| Summary: | In this dissertation, I systematically investigated the impacts of current model-data uncertainties on biogeophysical-biogeochemical interactions in a terrestrial modeling framework. To achieve this objective, I applied a land surface model, specifically the Integrated Science Assessment Model (ISAM), at spatial scales ranging from flux tower sites to the global land surface in order to explore how the modeling uncertainties translate from site to regional to global scales. I studied the impacts from three key factors that can largely contribute of modeling uncertainties in terrestrial processes, such as differences due to: (1) meteorological forcing datasets, (2) boundary conditions such as land-cover and land-use change datasets, and (3) representation of biogeophysical/biogeochemical processes in land surface models. A brief introduction containing the overall objectives and content of this dissertation has been provided in Chapter 1. For the current body of work, extensive model development was carried out to extend the capabilities of ISAM to be a coupled biogeophysical-biogeochemical land surface model (LSM). Therefore, in Chapters 2 and 3 I have first described in detail the completed model development and evaluation aspects. Subsequently In Chapter 2, I used the ISAM to quantify the causes and extents of biases in terrestrial gross primary production (GPP) due to the use of meteorological reanalysis datasets. I first calibrated the model using meteorology and eddy covariance data from 25 flux tower sites ranging from the tropics to the northern high-latitudes, and thereafter repeated the site simulations using two reanalysis datasets: NCEP/NCAR and CRUNCEP. The results show that at most sites, the reanalysis-driven GPP bias is significantly positive with respect to the observed meteorology driven simulations. Notably, the absolute GPP bias is highest at the tropical evergreen tree sites, averaging up to ~0.45 kgC/m2/yr across sites (~15% of site-level GPP). At the northern mid/high-latitude broadleaf deciduous and the needleleaf evergreen tree sites, the corresponding annual GPP biases are up to 20%. For the non-tree sites, average annual biases of up to ~20—30% occur within savanna, grassland and shrubland vegetation types. At the tree sites the biases in shortwave radiation and humidity strongly influence the GPP biases, while the non-tree sites are more affected by biases in factors controlling water stress (precipitation, humidity, air temperature). In this chapter, I also discussed the influence of seasonal patterns of meteorological biases on GPP, and finally the impacts of the results on GPP simulations for the entire global land surface. In a broader context, my results can have important consequences on other terrestrial ecosystem fluxes (e.g., net primary production, net ecosystem production, energy/water fluxes) and reservoirs (e.g., soil carbon stocks). In continuation from Chapter 2, I used the ISAM in Chapter 3 to extend the analysis for biases in ecosystem energy and water fluxes arising due to the use of the meteorological reanalysis of the NCEP/NCAR and the CRUNCEP. In comparison with the model simulations using observed meteorology from sites, the reanalysis-driven simulations produced several systematic biases in net radiation (Rn), latent heat (LE) and sensible heat (H) fluxes. These include: (1) persistently positive tropical/subtropical biases in Rn using the NCEP/NCAR, and gradually transitioning to negative Rn biases in the higher latitudes; (2) large positive H biases in the tropics/subtropics using the NCEP/NCAR; (3) negative LE biases using the NCEP/NCAR above 40oN; (4) high tropical LE using the CRUNCEP in comparison with observationally derived global estimates; and (5) flux-partitioning biases from canopy and ground components. Across vegetation types, I investigated the role of the meteorological drivers (shortwave and longwave radiation, atmospheric humidity, temperature, precipitation) and their seasonal biases in controlling these reanalysis-driven uncertainties. At the global scale, my site-level analysis explains several model-data differences in the LE and H fluxes when compared with observationally derived global estimates of these fluxes. Using these results, I discussed the implications of site-level model calibration on subsequent regional/global applications to study energy and hydrological processes. The flux partitioning biases presented in this study have potential implications on the couplings amongst terrestrial carbon, energy and water fluxes, and for the calibration of land-atmosphere parameterizations that are dependent on LE/H partitioning. Previous studies demonstrated that land-use and land-cover change (LULCC) can substantially influence terrestrial energy and water fluxes in climate models through the modification of land surface albedo, roughness, and the biosphere-atmosphere coupling of moisture, heat and carbon. However, current datasets and model representation of LULCC remain notably uncertain, thereby impacting the simulation of these fluxes. In addition, these fluxes are also strongly coupled to input meteorology (i.e., climate) to the land surface – the differences in which can introduce another component of uncertainty in models. In Chapter 4, I decomposed their relative impacts in a land surface model using multiple LULCC and climate datasets. My results show that uncertainties in model-derived annual-scale energy/water fluxes (latent heat, sensible heat, runoff) can be dominated by existing disagreements amongst climate datasets, with much smaller impacts from current LULCC uncertainties. These results appear to be robust for global and latitudinally averaged flux budgets, as well as in many regions that contain the largest differences amongst the used LULCC datasets. I also show that relatively small uncertainties in meteorology can produce large differences in terrestrial energy/water fluxes. Henceforth I argue that, in coupled climate models the accuracy of computed meteorology can potentially be a limiting factor in successfully assessing future changes in soil hydrology and water availability. Improving the LULCC datasets themselves remain important for multitude of terrestrial modeling applications, such as – biogeochemistry, ecosystems’ goods/services and regional socioeconomics. In Chapter 5, I used the ISAM to: (1) evaluate the influence of recent improvements in modeling soil/snow physics on historical permafrost area, temperature, and degradation rates in the northern high-latitudes (45—90oN), and (2) compare the relative impact of these improvements with modeling uncertainties due to climate, and land-use and land-cover change datasets. Specifically, by incorporating deep soils, soil organic carbon (SOC) driven soil properties, wind compaction of snow density, and depth-hoar formation in snow, the simulated near-surface permafrost area increased by 9.2 million-km2 (from 2.9 to 12.3) for current atmospheric conditions. In comparison, permafrost area using two reanalysis datasets (CRUNCEP and NCEP/NCAR) differed by up to 2.3 million-km2 in response to their mean annual air temperatures differing by ~0.5oC, and with lower impacts from current land cover uncertainties (up to 0.7 million-km2). Further, I show that incorporating all these soil/snow processes can lead to strongly increased (net) stability in permafrost temperatures, highlighting the importance of including them in future modeling studies of permafrost degradation. Analyses of relative contributions from soil/snow processes show that inclusion of deep soils lead to the largest increases in permafrost area, followed by contributions due to SOC and wind compaction of snow, while inclusion of depth hoar lead to slight decreases in permafrost. However, in the context of assessing the impacts of these processes on permafrost soil biogeochemistry, the results show the dominating influence of near-surface soil/snow processes (such as wind compaction of snow) on vegetation root zone temperatures with minimal impacts from deep soils. Finally, given that only a few of the currently available land surface models include the representation of the specific soil/snow processes considered in this study, I discussed their importance in modeling future permafrost physical characteristics, and for permafrost SOC stocks. In Chapter 6, I continued the analysis of Chapter 5 to study the impacts of the aforementioned four important soil/snow related thermal processes on permafrost SOC stocks. By including these processes, the modeled northern high-latitude permafrost carbon stocks increased from 313 to 445 gigatonnes of carbon (GtC = 1015 gC) in the top 1 m of soil, in better agreement with observational estimates of 495 GtC from the Northern Circumpolar Soil Carbon Database (NCSCD). In the model, net impact of these processes reduced the root zone soil temperature and liquid water content, and generally increased the litter input to the soil. Consequently, driven by stronger temperature and water stresses on soil/litter decomposition processes, the modeled SOC density increased throughout the circumpolar permafrost soils. In comparison to the NCSCD, these represent improvements at regional as well as for biome-aggregated (e.g., tundra, boreal, other grass/shrub) scales. However, given current uncertainties in SOC datasets, I also evaluated their implications in assessing modeling improvements by comparison with data from two other global sources: the Harmonized World Soil Database (HWSD) and the Global Soil Data Task (GSDT), and an Alaskan SOC dataset. My analysis highlights the need for evaluation using multiple datasets, and at site scales, to robustly delineate model estimation of SOC across various regions. Nonetheless, while continued improvements are required in the treatment of model biogeochemistry and in datasets, my study quantifies the importance of key soil/snow biogeophysical processes as drivers of soil biogeochemical processes. Hence, many earth system models that underestimate permafrost SOC and do not represent the processes may also similarly benefit by including them in their land surface schemes. Finally, in Chapter 7, I provided an overall summary and the future direction of research presented in this dissertation.
|
|---|