PubAg

Main content area

Evaluation of the performance of the EPIC model for yield and biomass simulation under conservation systems in Cambodia

Author:
Le, Kieu N., Jeong, Jaehak, Reyes, Manuel R., Jha, Manoj K., Gassman, Philip W., Doro, Luca, Hok, Lyda, Boulakia, Stéphane
Source:
Agricultural systems 2018 v.166 pp. 90-100
ISSN:
0308-521X
Subject:
Crotalaria juncea, Environmental Policy Integrated Climate Model, Urochloa ruziziensis, agricultural conservation practice, bananas, biomass, carbon sequestration, cassava, conservation tillage, corn, cover crops, crop yield, cropping systems, highlands, rice, soil organic carbon, soybeans, statistics, Cambodia
Abstract:
Limited field studies have been performed to evaluate the impacts of conservation agriculture (CA) on crop yields and soil organic carbon sequestration in tropical conditions. In this study, we used the Environmental Policy Integrated Climate (EPIC) model to evaluate the impact of CA and conservation tillage (CT) on crop yields in tropical conditions for unique upland rice, soybean, and cassava cropping systems in Cambodia. New crop parameters were developed and tested for cassava, sesame, banana, sunn hemp, stylo, and congo grass. The results show that EPIC successfully replicated crop yields of soybean, upland rice, maize, and cassava based on R2 statistics ranging from 0.62 to 0.88 and percent bias (PBIAS) values ≤10%. However, it cannot be concluded that the model can accurately capture the biomass for all the individual crops due to limitations in the observed biomass data. The cassava and maize biomass were simulated satisfactorily, resulting in R2 values of 0.81 and 0.75, respectively. However, the computed PBIAS for the biomass estimates of the two crops were >25%. In contrast, the predicted rice and soybean biomass met PBIAS criteria (≤23%) but resulted in weak R2 statistics of ≤0.20, indicating inaccurate replications of the measured biomass. Similarly, the cover crop mean biomass and PBIAS statistics were acceptable but the R2 values were not. Overall, the model tended to overestimate the measured crop biomass. No significant difference was found in the simulated crop yields between the CA and CT treatments. However, the predicted rice and soybean results reflect an increased yield trend over time for the CA treatments, versus no discernible trend for the cassava and maize yields.
Agid:
6132522