B22B-01 INVITED
Ecohydrology Controls on Feedbacks Between Northern Wetlands and Climate Systems
Boreal regions contain large stocks of soil carbon, mostly in poorly drained areas where peat accumulating wetlands have served as a long-term sink for atmospheric carbon. It is unknown whether northern wetlands globally will continue to represent a net sink for atmospheric carbon dioxide, or whether changes in the Earth's climate will cause these ecosystems to release stored carbon back to the atmosphere. Such feedbacks between northern wetlands and regional or global climate systems will depend on interactions between wetland vegetation, peat properties, and hydrology. Within many wetlands, hydrology is the dominant control on plant community structure and decomposition rates. In turn, both plant and microbial activity determine the quantity and quality of litter, which govern the nature of peat accumulation and soil properties critical to hydrology. Here, we will present research from our field and modeling studies investigating the effects of drought, permafrost degradation, and wildfire on vegetation, carbon cycling, and hydrological processes in northern wetlands at multiple spatial scales. At local scales, our findings show that interactions among vegetation, soil, and hydrology can lead to unexpected and often complex changes in soil environments, with potential ‘carbon surprises'. For example, in a nonpermafrost peatland, we found that sustained drought led to peat subsidence that limited the development of oxic surface peat layers and inhibited ecosystem respiration. The decrease in porosity and water content with drought reduced seasonal ice thaw, which also likely limited microbial activity. In contrast, peatlands underlain by permafrost are increasingly experiencing thermokarst and soil flooding with increasing active layer depth. Changes in moss productivity post-thaw led to increased rates of organic matter accumulation, with very different hydrologic and soil properties than peat accumulated in permafrost settings. In addition to local experimentation and gradient studies, we are conducting modeling studies to understand how responses of wetlands may effect the climate system. We argue that carbon, water, and energy fluxes in northern wetlands can respond quickly to changes in climate. However, a comprehensive understanding of whether the main feedback of wetland responses to climate change is through atmospheric carbon concentrations, or whether there also are direct feedbacks to local and regional climate through water and energy exchange pathways requires integration of observational and modeling studies at a variety of spatial scales.
B22B-02
On the Hydrological Impact of Future Climate-Induced Vegetation Changes
This study uses the NCAR CLM3 and CLM3-DGVM to investigate how predicted vegetation changes influences the surface hydrologic responses to elevated CO2 and attendant climate changes projected by 8 GCMs under the SRESA1B scenario. Pre-industrial control (PICNTRL) and SRESA1B scenarios of equilibrium vegetation structure (leaf area index (LAI) and vegetation coverage) are simulated by the dynamic global vegetation model CLM3-DGVM. For climate change forcing from each GCM, comparisons are made among three surface hydrology simulations using CLM3.0 driven with different combinations of climate forcing and potential natural vegetation in order to separate the effect of structural vegetation feedback from the combined influence of climate and CO2 changes. With the exception of the HadCM scenario, all other GCM scenarios broadly agree on spatial patterns of structural vegetation feedbacks on surface temperature and the surface water budget, although the response of soil moisture varies considerably among the GCM scenarios especially in the tropics. With the HadCM excluded, averages over the seven GCM scenarios indicate that the CO2-induced warming in winter is stronger than in summer in the northern mid- and high-latitudes, and structural vegetation feedback enhances the winter warming (due to masking of snow by increased LAI) and reduces the summer warming (due to increased evapotranspiration with increased LAI) over a large portion of these regions; the global hydrological cycle is expected to accelerate in a warmer climate, and structural vegetation feedback increases evapotranspiration under future climate, thereby further accelerating the hydrological cycle. Measured by evapotranspiration flux, this hydrological acceleration by structural vegetation feedback is at a magnitude comparable to that due to CO2 and climate changes. The impact of vegetation changes corresponding to the HadCM-projected climate changes is markedly different, being either more extreme or in a different direction than those corresponding to the other GCMs examined.
