| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Univ. of Arkansas, Dep. of Crop, Soil, and Environ. Sci., 1366 W. Altheimer Drive, Fayetteville, AR 72704
b USDA-ARS, Univ. of Florida, Agron. Physiol. and Genet. Lab., IFAS Bldg. no. 350, 2005 SW 23rd St., P.O. Box 110965, Gainesville, FL 32611-0965
c Agric. Stat. Lab., Agriculture Annex 101, Univ. of Arkansas, Fayetteville, AR 72704
* Corresponding author (lpurcell{at}uark.edu).
Received for publication October 29, 2002.
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
0.20 of drought. In the Southeast, there were approximately 48 d and 121 d at the beginning and end of the growing season, respectively, with P 0.20 of drought. For the Midwest, P of drought was 0.20 throughout the growing season for three of the four sites, and it was concluded that a 50-mm water deficit was not likely to be a production constraint on the deep soils of the Midwest. For the Northern Great Plains, P of drought was >0.20 for more than half of the region's growing season. This meteorological approach for assessing drought may provide insights for drought avoidance in breeding and crop management.
Abbreviations: ASW, available soil water • CVPD, crop vapor pressure deficit • CWD, critical water deficit • DOY, day of year • ea, actual water vapor pressure • emax, maximum water vapor pressure • es, average daily saturated water vapor pressure • Eto, potential evapotranspiration • FAO, Food and Agricultural Organization of the United Nations • Rs, total solar radiation at earth's surface • Tmax, maximum daily temperature • Tmin, minimum daily temperature • VPD, vapor pressure deficit • WUE, transpirational water use efficiency
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Although the ability to tolerate drought and have acceptable yields is limited among cultivars within a species (Serraj and Sinclair, 2002; Purcell and Specht, 2003), there are considerable differences among cultivars that allow them to avoid drought. Drought may be avoided by matching crop phenology with periods during the cropping season when water supply is likely to be more abundant. This approach has been an effective tool for crops grown in monsoonal climates where they are sown near the beginning of the wet season and mature before the dry season (Monteith and Virmani, 1991). Similarly, soybean [Glycine max (L.) Merr.] production in the southern portion of the Mississippi Delta has shifted to early sowing dates combined with early maturing cultivars to avoid droughts, which frequently occur in August (Bowers, 1995; Heatherly, 1999).
Despite the recognition that matching crop phenology with seasonal water supply is an effective means of avoiding drought, tools for evaluating weather patterns for the potential of avoiding drought are not widely used. Several of the most common meteorological methods for assessing drought have limited utility for evaluating the potential for a given crop production system to avoid drought in a particular region. For example, the Palmer Drought Index (Palmer, 1965) uses a soil water balance approach to evaluate drought severity, which is normalized for the departure from climatological norms. Thus, a Palmer Drought Index of -4 for an arid region and for a very humid region would both indicate extreme droughts, but because this index is normalized to long-term values, they reflect very different severities of drought from a cropping perspective. The Standardized Precipitation Index (Hayes et al., 1999) also classifies drought severity for a region on the departure of precipitation from long-term norms. Although the Crop Moisture Index (Palmer, 1968) provides current information on soil moisture availability for crop production, it does not provide a forward-looking assessment of the likelihood of drought for a cropping season.
In contrast to the previously mentioned methods of evaluating drought, Monteith and Virmani (1991) reviewed several techniques that have been used effectively for evaluating the probability of drought occurrence for sites in the semiarid tropics having a monsoonal rainfall distribution. By dividing weekly rainfall totals by the sum of weekly potential evapotranspiration (Eto) values, P indices for drought were developed to determine favorable sowing dates and the length of a crop's growth cycle that would lessen risk of drought (Virmani et al., 1982). Crop breeding programs at ICRISAT (International Crops Research Institute for the Semi-Arid Tropics) have used this approach to develop lines that mature within the growing season as defined by the period with decreased risk of drought (Monteith and Virmani, 1991). We are unaware of a similar systematic approach outside of the semiarid tropics of analyzing weather patterns for crop production purposes.
