Carbon and water dynamics in the Alaska Yukon River Basin: Analysis using process-based biogeochemistry models and satellite data.

摘要

Due to the recent warming trend, the arctic regions have experienced significant land cover and hydrology changes which include extended shrub coverage, shrinking water bodies and melting permafrost. All these changes have and will certainly continue to affect the carbon cycles of those regions which have the largest soil organic carbon pools in the world. Of these large soil organic carbon pools, we selected the portion of the Yukon River Basin in the state of Alaska to investigate the dynamic in land cover changes and methane (CH4) emission from 1980s onwards. We also developed a dissolved organic carbon (DOC) transport model to analyze the DOC trends for a watershed in the Yukon River Basin.;We used the newly-released Landsat data to construct land cover changes in the Yukon River Basin from 1986-2005. The NLCD (National Land Cover Dataset) provides the reference data required in the classification. The relative atmosphere correction and the Bayesian inference (BI) methods are used to extract change and non-change areas in the images. Non-change areas and the NLCD dataset are then used to provide the reference land cover types and reflectance required for the classification operations for change-occurred areas. This framework avoids the requirement of collecting atmospheric characterization data and reference data which is difficult for historic images. By analyzing the land cover maps for different periods, we concluded that: (1) forests decrease due to wild fires, (2) the shrinking or growing trends of closed water bodies depend on the permafrost types; the total areas of closed water bodies in the Yukon River Basin generally shrink, and (3) shrub land extends into bare land and grassland.;Methane dynamics in the Yukon River Basin is simulated by coupling the Variable Infiltration Capacity (VIC) model and TOPMODEL. One aspect of our model that is different from most previous methane models is the consideration of the topographic effect on soil moisture distribution. Water table depth, the most important factor in methane modeling, can be updated according to the soil moisture balance and the local topography at each time step. The freezing/thawing phenomena are also expressed in the methane model by incorporating six different scenarios which connect methane production and oxidation processes with soil thermal conditions. Our study shows that the average annual net emission of CH4 from the region is 4.01 Tg CH4 yr-1. El Nino and La Nina events lead to increases and decreases in net CH4 emissions, respectively. We also find that CH4 dynamics are more sensitive to temperature than to precipitation than soil temperature and active layer depth in the Yukon River Basin.;In the fourth chapter, we developed a DOC transport model and presented some preliminary results. This model contains three components: land surface processes, soil heat transport/water infiltration and DOC dynamic modules. The first two components provide soil thermal and moisture information to the DOC module which simulates DOC production, mineralization and sorption/desorption processes. The DOC on each grid is then transported to its neighbor by surface and subsurface flows and eventually joins the river system. From the simulation results, we conclude that the flush effect due to rainfall and snowfall is the main transport process in DOC transport from the terrestrial ecosystem. Soil temperature and root distribution largely determine the DOC production rate and location.