|
Titel |
A Bayesian Belief Network framework to predict SOC stock change: the Veneto region (Italy) case study |
VerfasserIn |
Nicola Dal Ferro, Claire Helen Quinn, Francesco Morari |
Konferenz |
EGU General Assembly 2017
|
Medientyp |
Artikel
|
Sprache |
en
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 19 (2017) |
Datensatznummer |
250149096
|
Publikation (Nr.) |
EGU/EGU2017-13413.pdf |
|
|
|
Zusammenfassung |
A key challenge for soil scientists is predicting agricultural management scenarios that
combine crop productions with high standards of environmental quality. In this context,
reversing the soil organic carbon (SOC) decline in croplands is required for maintaining soil
fertility and contributing to mitigate GHGs emissions. Bayesian belief networks (BBN) are
probabilistic models able to accommodate uncertainty and variability in the predictions of the
impacts of management and environmental changes. By linking multiple qualitative and
quantitative variables in a cause-and-effect relationships, BBNs can be used as a decision
support system at different spatial scales to find best management strategies in the
agroecosystems.
In this work we built a BBN to model SOC dynamics (0-30 cm layer) in the low-lying
plain of Veneto region, north-eastern Italy, and define best practices leading to SOC
accumulation and GHGs (CO2-equivalent) emissions reduction.
Regional pedo-climatic, land use and management information were combined with
experimental and modelled data on soil C dynamics as natural and anthropic key drivers
affecting SOC stock change. Moreover, utility nodes were introduced to determine optimal
decisions for mitigating GHGs emissions from croplands considering also three different
IPCC climate scenarios. The network was finally validated with real field data in terms of
SOC stock change.
Results showed that the BBN was able to model real SOC stock changes, since validation
slightly overestimated SOC reduction (+5%) at the expenses of its accumulation. At regional
level, probability distributions showed 50% of SOC loss, while only 17% of accumulation.
However, the greatest losses (34%) were associated with low reduction rates (100-500 kg C
ha−1 y−1), followed by 33% of stabilized conditions (-100 < SOC < 100 kg ha−1 y−1).
Land use management (especially tillage operations and soil cover) played a primary role to
affect SOC stock change, while climate conditions were only slightly involved in C
regulation within the 0-30 cm layer.
The proposed BBN framework was flexible to perform both field-scale validation and
regional-scale predictions. Moreover, BBN provided guidelines for improved land
management strategies in a perspective of climate change scenarios, although further
validation, including a broader set of experimental data, is needed to strengthen the outcomes
across Veneto region. |
|
|
|
|
|