B22B-03
Importance of Carbon-Nitrogen Interactions on Ecohydrology During the 21st Century
The effects of various aspects of global change (e.g., climate change, changes in the chemistry of the atmosphere, such as CO2 and O3, and land-use change) on the hydrologic cycle are becoming an important research area. For example, with respect to increases in atmospheric CO2, recent work supports the contention that there will be reduced evapotranspiration and therefore increased water availability in a CO2 -rich world. Our new research on this topic suggests that various aspects of global change combine to affect hydrology in terrestrial ecosystems, and that it is particularly important to include carbon-nitrogen interactions in these studies. We have developed a new version of the Terrestrial Ecosystems Model (TEM) to examine the effects of carbon-nitrogen interactions on the water cycle. This new version includes explicit modeling of the stomatal exchange of CO2 and water, as well as a new approach to carbon and nitrogen allocation in plants. Using this new version of TEM, we have performed a range of site-level and regional experiments across the eastern United States. For example, using data from Harvard Forest, MA, a predominantly deciduous mixed forest, we ran two transient simulations from 1700 to 2100, with and without considering nitrogen limitations on plant productivity. In both of these simulations, we allowed CO2 to double by 2100, but maintained present- day climate. In these two experiments, we found that runoff increased through the 21st century in response to elevated atmospheric CO2. Without nitrogen limitation on plant productivity, the increase in runoff was 12%. However, with nitrogen limitation on plant productivity, the increase in runoff nearly doubled to 21%. This difference in runoff response was the result of a stronger transpiration reduction associated with a smaller increase in photosynthesis in the nitrogen limitation case. In this presentation we will discuss a set of site-level and regional experiments that explore the effects of carbon- nitrogen interactions on the water cycle in the context of different combinations of global changes including climate changes, changes in nitrogen deposition, and changes in tropospheric ozone. Since the carbon and water cycles are tightly coupled, future considerations of ecohydrology must take into account carbon-nitrogen interactions and other multiple stresses that strongly influence the carbon cycle.
B22B-04 INVITED
An Ecohydrological Perspective of Drought-Triggered Vegetation Die-off and Land-Surface Changes
Ecohydrological linkages between vegetation and the water budget can be particularly pronounced in semiarid systems and in response to extreme drought events, which are projected to become more frequent and intense with progressing climate change. We provide an overview of how under extreme drought conditions, a prolonged period with low values of soil water can trigger basic land surface changes through tree mortality and of the potential land surface feedbacks related to such mortality. Drought-induced tree mortality might first be triggered at ecotones between vegetation types, at the drier end of the distribution of the less drought-sensitive species, but under more extreme conditions can span elevational gradients and regions. Warmer temperatures might amplify such trends. Previous observations and new experiments are providing insights into the drought threshold that triggers tree mortality. Changes in tree cover resulting from mortality can result in large changes in near-ground solar radiation because the relationship between near-ground solar radiation and amount of tree cover includes non-linear relationships; these relationships are dependent, of course, on topography and aspect. Loss of tree cover via drought-induced mortality can be compensated for by post-drought increases in herbaceous cover. Research from more mesic ecosystems highlights that tree mortality in response to drought depends not only on the magnitude and duration of the drought, but also on the sequence of conditions prior to the drought, including baseline and flooding conditions. Vegetation die-off is likely to be a major type of response to climate change that impacts land-surface conditions and a broad range of ecological and hydrological characteristics. Many important uncertainties remain relative to ecohydrological responses and climate change.
B22B-05
Examining Severe Drought-Induced Vegetation Change and its Influence on Water Resources
A "global-change-type" drought that occurred in the southwestern U.S. from 2000 to 2003, accompanied by increased temperatures and bark beetle infestations, induced large-scale woodland overstory mortality, the consequent redistribution of water, radiation, and nutrients, as well as modification of the ecosystem phenology. Our objectives in this research are to examine these vegetation changes in detail and to determine whether they translated to changes in hydrological processes. We chose the Rio Ojo Caliente, a subbasin of the Rio Grande, as a study site since a significant portion of the woodland ecosystem (piñon-juniper) was affected. Examining a remotely-sensed vegetation index (1-km AVHRR NDVI from 1989 to 2006), there is an increasing trend in the mean NDVI from 1989 to 1998 (pre-drought period), a decreasing trend from 1999 to 2003 (drought period), and a dramatic increasing trend from 2004 to 2006 (post-drought period) in which the mean NDVI rebounds to pre- drought magnitudes. Streamflow records from 1932 to 2006 show the watershed to be primarily spring snowmelt-driven, although monsoonal summer precipitation also plays a significant role. We compare the temporal variability in the streamflow to the NDVI, including the mean, anomalies from the mean, and seasonally- based duration curves, and find significant correlations (correlation coefficient ρ = -0.61) between the streamflow and NDVI at approximately a three-month lag (NDVI lagging streamflow). In analyzing the three phases of the drought, the correlation is slightly stronger during the pre-drought (ρ = -0.64) and drought (ρ = -0.65) periods, yet markedly stronger during the post-drought period (ρ = -0.74). This suggests that the coupling between vegetation water use and streamflow is tighter after the drought. This may be attributable to the reduction in the less-responsive overstory (pinñon mortality) and increase in the more-responsive understory (grasses and shrubs exploiting newly available resources). Temporal patterns in gauge-based precipitation (frozen and unfrozen) and air temperature, and spatial-temporal patterns in PRISM precipitation, air temperature, and a soil moisture index are also compared to the NDVI. While the vegetation composition was altered to a great degree in the Rio Ojo Caliente Basin, the system rapidly recovered both photosynthetically and hydrologically during the post-drought wet period, although the dynamic between vegetation water use and streamflow was slightly altered. The aim of this research is to explore the consequences of a severe drought married with elevated temperatures on vegetation and water resources. As the intensity and frequency of droughts are expected to increase in the southwestern U.S. with rising temperatures (IPCC 2007), this research contributes to our knowledge of ecosystem and hydrologic response to the changing climate.
