| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a (Consellería do Medio Rural, 27003 Lugo) Dep. de Producción Vexetal, Univ. de Santiago de Compostela, Campus Univ., 27002 Lugo, Spain
b (CIAM, Apdo. 10, 15080 A Coruña), Dep. de Producción Vexetal, Univ. de Santiago de Compostela, Campus Univ., 27002 Lugo, Spain
c Dep. of Agronomy, Univ. of Florida, Gainesville, FL 32611
d Dep. de Biología Vegetal, Escuela Técnica Superior de Ingenieros Agrónomos, Univ. Politécnica de Madrid, Avenida de la Complutense s/n, 28040 Madrid, Spain
* Corresponding autor (federico.sau{at}upm.es).
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
Abbreviations: ET, evapotranspiration
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
Received for publication March 7, 2007.
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
During the last 20 yr, crop modeling has proven to be a valuable tool that can shorten significantly the experimental process needed to improve rainfed cropping-strategies or water management in irrigated systems, using long term, multiyear weather simulations (Egli and Bruening, 1992; Hook, 1994; Muchow et al., 1994; Wilks and Wolfe, 1998; Ruíz-Nogueira et al., 2001; Jagtap and Abamu, 2003; Nijbroek et al., 2003). Crop models have also been used for predicting climate change impacts on agricultural systems (Tubiello et al., 2002; Mínguez et al., 2007). Nevertheless, the utility of crop models for this objective depends on the accuracy and reliability of their predictions of grain yield under the given environment and management systems, so there is a continuing need to test the models under a wide range of environments and cropping practices (Kiniry and Bockholt, 1998). This testing will allow finding possible flaws of the model, and help propose modifications to improve their predictions and reliability (Sau et al., 1999).
Possibly due to its relative simplicity, CERES-Maize (Jones and Kiniry, 1986; Jones et al., 2003) is one of the most used, among the multiple maize models that exist. Since its first release in 1986 (Jones and Kiniry, 1986), CERES-Maize has been widely applied under different environments and purposes (Epperson et al., 1992; Kiniry and Bockholt, 1998; Boote et al., 2001; Jagtap and Abamu, 2003; Tubiello et al., 2002; Mínguez et al., 2007). It is a deterministic model that predicts the timing of the different phenological stages of the crop (emergence, tassel initiation, anthesis, and harvest maturity), growth rate and the partitioning of biomass to growing organs (roots, stem, leaves, and kernels) with a daily time step. Under nonlimiting conditions (water and N fully available), the simulated processes are affected by main three environmental variables (daily solar radiation, maximum temperature, and minimum temperature), cultivar-specific factors (five genetic coefficients that define cultivar characteristics (Jones and Kiniry, 1986), and crop management practices (i.e., sowing date, density, row spacing, and dates and amounts of irrigation and N fertilization). This crop model, as all those included in the DSSAT software (Jones et al., 2003), allows simulation of crop development and growth under water and N-limiting environments when the water and N balance options are switched on. The daily-soil water balance in the CERES-4.0 model, as all the DSSAT 4.0 models, uses the Ritchie (1985) one-dimensional "tipping bucket" soil water balance, which predicts soil water flow and root water uptake for each of up to 10 soil layers. Using this method requires each layer (L) of the soil profile to be defined by a characteristic drained upper limit or field capacity (DUL(L), cm3 H2O cm–3 soil), a lower limit or wilting point [LL(L), cm3 H2O cm–3 soil], a saturated soil water content [SAT(L), cm3 H2O cm–3 soil], and a soil-rooting preference function [SRGF(L); relative root length distribution factor (0.00–1.00) in soil layer L] that quantifies the potential root hospitality of a particular layer.
The daily biomass accumulation (CARBO, for abbreviations, see appendix) and its partitioning between the different organs are described for the original CERES-Maize model by Jones and Kiniry (1986). The main differences from the older model versions to the current CERES-4.0 are highlighted by López-Cedrón et al. (2005). Because this present paper is focused on the ability of CERES-4.0 to predict water deficit effects on biomass and grain yield, the important equations of this model version that quantify water deficit and its effects on biomass accumulation and partitioning are described below.
