field locations and growing seasons ( Bell, 1950 , 1953 ; Wood, 1961 ). GDD models have been useful for predicting harvest dates of highbush blueberry cultivars in Michigan ( Carlson and Hancock, 1991 ) and cumulative flowering of ‘rabbiteye
Scott N. White, Nathan S. Boyd, and Rene C. Van Acker
Domingo R. Loero and Kent D. Kobayashi
Nine years of historical yield, meteorological, and soil data were input into a soil water balance simulation model to generate a daily soil water status value. The values for the number of days and millimeters of deficit (duration and magnitude) were grouped into trimesters and used to estimate yield. The greatest frequency of days with plant moisture stress occurred during the January–March and the October–December periods. The greatest magnitude of stress occurred during the January–March period. Annual coffee yields were best estimated by the model that incorporated variables for the previous year including, April–June deficit magnitude duration, July–September deficit magnitude duration, and the previous year's yield. Model testing with data from nine cultivars over an 8-year period showed that the model estimated yields with a mean error of 17%. The use of this model permitted yield estimation 2 months before anthesis and 8 months before the start of harvest.
Carmen Feller and Matthias Fink
To reduce nitrogen (N) losses from vegetable fields, fertilizer recommendations should be adjusted according to the large range in yield and thus in N uptake of vegetable crops. Therefore, a model was used to predict total N uptake based on expected yield. The model has been validated successfully in a series of studies for Brussels sprouts (Brassica oleracea L. var. gemmifera), white cabbage (Brassica oleracea L. var. capitata) and kohlrabi (Brassica oleracea L. var. gongylodes). The objective of this study was to validate the model for table beet (Beta vulgaris L. var. conditiva), a crop with a considerable variability in N uptake, which is caused by a large potential range of selecting sowing dates, plant densities and cultivars. Field experiments were carried out over two years. Fifty-five combinations of N fertilizer levels, plant densities, cultivars and sowing dates were tested. Plants were sampled at 2- or 3-week intervals, and fresh matter, dry matter and N content of leaves and roots were measured. Crop specific model parameters for table beets were determined from independent data. The model wverestimated N uptake for N-limiting conditions, but for optimally fertilized table beets measured and estimated N uptake showed a close correlation (R 2 = 0.93) when total yield was used as an input parameter for the model. Although the error of estimation (35 kg·ha-1) was considerable, studies with other vegetable crops using the model found the error even higher if other tools, such as look-up tables, were used for predicting N uptake.
Jay Frick, Cyrille Precetti, and Cary A. Mitchell
An artificial neural network (NN) and a statistical regression model were developed to predict canopy photosynthetic rates (Pn) for `Waldman's Green' leaf lettuce (Latuca sativa L.). All data used to develop and test the models were collected for crop stands grown hydroponically and under controlled-environment conditions. In the NN and regression models, canopy Pn was predicted as a function of three independent variables: shootzone CO2 concentration (600 to 1500 mmol·mol-1), photosynthetic photon flux (PPF) (600 to 1100 μmol·m-2·s-1), and canopy age (10 to 20 days after planting). The models were used to determine the combinations of CO2 and PPF setpoints required each day to maintain maximum canopy Pn. The statistical model (a third-order polynomial) predicted Pn more accurately than the simple NN (a three-layer, fully connected net). Over an 11-day validation period, average percent difference between predicted and actual Pn was 12.3% and 24.6% for the statistical and NN models, respectively. Both models lost considerable accuracy when used to determine relatively long-range Pn predictions (≥6 days into the future).
Michele R. Warmund and Joan Krumme
The time of rest completion of `Apache', `Arapaho', `Chickasaw', `Darrow', `Kiowa', `Navaho', and `Shawnee' blackberry (Rubus subgenus Rubus Watson) buds was compared and various models for estimating chilling were evaluated. `Kiowa' and `Arapaho' buds had the shortest rest periods, while those for `Shawnee', `Navaho', and `Chickasaw' buds were intermediate. `Apache' and `Darrow' buds had the longest rest periods. The model that accounted for the variation in percent budbreak among cultivars and temperatures during two dormant periods had the following two components: 1) a chilling inception temperature of –2.2 °C and 2) weighted chilling hours that accumulated after the chilling inception temperature. The chilling hours in this model were weighted as follows: 0 to 9.1 °C = 1; 9.2 to 12.4 °C = 0.5; 12.5 to 15.9 °C = 0; 16 to 18 °C = –0.5; >18 °C = –1. This study also elucidated that a blackberry model with a chilling inception temperature of –2.2 °C estimated chilling more accurately than one with chilling inception just after the maximum negative accumulation of chill units as used in the Utah chilling model. Also, temperatures between 0 and 2.4 °C must be weighted more heavily in a blackberry model than in the Utah peach model to accurately estimate chilling and rest completion.
