Calibration of ALMANAC model and APSIM model for simulating growth of WW-B.Dahl old world bluestem [Bothriochla bladhii (Retz.) S.T. Blake]



Journal Title

Journal ISSN

Volume Title



‘WW-B.Dahl’ old world bluestem [Bothriochloa bladhii (Retz.) S.T. Blake] (Bdahl), a warm-season perennial grass, has been established in the Southern High Plains (SHP) of Texas as a forage crop partly in response to declining supply of irrigation water from the Ogallala Aquifer. Bdahl was introduced to SHP because of its drought tolerance and ability to maintain biomass yield and nutritional value for ruminant animal production. Previous studies at Texas Tech University have investigated various aspects of Bdahl production such as response to water input levels, forage quality for livestock, integration of Bdahl into cotton systems, and insect ecology in pursuit of deep comprehension of Bdahl production capabilities. This project applied new technology including digital image analysis (DIA) and crop simulation models to further describe the potential of Bdahl to support profitable livestock production in a water-limited environment. The objectives were: 1) determine the potential for using canopy cover from DIA to estimate leaf area index (LAI), percentage photosynthetically active radiation interception (PARI) and forage aboveground biomass (AGB) of Bdahl; 2) calibrate and validate ALMANAC growth models for Bdahl AGB production; and 3) correlate the simulation of Bdahl growth under different levels of water supply by ALMANAC to actual plant biomass production using previously published data; 4) compare APSIM and ALMANAC on the power of accurately predicting AGB production. The DIA programs ImageJ and Canopeo were compared for ground cover accuracy, their ability to estimate canopy functions such as LAI and PARI, and their time savings for vegetation analysis relative to a manual method. The two applications did not differ in their measurements of ground cover and their relationships with LAI and PARI. When using their capabilities for batch analysis of large numbers of images, they reduced by 96% the time required for manual methods. The prediction regressions for canopy functions showed that 1) ground cover from DIA predicted PARI very closely to a 1:1 linear relationship; and 2) LAI and AGB accumulation were exponentially related to DIA-detected ground cover. Weaknesses were that the two-dimensional imagery of ground cover had limited power to predict three-dimensional LAI and AGB when ground cover exceeded 60%, LAI exceeded 1.8, and AGB exceeded 1500 kg DM ha-1. DIA-detected ground cover is a poor estimator of AGB at high levels of LAI owing to excessive overlapping of leaves. In order to calibrate and validate the simulation of Bdahl AGB by ALMANAC, default parameter settings were modified based on measurements of leaf area development and growth relationship to PAR interception, and on fractional biomass allocation to roots. This allowed optimization of the following parameters: DLAP1, DLAP2, PHU, DLAI, and WA (abbreviation defines in chapter 3). The result was close agreement between simulated and observed values of LAI and AGB in 2014, the year of calibration, and in the 2 yr of validation (2015-2016), as tested with the regression comparisons with 1:1 relationships and Willmott Agreement Index (d-stat). The model was further validated for a range of water supply treatments by comparing simulation outputs to published AGB data from the irrigation trial of previous study in 2001 by Dr. Dirk Philipp. Good agreements were achieved by modifying the model’s automatic irrigation function to a manual irrigation function with evenly spaced irrigation events, without having to recalibrate the plant growth functions. As a preliminary assessment of the utility of APSIM, the generic model of AgPasture within APSIM framework was modified for initial AGB, initial root depth, efficiency of irrigation, and initial soil nitrogen content. After inputting weather and on-site soil textures, APSIM was able to simulate AGB for 2014 and 2015 reasonably well with d-stat values greater than 0.94 in first period, but with a consistent underestimation of AGB relative to simulated ALMANAC values and observed values, especially in second periods. Therefore, APSIM would require more thorough calibration before consistently reliable simulations could be useful for research and management purposes. Calibration of ALMANAC for accurate simulation of Bdahl forage production was achieved by calibrating in one year and validating in two other years, plus validating with data from a previous experiment with a wide range of water supplies. Results indicate excellent potential of using ALMANAC to generate meaningful forage yield responses to weather and management input scenarios expected in the Southern High Plains on soils similar to Pullman clay loam. Based on user preference and skill level, both ALAMANC and APSIM are good candidates as viable methods to predict forage production; however, further calibration of APSIM is required for reliable accuracy. The calibrations includes but not limit to using accurate soil inputs with more updated model than AgPasture, manually inputting plant phenology parameters, and testing the adaptations on all years of 2001-2003, and 2014-2016.



‘WW-B.Dahl’ old world bluestem, Growth model