Drought has been evaluated for specific sites using crop simulation models that predict yield responses under rainfed conditions (e.g., Hammer and Muchow, 1991; Monteith and Virmani, 1991; Sinclair, 1994). Although many of these models have been useful in understanding environmental limitations to crop production, crop simulation models require inputs for specific crops and for specific soils that limit their use outside of the specific situation for which they were developed. A meteorological evaluation of long-term weather data avoids many of the assumptions and crop-specific coefficients inherent in crop simulation models and provides a complementary tool to modeling that has broad applicability.
There were two objectives of this research. The first objective was to develop an analytical and statistical framework that could be used to evaluate the likelihood of drought, from a cropping perspective, from long-term weather data. Further, a goal of this framework is that it should be simple and easily modified as appropriate for specific sites and soil conditions. The second objective was to illustrate the use of this approach with a standard set of conditions for 16 sites in four regions of crop production in the USA.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
|
Estimation of Weather Variables and Water Use Efficiency
Our approach for estimating water deficit required the determination of Eto (mm). Potential evapotranspiration is defined as the theoretical amount of water lost through evapotranspiration from a complete canopy of a cool-season grass actively growing with adequate soil moisture and with specified crop height, conductance, and albedo (Allen et al., 1998). Potential evapotranspiration for each day was calculated using the FAO (Food and Agricultural Organization of the United Nations, Rome, Italy) modified form of the Penman–Monteith equation (Allen et al., 1998; Annandale et al., 2002). Calculations of Eto require values for total solar radiation at the earth's surface (Rs, MJ m-2 d-1), Tmax, Tmin, wind speed (m s-1), and vapor pressure deficit (VPD, kPa). In the absence of Rs, VPD, and wind speed measurements, the FAO procedure gives methods by which these factors may be estimated. We used these methods in our analysis, and these steps are summarized in the following paragraphs.
Daily values for Rs were estimated using the procedure of Hargreaves and Samani (1982) and as subsequently modified by Allen (1997). Extraterrestrial solar radiation (Ra, MJ m-2 d-1) was first calculated at the top of the earth's atmosphere for each DOY based on latitude, longitude, and the solar constant (Allen, 1997; Ball et al., unpublished, 2003). Then, Rs was calculated using the following equation:
 | [1] |
Calculation of VPD was based on estimates of the differences between the average daily saturated water vapor pressure (es) and the actual water vapor pressure (ea). Relative humidity was assumed to be 100% at Tmin, and values for ea were calculated as (Allen et al., 1998):
 |
For the drier locations in the Northern Great Plains, there was a concern that relative humidity might not be 100% at Tmin. We evaluated the importance of this assumption for our driest location (Ogallala, NE) by assuming that ea occurred at 10°C < Tmin and found that there was very little effect on Eto (
1 to 2%, data not shown). Therefore, all Eto calculations were made assuming that the dew point was equal to Tmin.
The maximum saturated water vapor pressure during the day (emax) depended on Tmax:
 | [3] |
The es was assumed to be the mean of ea and emax, and VPD was calculated as the difference between es and ea. The calculation of es as the mean of ea and emax was used to weight VPD calculations to the portion of the day when air temperature was not at Tmax. Although others have found that (emax - ea) x 0.75 is a closer approximation to the es (Tanner and Sinclair, 1983), we have used the mean of ea and emax for the calculation of Eto in accordance with Allen et al. (1998).
In the absence of wind speed measurements, the FAO procedure suggests using a value of 2 m s-1 (Allen et al., 1998). This value is the average of wind speed measurements from over 2000 weather stations around the globe. For hot and dry conditions, Eto increases linearly with wind speed, but for warm and humid conditions, Eto is relatively insensitive to wind speed (Allen et al., 1998).
A 7-d water deficit for each DOY was estimated by calculating the 7-d running sum of Eto and subtracting the 7-d running sum of precipitation. Running sums were calculated using PROC EXPAND of SAS (Statistical Analysis System, Version 8.0, SAS Inst., Cary, NC). Summing the precipitation, Eto, and water deficit variables over 7-d periods was similar to the method used by Virmani et al. (1982) in assessing drought risks in India. The use of running sums also served to smooth data and lessen the impact of abnormally large precipitation events on any given DOY.
Transpirational water use efficiency (WUE, Pa kPa-1) was estimated using the procedure of Tanner and Sinclair (1983). They derived a simple expression of WUE based on characteristics of leaf gas exchange whereby:
 | [4] |
 | [5] |
For each DOY, when the P < 0.05 for Tmin being below 0°C, WUE was calculated using Eq. [4].