G.R. Panta and D.S. NeSmith
Eight muskmelon (Cucumis melo reticulatus L.) cultivars were selected to test whether a model could be developed to estimate leaf area across cultivars. Regression analyses of leaf area vs. leaf width and length revealed several models that could be used for estimating the area of individual muskmelon leaves. A linear model using leaf width squared was the best overall, yielding the equation A = 3.3 + 0.63 (W2), where A is area of an individual leaf lamina (square centimeter) and W is leaf width (centimeter) at the widest point perpendicular to the leaf midrib. Forcing the intercept through the origin did not significantly alter prediction capability and resulted in a simple model of the form A = 0.64 (W2) that was applicable to all eight cultivars.
Maynard E. Bates
A simple plant growth model has been developed based on the analysis of growth curves of lettuce and spinach in numerous controlled environment experiments. The model incorporates elements for genetic potential, plant spacing, photosynthetic photon flux, photoperiod, environment, and morphology. Predicted parameters are relative growth rate, mean plant weight, and plant growth efficiency. Prediction may be on an hourly or daily basis. Examples drawn from data on various species and cultivars will be presented.
Katharine B. Perry and Todd C. Wehner
A heat unit model developed in a previous study was compared to the standard method (average number of days to harvest) for ability to predict harvest date in cucumber (Cucumis sativus L.). Processing and fresh-market cucumbers were evaluated in 3 years (1984 through 1986), three seasons (spring, summer, and fall), and three North Carolina locations. The model predicted harvest date significantly better than the standard method for processing, but not for fresh-market cucumbers.
Vanessa Drouot, Eric H. Simonne, and James B. Witt
An irrigation scheduling model represented by 12.7 DAT * 0.5 * ASW = D(DAT – 1) + [Ep(DAT) * CF(DAT) – R – I] was tested in central Alabama for Spring-grown bell pepper (Capsicum annuum L.). In the model, DAT (days after transplanting) is crop age; effective root depth is 12.7 DAT with a maximum of 250 mm; usable water (mm3·mm–3) is 0.5 ASW; deficit on the previous day is D(DAT–1); evapotranspiration is pan evaporation [Ep(DAT)] times a crop factor value [CF(DAT) = 0.15 + 0.018 DAT – 0.0001 DAT * DAT]; rainfall (R) and irrigation (I) are in mm. The model called for 13 irrigations between 17 and 85 DAT. Under the current N recommendation rate for bell pepper (112 kg/ha), marketable yield increased quadratically from 36% to 148% of the model rate. Highest marketable yields occurred near the model rate. Under a N rate of 170 kg/ha, yields increased linearly. These results suggests that the model provided adequate moisture to maximize bell pepper marketable yields under the recommended N rate.
James E. Faust and Royal D. Heins
Leaf unfolding rate (LUR) was determined for `Utah' African violet plants grown in growth chambers under 20 combinations of temperature and photosynthetic photon flus (PPF). A nonlinear model was used to predict LUR as a function of shoot temperature and daily integrated PPF. The maximum predicted LUR was 0.27 leaves/day, which occurred at 25C and a daily integrated PPF of 10 mol/m2 per day. The optimum temperature for leaf unfolding decreased to 23C, and the maximum rate decreased to 0.18 leaves/day as the daily integrated PPF decreased from 10 to 1 mol/m2 per day. A greenhouse experiment using 12 combinations of air temperature and daily integrated PPF was conducted to validate the LUR model. Plant temperatures used in the model predicted leaf development more accurately than did air temperatures, but using average hourly temperature data was no more accurate than using average daily temperature data.