Enhancing ALMANAC for simulating switchgrass biomass production and macronutrient removal
A complete system for switchgrass (Panicum virgatum L.) feedstock and biofuel production has not yet been developed in the U.S. Distinct management practices may be required for different environments. Additionally, accurate decision–making tools to predict switchgrass biomass yield and nutrient removal are necessary for widespread switchgrass adoption by the bioenergy industry. ALMANAC (Agricultural Land Management Alternative with Numerical Assessment Criteria) is a process-oriented crop simulator that can predict growth of various crops; however, it lacks accuracy for predicting switchgrass biomass and outputs for nutrient removal. The overall objective of this study was to improve ALMANAC for supporting decisions on timing of harvest and amount of fertilization. The specific objectives were to (i) calibrate and validate ALMANAC for simulating switchgrass biomass in Arkansas, (ii) enhance the nitrogen (N) and phosphorus (P) uptake logic for simulating their removal in harvested biomass, and (iii) enhance frost-damage logic for improving simulation of post-season biomass losses. ALMANAC was calibrated by modifying the parameters: potential heat units, leaf area index decline rate, growing season fraction for LAI decline, first/second points of the leaf area development curve, plant population, and biomass-energy ratio. Those parameters were calibrated based on empirical research conducted at Fayetteville, AR. The calibrated values for Arkansas conditions were, respectively, 1922, 0.25, 0.42, 10.40/90.88, 50 plants m-2, and 29 dg MJ-1. Furthermore, the calibrated model was validated based on comparisons between observed and simulated yields for two locations located in contrasting ecoregions: Ozark Highlands (Decatur, AR) and Bottomland and Terrace (Pinetree, AR). In addition, a mathematical model for simulating changes in daily N concentration and removal was developed by adding new logic to the nitrogen (N) constraints-factor algorithm. This new logic incorporated N removal losses during peak yield and was developed based on data collected at Fayetteville, AR. The modified model was able to accurately simulate the N variables within 95% prediction intervals at Fayetteville, AR; however, these variables were not successfully validated for the two Arkansas ecoregions. Furthermore, no trend across years was elucidated when analyzing P data collected at Fayetteville, AR; therefore no new logic could be developed for P concentration and removal. In addition, the P logic in ALMANAC for the P constrain-factor calculation could not be adapted to simulate daily P concentration and removal. Finally, suggestions were proposed based on empirical data for improving the simulation of ALMANAC biomass losses by the frost-damage algorithm: (i) frost-damage algorithm may commence after 5 consecutive days of daily minimum temperature of -1°C or below, rather than starting at the first day. (ii) A minimum amount of aboveground biomass attached to the plant may remain after killing frost, which are mainly the lodging resistant stems; however, the amount of biomass resistant to shattering may differ for different stands.