B22B-06 INVITED
Large-scale evapotranspiration derived from MODIS GPP compared to MODIS evapotranspiration and 'observations'
Coupling of carbon and water exchanges between leaf surface and surrounding air by stomatal conductance is well known for decades and led to the concept of water-use efficiency (WUE). It is the recent development of flux towers, however, which allows the estimation of the equivalent property at ecosystem scale. Here, we present global maps of WUE extrapolated from flux tower sites by using soil texture maps and remotely sensed leaf area index. Such spatial details of WUE are used for estimating gross primary production (GPP) from evapotranspiration (ET), or vice versa for estimating ET from remotely sensed GPP. The latter case will be presented in detail, and comparisons to ET directly derived from MODIS data and to ET approximated by precipitation, interception and runoff are shown. Interestingly, these (partly) independent approaches lead to similar results in some areas but discrepancies exist, too. Such analysis allows inferences about related uncertainties in each of the extrapolation approaches.
B22B-07
Impacts of Land Cover Change and Increasing CO2 Concentrations on Amazon Basin Hydroclimatology: High-resolution Results From the Ocean-Land-Atmosphere Model (OLAM)
It has recently been projected that agricultural expansion will eliminate 40 percent of Amazon forests during the next 50 years. Previous studies have suggested that such dramatic land cover change can significantly alter the regional and possibly global hydroclimate. Furthermore, projections of vegetation response to increasing atmospheric carbon dioxide concentrations and the resulting feedbacks to the hydroclimate are sensitive to water use efficiency, which varies among different land cover types. To assess the influence of Amazonian land cover change on hydroclimate, general circulation models (GCMs) and regional meteorological models have been used. However, both approaches have important limitations: the grid resolution of the GCMs is coarser than the scale of actual land cover change and too coarse to resolve mesoscale circulations, while mesoscale models typically obtain their requisite lateral boundary conditions from global simulations not accounting for land cover change. In this study, we present a new model, the Ocean Land Atmosphere Model (OLAM), which alleviates the resolution problem and eliminates the problem of boundary conditions. OLAM is a global Earth System Model employing a state-of-the-art grid structure which enabled us to use an atmospheric resolution typical of a mesoscale model over South America while using a coarser resolution typical of other GCMs throughout the rest of the world. Vegetation is treated either by the LEAF-3 biophysical scheme or by the Ecosystem Demography (ED) model, a dynamic vegetation model. In this configuration, OLAM reasonably matched observed patterns of precipitation and radiation on monthly to decadal time scales, with particularly good agreement in South America. We investigated how regional and global hydrological cycles were altered in OLAM when the model was forced with current projections of land cover change in Amazon basin and with increased CO2 concentrations. We found that, because of mesoscale feedbacks, the impact of land cover change on basin-wide Amazon precipitation was small relative to the previous projections of coarser-resolution modeling studies. However, changes in land cover had important impacts on sub-basin scales with the eastern Amazon experiencing increased precipitation and strongly increased runoff, and the western Amazon experiencing decreased precipitation and only slightly increased runoff. Similarly, we found that the impact of increased CO2 concentrations on precipitation varied regionally, with both negative and positive impacts possible but relatively small on the basin-average.
B22B-08
Ecosystem-scale Modelling of the Response of the Amazon Rainforest to Drought
Some climate models predict that there may be a decrease in rainfall over the Amazon Basin by up to 50% over the next century. The uncertainties associated with this change in precipitation have been the focus of several recent studies, but less attention has been paid to the uncertainty in modelling the response of vegetation to drought stress. This study aims to compare the effect of drought in four dynamic global vegetation models (DGVMs), with regards to ecosystem carbon balance and shifts in plant functional type composition. Key objectives were to quantify the uncertainty due to model structures and parameters, and to highlight particular areas for model improvement. The study was carried out in three parts: 1) assessment of DGVMs in simulating changes in ecosystem processes along a rainfall gradient, 2) long-term future runs under a range of drought scenarios, including a factorial analysis of the effects of drought, other climate change and CO2 fertilisation and 3) a Monte Carlo based sensitivity analysis to identify key parameters influencing the uncertainty of the response to drought. Considerable differences were found in the responses of the models to drought stress, with some models, such as MOSES-TRIFFID and HYLAND predicting very little loss of carbon, even at rainfall reductions of up to 60%. This resilience was much lower, however, when climatic changes other than rainfall were included in the simulations. The reduction in rainfall needed to induce a transition in vegetation type was also considered, but this was made difficult by the inability of some models to simulate changes in plant functional type competition dynamics under drought. Other areas highlighted for future improvement include the ability of trees to access water stored in deeper soil layers and the effect of reduced soil moisture on nutrient uptake.