics, Chinese Academy of Sciences (2010), p. 154
2. Chen and Xie, 2010 F. Chen, Z.-H. Xie; Effects of interbasin water transfer on regional climate: A case study of the middle route of the south-to-north water transfer project in China; J. Geophys. Res., 115 (D11) (2010) D11112
3. Chen and Xie, 2011a F. Chen, Z.-H. Xie; Effects of crop growth and development on regional climate: A case study over East Asian monsoon area; Climate Dynamics (2011) http://dx.doi.org/10.1007/s00382-00011-01125-y
4. Chen and Xie, 2011b F. Chen, Z.-H. Xie; Effects of crop growth and development on land surface fluxes; Advances in Atmospheric Sciences, 28 (4) (2011), pp. 927–944
5. Gao et al., 2011 X.-J. Gao, Y. Shi, F. Giorgi; A high resolution simulation of climate change over China; Science China Earth Sciences, 54 (3) (2011), pp. 462–472
6. Gao et al., 2007 X.-J. Gao, D.-F. Zhang, Z.-X. Chen, et al.; Land use effects on climate in China as simulated by a regional climate model; Science in China Series D: Earth Sciences, 50 (4) (2007), pp. 620–628
7. Grell, 1993 G.A. Grell; Prognostic evaluation of assumptions used by cumulus parameterizations; Monthly Weather Review, 121 (3) (1993), pp. 764–787
8. Jones and Kiniry, 1986 C. Jones, J. Kiniry; CERES-Maize: A Simuation Model of Maize Growth and Development, Texas A & M University Press (1986), p. 194
9. Leff et al., 2004 B. Leff, N. Ramankutty, J.A. Foley; Geographic distribution of major crops across the world; Global Biogeochemical Cycles, 18 (1) (2004) GB1009
10. Liu et al., 2011 X.-P. Liu, H.-J. Wang, J.-B. Liu; Influence of spatial resolution in a regional climate model on summer precipitation simulation; Advances in Water Science (in Chinese), 22 (5) (2011), pp. 615–623
11. Liu et al., 2005 Y.-M. Liu, Y.-H. Ding, Q.-Q. Li; 10-year hindcasts and assessment analysis of summer rainfall over China from regional climate model; Quarterly Journal of Applied Meteorology (in Chinese), 16 (2005), pp. S41–S47
12. Pal et al., 2007 J.S. Pal, F. Giorgi, X.-Q. Bi, et al.; Regional climate modeling for the developing world: The ICTP RegCM3 and RegCNET; Bull. Amer. Meteor. Soc., 88 (9) (2007), pp. 1395–1409 2007
13. Reynolds et al., 2002 R.W. Reynolds, N.A. Rayner, T.M. Smith, et al.; An improved in situ and satellite SST analysis for climate; J. Climate, 15 (13) (2002), pp. 1609–1625
14. Schär et al., 1999 C. Schär, D. Lüthi, U. Beyerle, et al.; The soil-precipitation feedback: A process study with a regional climate model; J. Climate, 12 (3) (1999), pp. 722–741
15. Shi et al., 2001 X.-L. Shi, Y.-H. Ding, Y.-M. Liu; Simulation experiments of summer rainbelt in China with the regional climate model; Climatic and Environmental Research (in Chinese), 6 (2) (2001), pp. 249–254
16. Shi et al., 2010 Y. Shi, X.-J. Gao, F. Giorgi, et al.; High resolution simulation of changes in different-intensity precipitation events over China under global warming; Advances in Climate Change Research (in Chinese), 6 (3) (2010), pp. 164–169
17. Tsuji et al., 1998 G. Tsuji, G. Googenboom, P. Thornton; Understanding Options for Agricultural Production, Kluwer Academic Publishers (1998), p. 400
18. Tsvetsinskaya et al., 2001a E.A. Tsvetsinskaya, L.O. Mearns, W.E. Easterling; Investigating the effect of seasonal plant growth and development in threedimensional atmospheric simulations. Part I: Simulation of surface fluxes over the growing season; J. Climate, 14 (5) (2001), pp. 692–709
19. Tsvetsinskaya et al., 2001b E.A. Tsvetsinskaya, L.O. Mearns, W.E. Easterling; Investigating the effect of seasonal plant growth and development in threedimensional atmospheric simulations. Part II: Atmospheric response to crop growth and development; J. Climate, 14 (5) (2001), pp. 711–729
20. Xu et al., 2011 X. Xu, R.-Y. Lu, Y. Shi; Comparison between the results on seasonal evolution of summer precipitation over eastern China simulated by a regional climate model and the driving GCM; Chinese Journal of Atmospheric Sciences (in Chinese), 35 (6) (2011), pp. 1177–1186
21. Yan et al., 2005 H.-M. Yan, J.-Y. Liu, M.-K. Cao; Remotely sensed multiple cropping index variations in China during 1981–2000; Acta Geographica Sinica (in Chinese), 60 (4) (2005), pp. 559–566
22. Yu and Xie, 2013 Y. Yu, Z.-H. Xie; A simulation study on climatic effects of land cover change in China; Adv. Clim. Change Res., 4 (2) (2013) http://dx.doi.org/10.3724/SP.J.1248.2013.117
23. Yuan et al., 2008 X. Yuan, Z.-H. Xie, J. Zheng, et al.; Effects of water table dynamics on regional climate: A case study over East Asian monsoon area; J. Geophys. Res., 113 (D21) (2008) D21112
24. Zeng et al., 2009 X.-M. Zeng, J.-B. Liu, S. Song, et al.; Effects of vertical resolution on simulation of summer precipitation in China by a regional climate mode; Journal of Hydrodynamics (Ser A), 24 (1) (2009), pp. 71–81
25. Zhang et al., 1987 F.-C. Zhang, D.-H. Wang, B.-J. Qiu; Agrephenological Altas of China (in Chinese), Science Press (1987), p. 202