The odyssey of a national soil carbon model
The Ministry for the Environment (MfE) established the Land Use and Carbon Analysis System (LUCAS) for reporting on New Zealand’s land use, land-use change, and forestry (LULUCF) sector in the national greenhouse gas inventory, submitted each year to the United Nations Framework Convention on Climate Change (UNFCCC). As part of LUCAS, the Soil Carbon Monitoring System (the ‘Soil CMS’) was developed to extrapolate national soil carbon stocks and estimate and report the effect of land-use change on the soil carbon pool to meet the UNFCCC reporting guidelines. Under those guidelines, countries can rely on simple, prescribed methods using global default values (called a Tier 1 approach), use country-specific data with Tier 1 methods (a Tier 2 approach), or implement more specific and sophisticated methods, such as process-based models (a Tier 3 approach). The tier structure is hierarchical, with higher tiers implying increased accuracy of the method and/or emissions factors and other parameters used in the estimation of the emissions and removal. Following the IPCC good practice guidance, countries are expected to use appropriate available data and methods at the highest tier possible depending on national circumstances, especially for reporting on important (“key”) categories of greenhouse gas emissions and removals. In addition, the principle of continuous improvement encourages each Party to refine and improve its approach through time.
1990s: The Soil CMS Model emerges
From the outset, New Zealand sought to report on soil carbon using a Tier 2 approach, setting it on an odyssey of meeting both scientific research and international policy challenges. After the UN Conference on Environment and Development, the “Earth Summit” held in Rio de Janiero in 1992, Landcare Research Scientist Kevin Tate recognized the need to pull together New Zealand’s soil data for the future reporting requirements, and MfE initiated the development of the Soil CMS in 1996. The main focus of this fi rst version was the then major land use change in New Zealand of afforestation of former pastures. The underlying principle was to calculate the difference between assumed equilibrium soil C stocks where land-use change occurs, applying the IPCC default of linear change to a new equilibrium over a 20-year period (the approximate duration of exotic forests being harvested). With that aim in mind, the soil CMS model was developed based on biophysical principles using existing national soil data sets to estimate SOC taking into account site variables such as climate, topography, soil type, and land use. The soil CMS model was further developed with additional data, and a move from a linear regression model to a general least squares fitting procedure and a correction for spatial autocorrelation.
2010: International review
The soil CMS met resistance from an “external review team” (ERT) of New Zealand’s 2010 submission (for the 2008 reporting year), its first annual submission for Commitment Period 1 (2008–2012) of the Kyoto Protocol. The ERT commended New Zealand for undertaking the Tier 2 approach but questioned its statistical validity, especially detecting the effect of particular land-use transitions on soil carbon stock changes. The ERT encouraged New Zealand to re-examine the methodological approach, as well as collect more data for land-use categories under-represented in the calibration data set.
2015: A refined Soil CMS Model
The LUCAS programme sought to meet ERT expectations by collecting additional data and recalibrating the model. Although the first attempts to add new data sets (for cropland soils) were helpful in terms of expanding the data set, the model improvements were deemed insufficient to meet the ERT critique. So, New Zealand reverted to using a Tier 1 approach for estimating and reporting soil carbon emissions and removals for the 2010 and 2011 reporting years (submitted in 2012 and 2013, respectively). Further model development and recalibration involved adding a wetland soils data set to the calibration data and a thorough investigation into approaches used to model SOC and determine the significance of land-use transitions, resulting in an improved version of the soil CMS that could be used for a Tier 2 approach. Thus, in 2014, New Zealand returned to using a Tier 2 approach for soils for the 2012 reporting year. The ERT commended New Zealand for this improvement in its 2014 review, but also noted the ongoing need to verify SOC stocks for land-use categories currently lacking data.
Since the 2014 return to the Tier 2 approach, a further refinement has been made to the soil CMS model. After a data collection campaign in 2014 (Fig. 1), soil data from the post-1989 forests were added to the model, and the updated land-use coefficients were used in the 2015 submission (for the 2013 reporting year).
Although New Zealand has found its way back to reporting soil carbon at a Tier 2 level, further improvements could be made to modelling national soil carbon for greenhouse gas inventories, even after these 20 years of research and development (thus making it a longer quest than the fabled 20-year sojourn of Odysseus!). Despite these keen efforts made to develop the soil CMS model, it is still incomplete. As noted by the ERT, certain land uses are not well characterised due to inadequate, or a total lack of, data. Subcategories of grassland are ill-defined, which obscure any signifi cant effect on SOC of land-use change among them. Moreover, the model’s relatively large residual standard error indicates uncaptured effects on SOC. Evolving understanding, of course, also makes it necessary to reconsider underlying assumptions, such as equilibrium states. All this effort has not been for nought: the soil CMS model is currently the best estimate of national carbon stocks for New Zealand with existing datasets, which can be used to assess the effect of land-use change. For the purposes of greenhouse gas inventories, it is a sophisticated approach to Tier 2, which allows an unbiased estimate of national carbon stocks using country-specific data. Rather, the unfinished nature of this particular story has more to do with the complexity of the subject matter, improved (or shifting) understanding of the system from research findings, and funding limitations on data gathering and model development, all set against the backdrop of global environmental change and socioeconomic practices altering conditions on the ground. Odyssey indeed.