Evapotranspiration Options in CERES-4.0
Currently CERES-4.0 allows the use of different equations to compute daily potential evapotranspiration (ET) (E0, mm d–1), here considered equivalent to the "crop ET under standard conditions" (ETc) as defined and employed by Allen et al. (1998), including: (i) Priestley-Taylor (formulated by Ritchie (1972, 1985; called PT hereafter)), and (ii) Penman-Montieth-FAO56 (Allen et al., 1998; called PFAO56 hereafter). The Priestley-Taylor (i) is the default option and is the most widely used, mainly because it requires less input (daily solar radiation and minimum and maximum temperatures). PFAO56 (ii) requires additional weather data (daily average dew point temperature and wind speed) and is currently computed with a crop coefficient (Kc) of 1.00 for the entire crop cycle and thus does not allow E0 to exceed the hypothetical grass reference surface ET (ETo; E0 = ETc = Kc x ETo). Allen et al. (1998) uses a variable Kc coefficient that varies with crop cover and height and can reach 1.15–1.20 as a maximum for a maize crop, taking into account its lower aerodynamic resistance by comparison with the grass reference. The use of these different E0 options for the CROPGRO-faba bean model are described by Sau et al. (2004); however these ET options have not been tested for CERES-Maize model V4.0.
Partitioning of Potential Evapotranspiration between Evaporation and Transpiration in CERES-4.0
The models partition E0 to potential soil evaporation (ES0) and potential plant transpiration (EP0), following the Ritchie (1972, 1985) approach, which considers the portion of solar radiation (SR) reaching the soil that can be spent as latent energy to evaporate water from the soil surface if the soil is wet. The fraction of SR reaching the soil is a function of leaf area index (LAI) (Eq. [1]). Actual soil evaporation (ES) and plant transpiration (EP) subsequently depend on the availability of water to meet these potential rates.
The potential soil evaporation (ES0) in DSSAT V4.0 is calculated from E0 as follows:
 | [1] |
The DSSAT V4.0 models follow the premise (Ritchie, 1985) that actual soil evaporation takes place in two stages: (i) the constant stage or energy limited (stage 1) and (ii) the falling rate stage (stage 2). During stage 2, ES is smaller than ES0.
Partitioning of E0 to climatic potential transpiration of the crop (EP0) is calculated using Eq.[2] and [3], with the same extinction coefficient (KEP = 0.685) used for ES0 computation.
 | [2] |
 | [3] |
Note that, in V4.0 the same KEP = 0.685 correctly applies to both ES0 and EP0 (see Eq. [1] and [2]). In V3.5 and earlier CERES model versions, the extinction coefficients were different for these two processes: it was approximately 0.45 for ES0 but was equal to 1.00 for EP0.
Finally, the transpiration of the crop (EP), the water deficit factor on photosynthesis (SWFAC) and turgor factor (TURFAC) are calculated.
Kernel Number per Plant and Kernel Growth Rate in CERES-4.0
As in other cereals, maize grain yield is strongly related to kernel number m–2 at harvest and thus an accurate prediction of kernel number set per plant (GPP) is crucial for a maize crop model to correctly estimate production (Ritchie and Alagarswamy, 2003). Under rainfed conditions, a good simulation of the water deficit reduction on GPP will be needed. In CERES-4.0, GPP is a function of PSKER (average rate of photosynthesis during stage 4 (anthesis to effective grain filling period), Eq. [4] and [5]). The PSKER is directly related to the values of CARBO during stage 4 and therefore linearly influenced by the actual SWFAC effects on CARBO. Thus as SWFAC decreases from 1.00 to 0.00 (more water deficit), CARBO and GPP diminishes.
