| 1. |
- Bolinder, Martin, et al.
(author)
-
New calibration of the ICBM model & analysis of soil organic carbon concentration from Swedish soil monitoring programs
- 2018
-
Reports (other academic/artistic)abstract
- We are using the Introductory Carbon Balance Model (ICBM) within the national inventory reporting (NIR) system for estimating changes in soil organic carbon (SOC) stocks in Swedish arable land for mineral soils. The first version of ICBM was developing from a long-term experiment (LTE) in Ultuna Sweden, with a time-series of data from 1956 to 1991. The model is initialized using data from a soil database and information from Swedish soil monitoring programs (SMPs). For that purpose, it specifically uses data from the first inventory of the Swedish Environmental Protection Agency (SEPA) SMP conducted in the early 1990s. Since the 1990s, the SEPA SMP has conducted two further inventories in a 10-year cycles, with the last inventory completed in 2017. In complement to this SEPA SMP, the Swedish Board of Agriculture (SJV) was completing a spatially more detailed SMP in 2011 and 2012. In an earlier study, when the third inventory of SEPA was still not fully completed, we showed that SOC concentrations are increasing in Swedish arable mineral soils during the past two decades. Simulations with the current version of ICBM are not reproducing this long-term increasing trend.The main purposes of this project were to (i) calibrate a second version of ICBM by incorporating new knowledge gained from LTEs, (ii) evaluate the trends in SOC concentrations with the complete inventories (I, II and III) of SEPA, (iii) compare results from SEPAs last inventory (III) with that of SJV SMP conducted at approximately the same time period. A Bayesian approach was selected for calibrating the second version of ICBM, using an updated time series of data from Ultuna ending in 2017, and by also including experimental data (1996-2017) from a new sister experiment at Lanna in southern Sweden. For adequately comparing the SEPA and SJV SMPs, we were correcting data from the SJV samples so that we could express all results for SOC concentrations on a basis equivalent to dry combustion. Our analysis on the complete SEPA inventories are generally confirming our previous assessment. However, a more detailed exploration using data only from inventories II and III having identical sampling coordinates, clearly indicates that the increases in SOC concentrations were less in the last decade than in the previous one. Furthermore, when comparing SOC concentrations at different spatial scales, the positive trend in SOC concentrations is much more obvious on the scale of Swedish agricultural production regions (eight regions). Indeed, at the scale of the twenty-one Swedish counties, there are few significant differences and the increasing trend is unclear for many of the counties. This is reflecting difficulties related to the density of sampling points in the SEPA SMP; each of them roughly represents 1500 ha of arable land. Besides, only two decades or even only one decade since SEPA was regarding inventory II as a restart of their SMP, remain a short time period in detecting changes in SOC and highlight the importance of maintaining continuous SMPs.We found the comparisons between the SJV and SEPA SMPs useful but further work is necessary to refine the linkage. The overall changes in parameter values related to SOC dynamics for the calibrated second version of ICBM, will give simulations towards higher SOC stocks, compared to the current version. The main parameter explaining this differences is the increased contribution of below ground (i.e., roots) annual C inputs to soil contributing to the formation of more stable SOC. The approach we used here also offers the advantage of making a multi-site calibration and we could in the future, further refine parameter values by including other Swedish LTEs. Additionally, it allows us to include an estimate for uncertainty in the predictions of SOC stocks, which we could also develop for inclusion in the NIR system.
|
|