Statistical Evaluation of Weather Variables
The temperature limits of the growing season for summer annual crops were first determined for each location as the period when P < 0.05 for Tmin being below 0°C. The determination of P values was made using cumulative distribution functions (Anderson, 1974) for each DOY. An example of cumulative distribution functions of Tmin for Urbana, IL, on DOY 75, 100, 125, and 150 shows that P 0.70 for Tmin being less than 0°C on DOY 75 but that P 0.35 for Tmin being less than 0°C on DOY 100 (Fig. 1A)
.
|
0 were linearly interpolated between consecutive values of Tmin that bracketed 0°C and their associated P values. For each location, we determined the DOY in spring and in fall when P 0.05 for Tmin being less than 0°C, and these dates were considered the temperature limits for a warm-season, summer annual crop. For the example of P < 0.05 of Tmin being less than 0°C for Urbana, IL (Fig. 1B), the limits for the growing season correspond to DOY 118 and 282. Similar P data are presented for Tmin being less than 5°C, and using these limits, the growing season would be approximately 40 d shorter. The appropriate temperature limits would depend on specific crops, but we have chosen 0°C throughout the remainder of the paper. This period is hereafter referred to as the growing season. The water deficit at which a crop will have decreased growth and yield depends primarily on rooting depth (Sinclair and Muchow, 2001; Purcell et al., 2002; Purcell and Specht, 2003). If an approximate rooting depth (mm) is known for a location, then the total amount of available soil water (ASW, mm) to plants may be estimated as the product of the rooting depth and 0.13 (Ratliff et al., 1983; Sinclair et al., 1998). The value of 0.13 was derived from an extensive survey of 401 soils across the USA as the average difference in the volumetric water content of soils at field capacity and the volumetric water content when the soil was very dry and plants were dormant or dead (Ratliff et al., 1983). For soils with a fractional sand content greater than 0.55, the difference between volumetric water content of soils at field capacity and the lower limit for plants was significantly less than 0.13 (Ratliff et al., 1983; Sinclair et al., 1998).
Plants typically begin to undergo water-deficit stress when approximately 0.65 of the ASW has been depleted (Ritchie, 1981; Ray and Sinclair, 1998). Therefore, a critical water deficit (CWD, mm) at which plants begin to undergo water-deficit stress may be estimated as:
 | [6] |
Cumulative distribution functions were generated for 7-d water deficit. Similar to the analysis of Tmin, cumulative P values for 7-d water deficit were made for each DOY. Probabilities of 7-d water deficit exceeding 50 mm were determined for each location for each DOY. Mean values of Tmax, Tmin, Rs, 7-d cumulative precipitation, 7-d cumulative Eto, 7-d water deficit, and WUE were determined for the entire growing season and for the period from DOY 183 to 244 (July and August). It was of particular interest to evaluate weather variables during these months because evaporative demand is greatest at this time, and water deficit would develop quickly. Additionally, mean values of WUE were determined for each day during the growing season and each location.
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
The temperature limits for production of warm-season summer annuals was estimated from cumulative distribution functions, such as that illustrated for Urbana, IL, in Fig. 1B. The DOY in the spring and fall at which the P was 0.05 for Tmin being 0°C or less was used to set the limits on the growing season. Averaged over locations within each region, the length of the growing season was 207 d for the Southeast, 200 d for the Midsouth, 159 d for the Midwest, and 131 d for the Northern Great Plains (Table 2).
|
|
The likelihood of water deficit occurring on each day during the growing season was assessed from cumulative P functions of the running sum of 7-d water deficits being greater than 50 mm (Fig. 2)
. The interpretation of P 0.20 of a 50-mm water deficit on a given DOY is restricted to the water balance for the particular week centered on that DOY and disregards long-term water-deficit cumulation and soil water storage. Virmani et al. (1982) used a similar approach to determine whether or not weekly rainfall totals would meet weekly totals of Eto.