 | [4] |
 | [5] |
In CERES-4.0, grain growth rate per day (GROGRN) depends on temperature (RGFILL (relative rate of grain fill) is a temperature function), number of grains per plant, potential kernel growth rate per day (G3, mg kernel–1 d–1) and soil water stress factor on photosynthesis:
 | [6] |
The objective of this research was to predict accurately maize biomass, grain yield and grain yield components under water-limiting conditions in northwest Spain, an environment where CERES-4.0 proved to be the most adequate of three tested CERES-Maize versions for good predictions of biomass at harvest and grain yield under nonlimiting water conditions (López-Cedrón et al., 2005). In the present paper, we observed that CERES-4.0 with default settings failed to simulate sufficiently high growth and yield under water-limited conditions. Thus we re-evaluated potential ET, energy partitioning to E0 and EP0, grain number, and grain growth rate aspects of model code responsible for poor performance under water-limited rainfed environment.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
Experiments in 1999 and 2000 had two sowing dates (first and second), while in 1998, 2001, and 2002, only one sowing date was implemented (see Table 1 ). Before maize, during fall and winter, the experimental field was cultivated with Italian ryegrass (Lolium multiflorum L.) as a cover crop that was harvested approximately 2 wk before the maize sowing date. Soil was fertilized with ample amounts of N, P, and K (see López-Cedrón et al., 2005; irrigated and rainfed treatments were fertilized the same) to minimize limitation from these nutrients. Temperatures of the different growing seasons can be seen in Table 2 and were relatively cool when compared to the main world growing areas for maize. In all experiments, row spacing and sowing depth were 0.75 and 0.05 m, respectively. Plant density was close to 10 plants m–2 (Table 1 lists the actual field-sampled plant densities used for model simulations). Plot size was 12.0 by 7.0 m (16 rows of 7 m length) and had four replications. Drip irrigation was established for all irrigated plots and managed to avoid plant water stress.
|
|
|
Soil Characteristics, Soil Water Measurements and Initial Conditions
The soil at the experimental site was a Typic Haplumbrept (USDA Soil Taxonomy) with a sandy-loam texture and had a pH in water of 5.6. The bulk density, texture, and carbon content of the different layers were measured for two sites within the experimental field. These data were included in the file (SOIL.SOL) that DSSAT model uses to define soil characteristics (Table 4
).
|
|
Finally, to monitor water extraction during the 2001 growing season, soil water contents were determined gravimetrically to 150 cm depth with three replications at four dates, 22 June, 5 July, 31 July, and 26 September.
Volumetric Soil Water Content at Saturation and Soil Rooting Preference Function—Final Adjustment of Drained Upper Limit (L) for Simulations
Soil water content at saturation [SAT(L); % vol] and soil rooting preference function for each layer [SRGF(L)] (see Table 4) was calculated by the algorithm included in DSSAT-4.0 based on input of measured soil characteristics (layer thickness and depth, clay, silt, coarse, and organic fractions C). This same algorithm was used to set soil albedo (SALB; fraction) as 0.13, evaporation limit for stage 1 (SLU1; mm) as 6.0 mm, drainage rate (SLDR; fraction d–1) as 0.60 d–1, and soil run-off curve number (SLRO; Soil Conservation Service) as 73.
The LL(L) values placed in the file for the model runs were obtained through the method described previously (Table 5). Similarly, the DUL(L) values placed in the file for model runs came from data described previously in Table 5, although minor adjustments (<1%) to the experimentally-determined values were made to fit better simulated to observed water extraction dynamics of the 2001 treatment.
CERES-Maize Model Version
The CERES-Maize version (CERES-4.0) used in this work is the one included in DSSAT-4.0. The code and the model can be purchased through the International Consortium for Agricultural Systems Applications (ICASA-DSSAT) webpage (www.icasa.net/dssat/index.html; verified 3 Oct. 2007).
