PROJECT RESULTS

Estimation of Ground Cover and Phenological Development of Canola From Weather Data

Applicant: 

Dr. Carl F. Shaykewich
Department of Soil Science
University of Manitoba
Winnipeg, Manitoba  R3T 2N2  Canada

Table of Contents:

• Background and Objectives
• Procedure and Project Activities
• Results and Discussion
• Conclusion

ARDI Project:

#98-201

Project Status:

Completed November 1, 2001

Background and Objectives:

Since 1995, canola (Brassica napus L.) has been the second most successful cash crop in Canada. The 1999 growing season had a record area seeded of 5,598,700 hectares, declining slightly to 4,894,600 hectares in 2000 (Statistics Canada 2001). Although canola is an important contributor to the Canadian economy, little research has been conducted at the field level to determine how crop phenological stage and ground cover respond to weather variables such as temperature. The fungal infection Sclerotinia (Sclerotinia sclerotiorum (Lib.) de Bary) is a serious disease of canola in western Canada. The current model for predicting Sclerotinia disease for the Canadian Prairies predicts the risk of infection based on crop stage and top-zone soil moisture estimates from a Canola Phenology and Water-Use Model (Raddatz 1993, Raddatz et al. 1996). Crop stage and fractional leaf area are estimated using accumulated growing degree-days above 5^oC and utilized in the estimation of top-zone soil moisture.

The accurate estimation of ground cover is an important component of determining sclerotinia risk. The amount of canopy cover influences the relative humidity of the environment of the disease organism. If there is little or no canopy cover, air near the soil surface can mix with the air above, thereby lowering the relative humidity near the soil surface, regardless of the surface soil moisture content. If there is complete ground cover, air near the soil surface is prevented from mixing with the air above, and thus relative humidity in the canopy is strongly influenced by surface soil moisture (Oke 1987). Thus, knowledge of the fractional leaf area is vital in accurately assessing disease risk.

The overall aim of this project was to improve the current method for estimating the risk of Sclerotinia infection on the Canadian Prairies. Presently, Sclerotinia risk is based on soil moisture and growth stage information derived from a Canola Phenology and Water-Use Model (Raddatz 1993). However, this model is limited because it uses a simple heat unit to estimate phenological stage and percent ground cover. The accumulated growing degree-days (GDD) above 5^oC is a crude predictor of canola phenology because it assumes a linear plant development-temperature relationship. The specific project objectives were:

1. Develop a heat unit specific for canola. A non-linear heat unit system such as the P-Days system used to predict potato phenology will be adapted to reflect the cardinal temperatures of canola and determine if this improves estimates of canola phenology.
2. Determine the relationship between fractional leaf area and a heat unit developed specifically for canola.
3. Evaluate the accuracy of top-zone soil moisture estimated from the Canola Phenology and Water-Use Model (Raddatz 1993; Raddatz et al. 1996).

Procedure and Project Activities:

Field Site Locations

Two varieties of canola (Brassica napus L. cv. Quantum and 2273) were seeded at eight test sites within Agro-Manitoba (in collaboration with Aventis Canada) during the 1999 and 2000 growing season. The site locations were: 1) Brandon (1999), 2) Carman 1999, 3) Carman 2000, 4) Franklin 1999, 5) High Bluff, 6) Roblin 1999, 7) Roblin 2000 and 8) Stonewall (1999). Although 12 sites were originally proposed as test locations, difficulties with weather, seeding, and management reduced this number to 8.

Design

This project consisted of weekly field observations throughout the 1999 and 2000 growing season to determine the crop developmental stage, the fractional leaf area, and top-zone soil moisture (10-cm depth). Daily maximum and minimum temperatures and precipitation were obtained from the nearest Environment Canada Weather station.

Data Analysis

Fractional Leaf Area:  The fractional leaf area was assessed using an innovative technique using an overhead picture of the plot. This data was analysed using an image processing package called ASSESS for Windows (formerly ImageX32 for Windows) (Lamari 2002).

Top-zone Soil Moisture:  Modelled top-zone soil moisture was compared to observed soil moisture values and was further compared between sites that had on-site and off-site precipitation data.

Heat Units:  Heat units for each observed developmental stage were calculated using the maximum and minimum daily temperatures from the nearest Environment Canada weather station. Coefficients of variation were calculated for calendar days, growing degree-days above 5^oC, and several modified P-Day equations for both cultivars (Brassica napus L. cv. 2273 and Quantum).  The modified P-days equation will be abbreviated using the following notation:  P-Days_{(base temperature, optimum temperature, maximum temperature).}

Other Sources of Funding

1. Natural Sciences and Engineering Research Council of Canada Scholarship for Janna Wilson
2. Department of Soil Science: office space and laboratory facilities (in kind)
3. Aventis Canada: seeded and maintenance of plots (in kind)
4. Parkland Crop Diversification Centre in Roblin, MB: seeded and maintenance of plots in Roblin (in kind)
5. Summer Career Placement (Federal Government Student Employment Program)

Results and Discussion:

Heat Units

A paired t-test showed no significant difference between the two cultivars using P-Days_(5,17,30), except for the Brandon site. The Brandon site was eliminated because the data from the site was unreliable. This was apparent early in the growing season as it was seeded late (June 19, 2000) and had severe flea beetle damage, which completely obscured emergence counts and caused abnormal phenological development.

