Date of Graduation
Bachelor of Science in Biological Engineering
Biological and Agricultural Engineering
Runkle, Dr. Benjamin R. K.
Haggard, Dr. Brian
Committee Member/Second Reader
Naithani, Dr. Kusum
Committee Member/Third Reader
Moreno-Garcia, Dr. Beatriz
As the global population increases and food security is recognized as a critical issue, crop growth prediction models help ensure the sustainability of reliable food sources. Using a prediction model based on temperature and simple, measurable field parameters, e.g., Leaf Area Index (LAI) or Canopy Height (Hcan), may allow farmers and others to intervene mid-season with fertilizer, irrigation, or other inputs to obtain a better harvest.
This study aims to create a general model that could predict LAI and Hcan values for numerous rice varieties using Growing Degree Days (GDD) as the time scale. The models use data collected during the 2018-2020 growing seasons for 16 fields in east-central Arkansas. After comparing model performance indicators (coefficient of determination (R2), root mean square error (RMSE), percent bias (pbias), percent difference, and Akaike Information Criterion values (AIC) of quadratic and sigmoid regression forms, a sigmoid regression with GDD as its time scale was chosen as the best functional form for the datasets provided. The sigmoid with GDD was chosen due to its higher R2 values and lower AIC values (LAI: R2= 0.82, AIC= 14.97; Hcan: R2= 0.88, AIC= 83.01), compared to the other models. The data was then divided into calibration and validation datasets, accounting for field and rice variety differences. The calibration dataset created a generalizable model, and the validation dataset ensured the model could be applied successfully over varying field conditions (LAI: R2= 0.78, RMSE= 1.15 m2m-2; Hcan: R2= 0.85, RMSE= 13.7 cm).
Three cultivar-specific models for the CL-XL745, XP753, and Gemini214 CL cultivars were created and compared to the general model. Overall, there were only minor differences between each model, with the statistics values remaining within a tight range between the general and cultivar-specific models. Further work is being pursued on the benefits of dividing the data based on field cultivar. The uncertainties due to less representative calibration datasets within the cultivar-specific models make the general model the preferred choice for a future wide-scale application for farmers to make field management decisions concerning improving yield and general field management practices within Arkansas.
Growing Degree Days, GDD, LAI, Canopy Height, Crop Modeling
Kuhn, E. J. (2022). Modeling Leaf Area Index and Canopy Height Using Growing Degree Days. Biological and Agricultural Engineering Undergraduate Honors Theses Retrieved from https://scholarworks.uark.edu/baeguht/86