Model Options and Coefficients Based on Irrigated Treatments
Model runs with CERES-4.0 were performed, with water balance simulation switched "on," using (i) the PT ET as modified by Ritchie (PT; Ritchie, 1985) and (ii) the Penman–Monteith-FAO56 (PFAO56; Allen et al., 1998) method. Initial measured soil water content values were input at sowing date, except in 2001 when initial water content was adjusted to fit simulated water content to soil water content measurements made on 22 May 2001. As N supply proved to be nonlimiting, the model N balance was switched "off." The five genetic coefficients used to define the Clarica cultivar had been previously estimated in a low density experiment (see López-Cedrón et al. (2005)). Nevertheless, with these low density "inputs" CERES-4.0 slightly overpredicted biomass and grain yield of the seven irrigated treatments by approximately 1.9% and 9.4% (due to excessive simulated weight per average kernel (38%) and slightly underpredicted simulated seed number per land unit (21%)). Thus we calibrated G2 (potential kernel number per plant) and G3 (potential kernel growth rate) to optimize the model fit to grain number and weight of all high density irrigated treatments. After this change (now G2 = 1205.0 potential kernel number per plant instead of 936.0, and G3 = 5.75 mg seed–1 d–1 instead of 8.00) the average biomass, grain yield, harvest index (HI), kernel weight and number of kernels per unit land area of the seven irrigated treatments were accurately predicted. Average observed and simulated biomass, grain yield, kernel weight and kernel number per unit of land area of the seven irrigated treatments before evaluating rainfed treatments were respectively: 22,228 and 22,113 kg ha–1; 11,138 and 11,184 kg ha–1; 231 and 231 mg seed –1; 4847 and 4832 kernel number m–2. For these variables, the respective computed root mean square errors (RMSE) were: 2064 kg ha–1, 1330 kg ha–1, 24 mg seed –1, and 418 kernel number m–2.
Statistical and Graphical Procedures to Evaluate Evapotranspiration Options, Extinction Coefficients and Model Modifications
The following criteria were used to assess performance of the CERES-Maize model: (i) intercept (a) and slope (b) values of linear regression between simulated and observed biomass at harvest, grain yield at harvest, and grain harvest index using the seven treatments shown in Table 1; (ii) the RMSE of these variables; and (iii) an index of agreement (d; Willmott, 1982) that is an aggregate overall indicator that is of more value than R2.
RMSE and d were computed as follows:
 | [7] |
 | [8] |
(average of the observed) and Oi'=Oi–. According to Willmott (1982), the model fit improves as d-index approaches unity and RMSE approaches zero.
Because we wanted to test the ability of the model to simulate water deficit effect on reduction in biomass, grain yield, and yield components by comparison to irrigated predictions, we also calculated the same statistics above applied to (Predicted IR – Predicted RF) vs. (Observed IR – Observed RF).
Finally, predicted time series graphs of biomass, grain, and LAI were also compared visually with measurements to assess accuracy of time-series performance of the different CERES-Maize versions. Data from the in-season dry matter samplings (simulated vs. observed) were used for calculating a, b, RMSE, and d-index to compare the different model versions.
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
|
|
|
|
|
|
Soil Characteristics, Initial Conditions, and Soil Water Extraction Dynamic
Because soil water holding characteristics (LL, DUL) (Tables 3 and 4), and initial conditions used in the simulations were established experimentally, uncertainty of water supply and water holding traits cannot be viewed as responsible for model failure to predict biomass and grain yield. Moreover, measured soil water holding capacity (DUL-LL) was high (e.g., 13.2–23.7% vol.; Table 4) for a soil of this texture and a soil pedotransfer function would actually have assigned lower values for many of the soil layers. In addition, time series data on soil water content during the 2001-RF experiment (a relatively rainy season) (Fig. 3
), showed that the PT default option generally predicted too much water extraction in all layers below 15 cm. Likewise, the soil water data showed no observed extraction below 1.20 m and visual observation of the rooting profile made in the four rainfed plots at the end of 2002 growing season showed that very few roots reached this depth and none surpassed it. Furthermore, runoff was insignificant (averaged 5.4 mm over the seven rainfed treatments) because of a relatively moderate curve number, low rainfall and low intensity per day. Therefore, runoff, even if it could be reduced to zero, would have given less than a 0.5% yield increase, and was therefore not a significant cause for soil water depletion, reduced growth, and yield in this study. Thus it appears that the PT ET option overestimated the actual ET of the crop, and was therefore responsible for the generally too early and too severe predicted water deficit, which was the cause for the poor prediction of biomass and yield of CERES-4.0 under rainfed conditions.
|
Evaluation of Alternate Potential Evapotranspiration Equations
Penman–Monteith-FAO56 equation (PFAO56) with a crop coefficient (Kc) of 1.00 is available in CERES-4.0, and was one of the ET options that Sau et al. (2004) suggested to improve biomass and grain yield estimations. This is the equation presently recommended by the FAO to compute reference ET. Even with the default KEP coefficients (0.685), the PFAO56 option resulted in greater predictions of biomass, yield, and HI for rainfed treatments (Tables 6, 7, and 8). Average predicted biomass at harvest, grain yield, and HI increased from 11,426 to 13,830 kg ha–1, 4405 to 5759 kg ha–1, and 0.360 to 0.392, respectively, along with a clear reduction of RMSE and increase in d statistic closer to one. Under this cool environment for maize, the PFAO56 ET option predicted less accumulated E0 than PT. The average accumulated predicted E0 for the seven seasons was 534.8 and 468.4 mm with PT and PFAO56, respectively.
