Skip to main content
ARS Home » Research » Publications at this Location » Publication #114493
Title: THE ALMANAC MODEL'S SENSITIVITY TO INPUT VARIABLES

Author
item XIE, YUN - BEIJING NORMAL UNIVERSITY
item Kiniry, James
item WILLIAMS, JIMMY - TEXAS AGRIC EXP STATION

Submitted to: Agricultural Systems
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/26/2002
Publication Date: 10/1/2003
Citation: Xie, Y., Kiniry, J.R., Williams, J.R. 2003. The Almanac Model's Sensitivity to Input Variables. Agricultural Systems. 76:1-16.

Interpretive Summary: Computer models for crop growth often require complicated input data sets. Due to the difficulty in developing such input data, it would be valuable to know which variables are most important. This will help model users in showing which variables need accurate values and which ones can be estimated. We used results from hybrid performance trials in eight Texas counties, developing standard data sets of 10-year simulations of maize and sorghum with the ALMANAC (Agricultural Land Management Alternatives with Numerical Assessment Criteria) model. The simulation results were close to the measured county yields with mean errors less than 1.0 Mg ha**-1 in each county. We then analyzed the sensitivity of grain yield to solar radiation, rainfall, soil depth, soil plant available water, and runoff curve number, comparing simulated yields to those with the original, standard data sets. Runoff curve number changes had the greatest impact on simulated maize and sorghum yields for all the counties The next most critical input was rainfall, and then solar radiation for both maize and sorghum, especially for the dryland condition. For irrigated sorghum, solar radiation was the second most critical input instead of rainfall. Maize yields were more sensitive than sorghum yields for all variables except for solar radiation. Researchers using these models need accurate values for runoff curve number and rainfall. Many models use a USDA curve number approach to represent soil water redistribution, so it will be important to have accurate curve numbers, rainfall, and soil depth to realistically simulate yields.

Technical Abstract: Crop models often require extensive input data sets to realistically simulate crop growth. Development of such input data sets can be difficult for some model users. The objective of this study was to evaluate the importance of variables in input data sets for crop modeling. Based on published hybrid performance trials in eight Texas counties, we developed standard data sets of 10-year simulations of maize and sorghum for these eight counties with the ALMANAC (Agricultural Land Management Alternatives with Numerical Assessment Criteria) model. The simulation results were close to the measured county yields with bias values and root mean square errors less than 1.0 Mg ha**-1 in each county. We then analyzed the sensitivity of grain yield to solar radiation, rainfall, soil depth, soil plant available water, and runoff curve number, comparing simulated yields to those with the original, standard data sets. Runoff curve number changes had the greatest impact on simulated maize and sorghum yields for all the counties. The next most critical input was rainfall, and then solar radiation for both maize and sorghum, especially for the dryland condition. For irrigated sorghum, solar radiation was the second most critical input instead of rainfall. The degree of sensitivity of yield to all variables for maize was larger than for sorghum except for solar radiation. Many models use a USDA curve number approach to represent soil water redistribution, so it will be important to have accurate curve numbers, rainfall, and soil depth to realistically simulate yields.