The impacts of climate change and agricultural activities on water cycling of Northern Eurasia.

摘要

The ecosystems in Northern Eurasia (NE) are important due to their vast land coverage, high rate of observed and projected warming, and the potential feedbacks they can cause on the global climate system. To understand the impacts of climate change and agricultural activities on water cycling in NE, I analysed a variety of datasets and conducted series of studies by applying a combination of modeling, in-situ observations and remote sensing data, uncertainty analysis, and model-data fusion.;Long-term unique in-situ measurements on soil moisture across multiple stations and discharge records at the outflow basins in Northern China (NC) provide us robust evidence to assess the trends of soil moisture and discharge in this region (Chapter 2). NC overlaps with NE and is one of the hot-spots experiencing the most severe water shortage in the world. Declines in soil moisture and stream flow detected via in-situ measurements in the last three decades indicate that water scarcity has been exacerbated. Multiple linear regression results indicate that intensification of agricultural activities including increase in fertilizer use, prevalence of water-expensive crops and cropland expansion appear to have aggravated these declines in this region.;Scarce evapotranspiration (ET) measurements make ET estimation via model a necessary step for better regional-scale water management. Penman--Monteith based algorithms for plant transpiration and soil evaporation were introduced into the Terrestrial Ecosystem Model (TEM) to calculate ET (Chapter 3). I then examined the response of ET and water availability to changing climate and land cover on the Mongolian Plateau during the 21st century. It is shown that use of the Penman--Monteith based algorithms in the TEM substantially improved ET estimation on the Mongolia Plateau. Results show that regional annual ET varies from 188 to 286 mm yr-1 -- with an increasing trend -- across different climate change scenarios during the 21st century. Meanwhile, the differences between precipitation and ET suggest that the available water for human use will not change significantly during the 21st century. In addition, analyses also suggest that climate change is more important than land cover change in determining changes in regional ET.;Improvement in the accuracy of ET estimation by introducing Penman--Monteith based algorithms into the TEM motivated me to further improve the model representation of ET processes. I further modified the TEM to incorporate more detailed ET processes including canopy interception loss, ET (evaporation) from wetland surfaces, wetlands and water bodies, and snow sublimation to examine spatiotemporal variation of ET in NE from 1948 to 2009 (Chapter 4). Those modifications lead to substantial enhancement in the accuracy of estimation of ET and runoff. The consideration of snow sublimation substantially improved the ET estimates and highlighted the importance of snow in the hydrometeorology of NE. The root mean square error of discharge from the six largest watersheds in NE decreased from 527.74 km 3 yr-1 to 126.23 km3 yr-1. Meanwhile, a systematic underestimation of river discharge after 1970 indicates that other water sources or dynamics not considered in the model (e.g., melting glaciers, permafrost thawing and fires) or bias in the precipitation forcing may also be important for the hydrology of the region.;To better understand the possible causes of systematic bias in discharge estimates, I examined the impacts of forcing data uncertainty on ET and runoff estimation in NE by driving the modified TEM with five widely-used forcing data sets (Chapter 5). Estimates of regional ET vary between 263.5-369.3 mm yr-1 during 1979-2008 depending on the choice of forcing data, while the spatial variability of ET appears more consistent. Uncertainties in ET forcing propagate as well to estimates of runoff. Independent of the forcing dataset, the climatic variables that dominate ET temporal variability remain the same among all the five TEM simulated ET products. ET is dominated by air temperature in the north and by precipitation in the south during the growing season, and solar radiation and vapour pressure deficit explain the dynamics of ET for the rest of the year. While the Climate Research Unit (CRU) TS3.1 dataset of the University of East Anglia appears as a better choice of forcing via our assessment, the quality of forcing data remains a major challenge to accurately quantify the regional water balance in NE.