|
0.20 or P > 0.20 of 7-d water deficit exceeding 50 mm (Fig. 2). The specific DOY at which the P of 0.20 was reached was determined by regressing the P levels as they increased (in spring) or decreased (in fall) against DOY. For example, the 0.20 P threshold of 7-d water deficit exceeding 50 mm was from DOY 143 to DOY 189 for Athens, GA, and from DOY 180 to DOY 234 for Fayetteville, AR (Fig. 2). Comparable periods of water deficit for each location are given in Table 2. Obviously, the period identified as likely to undergo a water deficit depends on the selection of the 7-d water-deficit value, the P value, and the period over which water deficit is summed (7 d in this example). The appropriate P level that producers are willing to risk will change depending on various factors including commodity, inputs costs, and potential net returns. The first occurrence in the spring of a 7-d water deficit >50 mm at P = 0.20 ranged from DOY 128 at Tifton, GA, to DOY 180 at Fayetteville, AR (Table 2). In contrast, the probability of a 7-d water deficit exceeding 50 mm did not reach P = 0.20 for three sites in the Midwest. The duration of the period in which P > 0.20 for the 50-mm water deficit ranged from 25 to 55 d for the Southeast region, 54 to 99 d for the Midsouth region, and 78 to 138 d for the Northern Great Plains region.
The period of nonfreezing temperatures before the increased probability of water deficit in spring and after the decreased probability of water deficit in the fall might offer opportunities for drought avoidance for crop production in the southern regions. At the beginning of the growing season, there was, on average, a 48-d period with decreased likelihood of water deficit for the Southeast region and a 62-d period with a decreased likelihood of water deficit for the Midsouth region (Table 2). The early sown, early maturing soybean production system that is used in the Midsouth (Bowers, 1995; Heatherly, 1999) for drought avoidance has capitalized on this period when the probability of water deficit is low.
The later portion of the growing season may also offer opportunities for decreased risk of water deficit. In the Southeast region, the period when P < 0.20 for a 7-d water deficit exceeding 50 mm ended around DOY 182, which offers an 121-d period for crop production with a decreased probability of water deficit (Table 2). For the Midsouth, the same time period ended at approximately DOY 235, offering a 62-d period at the end of the growing season for a decreased probability of water deficit. Before canopy closure and as the crop nears maturity, Eto likely exceeds actual evapotranspiration and, therefore, likely overestimates actual water loss. This may provide a longer period with decreased probability of drought than is indicated in Table 2.
In addition to meteorological effects on water availability for crop production, the ability of crops to utilize water in the production of biomass changes in response to environmental conditions and determines WUE. Because emax increases exponentially with Tmax (Eq. [3]), small changes in Tmax have large effects on WUE. Figure 3A illustrates the response of average WUE to average Tmax for July and August for each of the 16 sites included in our analysis. For sites in the Southeast, Midsouth, and Midwest, WUE decreased linearly as the average Tmax increased. The regression equation predicted that average WUE would decrease by 21% for the 3.9°C increase in average temperature from Stoneville, MS, to Burlington, IA. For rainfed crop production in the Midsouth, low WUE and high probability of water deficit during July and August decrease the likelihood of high yields.
|
During the growing season, WUE varied approximately twofold for each of the representative sites illustrated in Fig. 4 . For Menno, SD, the maximum WUE near the beginning and end of the growing season was 4.3 Pa kPa-1, which was considerably less than the maximum WUE for other sites and reflects the higher CVPD in the Northern Great Plains region. The minimum WUE for Menno occurred on DOY 199 with a value of 2.3 Pa kPa-1, which was lower than the minimum WUE for the other sites. Average WUE values from Tifton and Stuttgart at the beginning of the season were >4.5 and 5.5 Pa kPa-1, respectively. By approximately DOY 140 at Tifton and DOY 160 at Stuttgart, WUE values were near minimum values of 2.6 Pa kPa-1, and they remained near the minimum values until DOY 250. Maximum WUE at the beginning of the season at Urbana was similar to that at Stuttgart with values >5.5 Pa kPa-1, but the minimum WUE at Urbana was greater than the minimum WUE at Tifton and Stuttgart with a value of approximately 2.9 Pa kPa-1. Furthermore, the duration of the minimum WUE at Urbana was approximately 40 d compared with 110 d at Tifton and 90 d at Stuttgart.