We also tried the PFAO56 option with a Kc value of 1.10, following the approach used by Sau et al. (2004) in which Kc allows E0 to increase from ETo of 1.00 to 1.10 x ETo as LAI increases from 0.00 to maximum LAI of 6.00). This option (PFAO56-Kc = 1.10) is not presently available in CERES-4.0 and would require a small FORTRAN code modification to be implemented. It provided a better fit than PT for biomass, yield, and HI but poorer predictions than PFAO56-Kc = 1.00. Thus, despite FAO recommendation to allow Kc to attain values above 1.00 for tall crops (Allen et al., 1998), for the present formulation and simulations this option proved to be too stressful.
Evaluation of Extinction Coefficient to Partition Potential Evapotranspiration to Potential Transpiration and Potential Evaporation
As Sau et al. (2004) showed that simulations could benefit using a KEP of 0.500 to partition ET between ES0 and EP0, and because this change is supported experimentally (Villalobos and Fereres, 1990) and by radiation extinction theory (Goudriaan, 1977; Goudriaan and van Laar, 1994), we tried a KEP of 0.500 in place of default CERES-4.0 KEP (0.685).
This reduction in KEP improved predictions with all three tested ET options (PT, PFAO56-Kc = 1.00) and PFAO56-Kc = 1.10) by systematically increasing average predicted biomass, grain yield, and HI of the rainfed treatments, reducing RMSE and increasing the d statistic (Tables 6, 7, and 8). In addition, reducing KEP from 0.685 to 0.500 improved the PT option predictions of soil water content for the different soil layers, reducing early simulated soil water extraction by the crop (Fig. 3) and improved the fit and the statistics of the time series simulations (Fig. 2), reducing RMSE for biomass from 3.870 to 3.002 t ha–1 and increasing d from 0.8950 to 0.9401, and reducing RMSE for grain yield from 5.580 to 2.954 t ha–1and increasing d from 0.7292 to 0.8066. This KEP change allowed slopes (b) closer to one and intercepts (a) closer to zero for simulated vs. measured time series biomass and grain data. These results are consistent with those shown by Sau et al. (2004) for faba bean (Vicia faba L.) in a cool-season Mediterranean climate and with those obtained for wheat in New Zealand by Jamieson et al. (1998). Both papers showed that DSSAT models with PT option tended to overpredict water extraction under rainfed conditions, especially during early season. Sau et al. (2004) attributed this problem of the model to the use of an excessive KEP (1.00 in CERES-3.5, 0.85 in CROPGRO in DSSAT V3.5 release) and to a possible overestimation of E0 by PT option under cool and low VPD environments. Three wheat crop models, SIRIUS, SWHEAT and AFRCWHEAT2 (Jamieson et al., 1998) use a KEP of 0.45. All these examples suggest that the KEP used by CERES-4.0 is still high despite being reduced from 1.00 in CERES-3.5 to 0.685 in CERES-4.0. When CERES-4.0 with PT option for ET was run with the ET partitioning equations of CERES-3.5, all statistics were worse for prediction of biomass, yield, and HI. The RMSE increased from 5027 kg ha–1, 3728 kg ha–1, 0.155 to 6378; 4263; and 0.161 for biomass, yield and HI, respectively.
