Saturday, 15 July 2006
170-31

Uncertainty Analysis of Human and Environmental Factors on Nitrogen Flow at Different Spatial Scales.

Sonoko D. Kimura, Graduate School of Bio-Application and Systems Engineering, Tokyo Univ of Agriculture and Technology, Nakacho 2-24-16, Koganei, 184-8588, Japan and Ryusuke Hatano, Graduate School of Agriculture, Hokkaido Univ, Sapporo, 060-8589, Japan.

To create a sustainable Nitrogen (N) management system, N flows in the agroecosystem must be able to be analyzed in a quantitative manner. However, uncertainty accompanying such calculation can make the whole calculation meaningless. This study quantifies the N flow at field and watershed scale using the N balance method. The data source was monitoring, farmer's inquiry, statistics and literature data. The field scale balance was calculated for the major landuses (rice, wheat, soybean, onion, vegetable and grassland). The watershed scale balance was calculated for the farmland, livestock and human subsystems. Special attention was paid on the surplus N. At field scale, farmland surplus N was defined as the difference between total N input to the farmland (chemical fertilizer, manure, biological N2 fixation, deposition and irrigation) and total N output of the farmland (harvested plants, denitrification, NH3 volatilization, N2O emission). At watershed scale, surplus N stands for the difference among the inter-system inflow N, N loss from the total system and food and feed export. This unaccounted N flow equals the change in stock of all subsystems, farmland surplus N in case of farmland subsystem, and excreta disposal N in case of human and livestock subsystems. The uncertainty of the calculated N balance was analyzed by Monte Carlo simulations. Influence of the variability in the data on the uncertainty of the output data was analyzed according to a regression analysis of the simulation runs. At field scale, farmland surplus N was highest for vegetables with 110 kgN ha-1 and lowest for soybean with 6 kgN ha-1. Among the N flows, denitrified N2 and N2O emission had the highest uncertainty with Coefficient of Variance (CV) was between 77 and 128%, except rice. The CV of farmland surplus N for individual landuses ranged from 51 to 630%, and was highest for soybean. Their associated uncertainties were influenced by the variability of biological N2 fixation for soybean, and the amount of fertilizer and N2 volatilization for other land uses. The biggest N flow at the watershed scale was the chemical N fertilizer input from outside the system with 5.4 kgN ha-1 per total watershed-area, followed by the harvested crops (including rice straw) with 4.2 kgN ha-1. The surplus N of the total system was 3.8 kgN ha-1 accounting for 68% of the total N loss. Farmland surplus N was 1.8 kgN ha-1 while disposal N of human and livestock subsystems was 2.0 kgN ha-1. CV was highest in flows from the livestock subsystem, ranging from 23 to 147%. For both farmland and human subsystems the highest uncertainty was found for denitrification (CV=67% and CV=41%, respectively) followed by the farmland surplus N (CV=51%). The uncertainty of livestock and human subsystems were generated by the assumed high variability of the literature data. Since farmland surplus N was the difference of input and output of the farmland subsystem, the uncertainty was caused by the input value with the highest variability and amount. The sensitivity analysis revealed that the variability of the denitrification and amount of applied fertilizer, especially those of onion and rice, influenced mostly the uncertainty of the farmland surplus N. Both scales were highly influenced by the variability in fertilizer input, harvested amount and denitrification. The total N load of the system cannot be significantly reduced unless the variables contributing to the input data are quantified more accurately.

Back to 4.3P Intensification of Agricultural Production Systems and the Environment - Poster
Back to WCSS

Back to The 18th World Congress of Soil Science (July 9-15, 2006)