|
In contrast to locations in the Southeast, Midsouth, and Northern Great Plains, generally P < 0.20 for a 7-d, 50-mm water deficit for sites in the Midwest (Table 2). This does not indicate, however, that drought does not occur for these Midwestern locations. The analysis presented in Table 2 indicates that for sites in the Midwest, it is unlikely that for any given 7-d period, Eto will exceed rainfall by 50 mm or more. As shown for long-term water-deficit values for Urbana, IL, the period of summation for a water deficit greatly affected its magnitude (Fig. 5) . Average water-deficit values that were summed over 7, 15, 21, and 27 d progressively increased as the period of summation increased. For the 7-d summation, the average water deficit reached a maximum value of 32 mm on DOY 204 whereas a maximum water deficit of 111 mm was reached on DOY 201 for the 27-d summation. The increase in water deficit associated with increasing period of summation indicates that soil water deficit was accumulating over time. The interpretation of a 7-d water deficit, therefore, is restricted for a 7-d period centered on each DOY and should not be interpreted as a cumulative water deficit. Cumulating the water deficit for a 27-d period at Urbana indicated an average water deficit >100 mm occurring during the summer (Fig. 5). Crops with rooting depths <1183 mm would likely be adversely affected by water deficits >100 mm (Eq. [6]).
|
The meteorological approach that we have presented for assessing drought from long-term weather data for warm-season annuals is simple and flexible. The minimum requirement for using this analysis is long-term, daily weather data that include Tmax, Tmin, and rainfall. These data are available (http://lwf.ncdc.noaa.gov/oa/ncdc.html) for thousands of sites throughout the USA.
The assumptions used to estimate water deficit and Eto (Allen et al., 1998) were that the relative humidity was 100% at Tmin and that the average wind speed was 2 m s-1. For arid locations, relative humidity may not be 100% at Tmin, but we found that having the dew point 10°C < Tmin had little effect (<2%) on Eto. For humid regions, Eto is not greatly affected by wind speed (Allen et al., 1998), but for arid regions, Eto accuracy may be improved by wind speed values appropriate for a specific location. A final assumption that may require adjustment is the amount of water in soil that is available to plants. We assumed that the difference in soil volumetric water content at field capacity and when plants had exhausted the water supply was 0.13, which is appropriate for all soils except those having sand contents >0.55 (Ratliff et al., 1983). For sandy soils, a value of approximately 0.08 should be used (Ratliff et al., 1983).
This meteorological analysis provides a broad overview of the risks of drought throughout a cropping season and complements crop simulation models. For more specific risk analysis, this meteorological approach can be modified easily to account for different crop temperature limits and soil characteristics. Periods during the growing season in which weekly Eto exceeded weekly precipitation by a nominal value of 50 mm were readily identified, and this information can provide insights for crop management and breeding programs in mitigating drought.
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  T. M. Seversike, L. C. Purcell, E. Gbur, P. Chen, and R. Scott Radiation Interception and Yield Response to Increased Leaflet Number in Early-Maturing Soybean Genotypes Crop Sci., January 28, 2009; 49(1): 281 - 289. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Popp, J. Edwards, P. Manning, and L. C. Purcell Plant Population Density and Maturity Effects on Profitability of Short-Season Maize Production in the Midsouthern USA Agron. J., May 3, 2006; 98(3): 760 - 765. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. T. Edwards and L. C. Purcell Soybean Yield and Biomass Responses to Increasing Plant Population Among Diverse Maturity Groups: I. Agronomic Characteristics Crop Sci., August 1, 2005; 45(5): 1770 - 1777. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. T. Edwards, L. C. Purcell, and D. E. Karcher Soybean Yield and Biomass Responses to Increasing Plant Population among Diverse Maturity Groups: II. Light Interception and Utilization Crop Sci., August 1, 2005; 45(5): 1778 - 1785. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. T. Edwards, L. C. Purcell, and E. D. Vories Light Interception and Yield Potential of Short-Season Maize (Zea mays L.) Hybrids in the Midsouth Agron. J., January 1, 2005; 97(1): 225 - 234. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Popp, J. Edwards, L. Purcell, and P. Manning Early-Maturity Soybean in a Late-Maturity Environment: Economic Considerations Agron. J., November 1, 2004; 96(6): 1711 - 1718. [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 | |||
0.20 for water deficit exceeded 50 mm.