Evaluation of Alternative Functions to Estimate Daily Growth of the Grain and Grain Number per Plant
The PFAO-56 option with KEP = 0.500 was sufficient to accurately predict average biomass of the rainfed treatments at harvest (15,780 kg ha–1 predicted vs. 15,654 measured) but its simulations were still low for grain yield (6821 kg ha–1 predicted vs. 7728 measured) due to too small simulated kernel number per square meter (3370 seed m–2 estimated vs. 3900 measured). This failure to adequately predict drought effect on grain number (Table 9), led us to re-evaluate the grain number functions of the model. Literature review shows that CERES-Maize predictions of grain number per plant are generally less accurate than those for biomass and yield, despite good predictions of grain yield (Piper and Weiss, 1990; Jagtap et al., 1993; Lizaso et al., 2001). The existing function (Eq. [4]) was based on experiments done in Kenya by B.A. Keating and B.M. Wafula (unpublished data, 1992). It is a linear function of average rate of photosynthesis during stage 4 (PSKER) that establishes a minimum of 50 kernels per plant. On the other hand, the GPP version of the original CERES-Maize model (Jones and Kiniry, 1986) was the one proposed by Edmeades and Daynard (1979):
 | [9] |
In Eq. [9], a minimum rate of photosynthesis during stage 4 of 195 mg plant–1 d–1 is needed for the plant to set kernels. Upon comparison of the two equations where both were calibrated to the irrigated crop, it appeared that, under our experimental conditions (seven rainfed treatments), environmental factors that reduced PSKER via SWFAC had a much stronger reduction effect in Eq. [4] than in the Edmeades equation (Fig. 4 ). In addition, Ritchie and Alagarswamy (2003) in their paper recommended that CERES-Maize should be using a nonlinear function for kernel number vs. cumulative intercepted radiation.
|
|
|
|
CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
|
APPENDIX |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
d: index of agreement
DUL(L): drained upper limit of soil layer L (fraction in volume)
E0: potential daily evapotranspiration of the crop and soil (mm d–1)
EP: transpiration of the crop (mm d–1)
EP0: potential transpiration of the crop (mm d–1)
ES: soil evaporation (mm d–1)
ES0: potential soil evaporation (mm d–1)
ET: evapotranspiration
ETc: crop evapotranspiration under standard conditions (mm d–1)
ETo: crop reference evapotranspiration (mm d–1)
GROGRN: grain growth rate per plant per day (g plant–1 d–1)
GPP: kernel number per plant
G2: genetic coefficient that defines the potential number of kernels per plant
G3: genetic coefficient that defines the potential kernel growth rate per day (mg kenel–1 d–1)
HI: harvest index
IR: irrigated treatment
Kc: crop coefficient that increases daily crop reference evapotranspiration linearly from 1.0 times ETo to Kc times ETo, as LAI increases from 0 to 6
KEP: LAI extinction coefficient for partitioning potential evapo- transpiration between soil evaporation and plant transpiration
LAI: leaf area index
LL(L): soil layer L water content at the lower limit or permanent wilting point (fraction in volume)
PSKER: average rate of photosynthesis per plant from anthesis to beginning effective grain filling period (mg (CH2O) plant–1 d–1)
RGFILL: temperature function that describes relative rate of grain fill (0.00–1.00)
RF: rainfed treatment
RMSE: root mean square error
RWU: potential root water uptake of the crop (mm d–1)
RWU(L): potential root water uptake of the crop from soil layel L (mm d–1)
SALB: soil albedo (fraction)
SAT(L): soil layer L water content at saturation (fraction in volume)
SR: daily total solar radiation (MJ m–2 d–1)
SRGF(L): relative root length distribution factor (0.00–1.00)
SWFAC: soil water stress factor that affect photosynthesis (0.00–1.00)
Tmax: mean daily maxium temperature
Tmin: mean daily minimum temperature
TURFAC: turgor stress factor (0.00–1.00)
|
ACKNOWLEDGMENTS |
|---|
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSAPPENDIXREFERENCES |
|---|
. 2004. Comparison of CERES, WOFOST and SWAP models in simulating soil water content during growing season under different soil conditions. Ecol. Modell.
171
:223–246.[Web of Science]This article has been cited by other articles:
|
|
 |
|
  Z. Yang, G. G. Wilkerson, G. S. Buol, D. T. Bowman, and R. W. Heiniger Estimating Genetic Coefficients for the CSM-CERES-Maize Model in North Carolina Environments Agron. J., August 31, 2009; 101(5): 1276 - 1285. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Crop Science | Vadose Zone Journal | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
