5.5 Uncertainties and Outlook
5.5.1 Uncertainties
These N budgets developed via CHANS could provide essential and vital information for further studies and policy-making. The input-output calculations we used to
estimate fluxes in CHANS represent a tightly integrated system in which variations
in one parameter can cascade through other subsystems and have feedbacks to the
outputs. Though our estimates in this study are based on current best available
information, the accuracy and robustness of our estimates are still limited by the
quality of the data, the applicability of the calculated coefficients for scaling and the
validity of the assumptions made. For example, inputs of N as NBNF, denitrification
to N 2 , and Nr accumulation are well recognized as the most uncertain fluxes. It’s
very hard to well constrain the natural biological N fixation and denitrification on a
regional scale anywhere in the world (Galloway et al. 2008), and Nr accumulation is
calculated as the difference between inputs and outputs and so reflects all of their
uncertainties.
In order to analyse the uncertainty in such studies, we used qualitative and semiquantitative approaches that classify all the N-related parameters, coefficients, and
activity data into several confidence levels and introduced Monte Carlo simulation to
estimate and refine the uncertainties of Nr fluxes. Based on existing statistical or
experimental data and expert judgement, secondary N fluxes or parameters were
derived or assumed for this study. For example, we applied a relatively low (Æ30%
to 50%) confidence rating to irrigation, CBNF, and food/feed import because these
calculated parameters are not well constrained and could easily introduce uncertainty. In contrast, national statistics could usually provide reliable data source of
crop/livestock production, food consumption, and human activity level; their uncertainties are thus believed to be small (within a range of $5%) (NBSC 2016).
Partitioning N surplus into different loss pathways introduces larger uncertainties
due to many affecting factors, e.g. temperature, humidity, management, pH, and
oxygen availability (e.g. Ju et al. 2009); however, field measurement and atmospheric deposition could offer useful constraints on estimates of N flux from
terrestrial subsystems (MEPC 2016; EPA 2011), and as discussed above, they
align well with our analyses. Besides, large-scale remote sensing and satellite
observation could improve our understanding of Nr sources and chemistry by
constraining Nr emission inventories through inverse modelling techniques (Van
Damme et al. 2015).
5 Reactive Nitrogen Budgets in China
105
5.5.1 Uncertainties
These N budgets developed via CHANS could provide essential and vital information for further studies and policy-making. The input-output calculations we used to
estimate fluxes in CHANS represent a tightly integrated system in which variations
in one parameter can cascade through other subsystems and have feedbacks to the
outputs. Though our estimates in this study are based on current best available
information, the accuracy and robustness of our estimates are still limited by the
quality of the data, the applicability of the calculated coefficients for scaling and the
validity of the assumptions made. For example, inputs of N as NBNF, denitrification
to N 2 , and Nr accumulation are well recognized as the most uncertain fluxes. It’s
very hard to well constrain the natural biological N fixation and denitrification on a
regional scale anywhere in the world (Galloway et al. 2008), and Nr accumulation is
calculated as the difference between inputs and outputs and so reflects all of their
uncertainties.
In order to analyse the uncertainty in such studies, we used qualitative and semiquantitative approaches that classify all the N-related parameters, coefficients, and
activity data into several confidence levels and introduced Monte Carlo simulation to
estimate and refine the uncertainties of Nr fluxes. Based on existing statistical or
experimental data and expert judgement, secondary N fluxes or parameters were
derived or assumed for this study. For example, we applied a relatively low (Æ30%
to 50%) confidence rating to irrigation, CBNF, and food/feed import because these
calculated parameters are not well constrained and could easily introduce uncertainty. In contrast, national statistics could usually provide reliable data source of
crop/livestock production, food consumption, and human activity level; their uncertainties are thus believed to be small (within a range of $5%) (NBSC 2016).
Partitioning N surplus into different loss pathways introduces larger uncertainties
due to many affecting factors, e.g. temperature, humidity, management, pH, and
oxygen availability (e.g. Ju et al. 2009); however, field measurement and atmospheric deposition could offer useful constraints on estimates of N flux from
terrestrial subsystems (MEPC 2016; EPA 2011), and as discussed above, they
align well with our analyses. Besides, large-scale remote sensing and satellite
observation could improve our understanding of Nr sources and chemistry by
constraining Nr emission inventories through inverse modelling techniques (Van
Damme et al. 2015).
5 Reactive Nitrogen Budgets in China
105