Calendar days from planting were a better estimator of phenological development than GDD above 5^oC from planting, i.e. the coefficient of variation for GDD above 5^oC was greater for each stage of development and for the averaged coefficient of variation (Table 1). Calculation from 50% emergence also lowered the average coefficient of variation for all of the thermal time systems tested. The heat units which showed the lowest coefficients of variation for the individual cultivars were recalculated for the combined cultivar phenological development data. The P-Day_{(5,17 30)} had the lowest average coefficient of variation from planting for the critical stages 3.1 to 5.4. Although there were other combinations that had lower average coefficients of variation when calculated from 50% emergence, the lowest method from planting was chosen because it would be more applicable to the Raddatz (1993) canola model. The following procedure patterned after the P-Days_(7,21,30) for potatoes (Sands et al. 1979) for calculating the P-Days_(5,17,30) is recommended:

P-days_(5,17,30) are calculated from the following equation:

(1)

Where:

The accumulation of heat is calculated from a function of temperature, P(T), where the temperatures T₁ through T₄ are used to define the value of P by the following formula:

When:

When:

When:

When:
Where: k is a scale factor set to a value of 10

Average P-Days_(5,17,30) required for several stages of development are given in Table 2.

Fractional Leaf Area

Comparison of observed and modeled values showed that fractional leaf area was overestimated by the current method. A regression equation relating observed and modeled values was calculated. The R² value of 0.61 and root mean square error (RMSE) of 0.15 suggest that there is room for improvement in this model.

Table 1.  Coefficients of variation for seven heat unit systems for Brassica napus L. cv. 2273 and Quantum.

Table 2.  Mean P-Days_(5,17,30) for several stages of development.

Since there is a lack of empirical data estimating the fractional leaf area (L_A) from temperature, observed fractional leaf area (up to stage 5.2) was plotted against growing degree-days above 5^oC to determine the nature of the actual relationship from stages 1.0-5.2 (Figure 2). The L_A for the senescence phase of the crop, that is stages 5.3 to 5.5 was not investigated in this study.

Figure 3 shows observed ground cover plotted against the P-Days_(5,17,30). The data was analyzed as two separate populations. The inflection point of 300 P-Days_(5,17,30) was chosen based on a visual assessment of the data. A linear portion from 0 to 400 P-Days_(5,17,30) had an improved linear relationship over that for GDD above 5^oC.

Soil Moisture

Modeled soil moisture was better estimated when on-site precipitation data was available (Figure 4). Roblin 1999 and 2000 (on-site precipitation data) had an R² of 0.83 and a RMSE of 2.32 mm while sites with off-site precipitation had a substantially lower R² of 0.61 and a RMSE of 4.45 mm. Thus, precipitation is a key component for modeling top-zone soil moisture in the canola phenology and water-use model. The spatial variability of rainfall and the poor estimation of soil moisture at sites with off-site precipitation data indicates that rainfall is the most important parameter (R. L. Raddatz, personal communication, Environment Canada, Winnipeg, MB). In order to more adequately assess the accuracy of the Canola Phenology and Water-Use Model, on site precipitation data should be included. Thus, the third objective to assess modeled top-zone soil moisture was only partially achieved. 

Conclusions:

Canola, Brassica napus L. is an important cash crop on the Canadian Prairies. Sclerotinia stem rot is the most devastating disease of canola and afflicts all canola-growing areas of Canada. Prairie canola producers expend $260 million annually as a result of yield loss and management techniques requiring expensive fungicide applications (G.B.H. Ash, personal communication, Canadian Wheat Board, Winnipeg, MB.). In terms of farm gate value, canola was the number one crop in 1999 in Manitoba, with production estimates of $401.3 million (Manitoba Agriculture 1999). As a result of this study, improvements can be made to the Canola Phenology and Water-use Model (Raddatz 1993). This will improve disease prediction of sclerotinia and facilitate the timely and efficient application of fungicide thereby reducing losses due to sclerotinia stem rot and the inappropriate use of fungicide. In addition, a better understanding of the factors that affect the rate of phenological development of canola allows the incorporation of this basic agronomic knowledge into other agrometeorological models to help improve the yields of an extremely important cash crop on the Canadian Prairies.

Acknowledgements:

This project was made possible due to the funding from the Governments of Manitoba and Canada through the Manitoba Canada Agri-Food Research and Development Initiative (ARDI).

References:

Lamari, L. 2002. Assess for Windows: Image Analysis for disease quantification. APS – Press ed. (In Press) (Formerly ImageX32 for Windows, Version 1.0)

Manitoba Agriculture. 1999. Manitoba Agricultural Yearbook, 1999. 203 pp.

Oke, T. R. 1987. Boundary Layer Climates. 2^nd ed. Routledge, London. 435 pp