A Bayesian Ensemble Data Assimilation to Constrain Model Parameters and Land Use Carbon Emissions

Abstract

Abstract.A dynamic global vegetation model (DGVM) is applied in a probabilistic framework and benchmarking system to constrain uncertain model parameters by observations and to quantify carbon emissions from land-use and land-cover change (LULCC).Processes featured in DGVMs include parameters which are prone to substantial uncertainty. To cope with these uncertainties Latin Hypercube Sampling (LHS) is used to create a 1000-member perturbed parameter ensemble, which is then evaluated with 5 a diverse set of global and spatio-temporaly resolved observational constraints. We discuss the performance of the constrained ensemble and use it to formulate a new best-guess version of the model (LPX-Bern v1.4). The observationally constrained ensemble is used to investigate historical emissions due to LULCC (E LUC ) and their sensitivity to model parametrization.We find a global E LUC estimate of 158 (108, 211) PgC (median and 90% confidence interval) between 1800 and 2016. We compare E LUC to other estimates both globally and regionally. Spatial patterns are investigated and estimates of E LUC of the 10 ten countries with the largest contribution to the flux over the historical period are reported. We consider model versions with and without additional land-use processes (shifting cultivation and wood harvest) and find that the difference in global E LUC is on the same order of magnitude as parameter induced uncertainty and in some cases could potentially even be offset with appropriate parameter choice.

Effect of Land Use and Land Cover Change in Context of Growth Enhancements in the United States Since 1700: Net Source or Sink?

“…The losses in carbon due to land use conversion (E 1.2 ) are largest in the eastern United States and West Coast with increases in the interior of the country driven by agricultural fertilization (Figure ). The net effect (Table ) shows larger emissions earlier in the 20th century than after 1950, consistent with results of Lienert and Joos (), but other studies (Houghton & Nassikas, ) show a sink after the 1950s from land use and land cover change. The regrowth effect, which is effect of land use and land cover change on NEP itself, is 29 TgC/year, or 1.78 PgC since 1950.…”

Global soil nitrous oxide emissions since the preindustrial era estimated by an ensemble of terrestrial biosphere models: Magnitude, attribution, and uncertainty

“…Finally, 7 of the 10 participating models were selected to estimate N 2 O emissions from both cropland and natural soils and to quantify the relative contributions of each driving factor. These seven models are as follows: (1) the Dynamic Land Ecosystem Model (DLEM; Tian et al, ; Xu et al, ), (2) Lund‐Potsdam‐Jena‐General Ecosystem Simulator (LPJ‐GUESS; Olin et al, ; Xu‐Ri & Prentice, ), (3) Land Processes and eXchanges model‐Bern (LPX‐Bern v1.4; Lienert & Joos, ; Stocker et al, ; Xu‐Ri & Prentice, ), (4) O‐CN (Zaehle et al, ), (5) Organising Carbon and Hydrology In Dynamic Ecosystems (ORCHIDEE, Vuichard et al ), (6) Organising Carbon and Hydrology In Dynamic Ecosystems‐Carbon Nitrogen Phosphorus (ORCHIDEE‐CNP; Goll et al, ), and (7) Vegetation Integrated Simulator for Trace gases (VISIT, Inatomi, Ito, Ishijima, & Murayama, ; Ito & Inatomi, ; See more model information in Supporting Information Table S1). Five models (DLEM, LPJ‐GUESS, ORCHIDEE, ORCHIDEE‐CNP, and VISIT) considered the effects of manure use in cropland and ran all the seven simulation experiments (S0–S6), while the other two models (LPX‐Bern and O‐CN) did not include manure effects and ran six model experiments (all except SE1).…”

Rainfall manipulation experiments as simulated by terrestrial biosphere models: Where do we stand?

“…For all sites, we conducted simulations using 10 terrestrial biosphere models: CABLE r54482.0 (Wang et al, ), DLEM v2.0 (Tian et al, ), JULES v5.2 (Clark et al, ), JSBACH v3.2 (Kaminski et al, ; Mauritsen et al, ), LPX v1.4 (Lienert & Joos, ), ORCHIDEE rev5150 (Krinner et al, ), ORCHIDEE MICT rev5308 (Guimberteau et al, ), ORCHIDEE CNP rev4520 (Goll et al, ), T&C v1.0 (Fatichi et al, ; Paschalis, Katul, Fatichi, Palmroth, & Way, ) and TECO v2.0 (Huang et al, ). All models include a land surface scheme, a hydrological component and a dynamic vegetation module.…”