Dissertations / Theses on the topic 'Climate change model'

Unforced internal variability abounds in the climate system and often confounds the identification of climate change due to external forcings. Given that greenhouse gas concentrations are projected to increase for the foreseeable future, separating forced climate change from internal variability is a key concern with important implications. Here, we leverage a 40-member ensemble, the Community Earth System Model Large Ensemble (CESM-LE) to investigate the influence of internal variability on the detection of forced changes in two climate phenomena. First, using cyclone identification and compositing techniques within the CESM-LE, we investigate precipitation changes in extratropical cyclones under greenhouse gas forcing and the effect of internal variability on the detection of these changes. We find that the ensemble projects increased cyclone precipitation under twenty-first century business-as-usual greenhouse gas forcing and this response exceeds internal variability in both near- and far- futures. Further, we find that these changes are almost entirely driven by increases in cyclone moisture. Next, we explore the role of internal variability in projections of the annual cycle of surface temperature over Northern Hemisphere land. Internal variability strongly confounds forced changes in the annual cycle over many regions of the Northern Hemisphere. Changes over Europe, North Africa and Siberia, however, are large and easily detectable and further, are remarkably robust across model ensembles from the Coupled Model Intercomparison Project Phase 5 (CMIP5) archive. Using a simple energy balance model, we find that changes in the annual cycle over the three regions are mostly driven by changes in surface heat fluxes.

The thesis also presents a novel ensemble-based framework for diagnosing forced changes in regional climate variability. Changes in climate variability are commonly assessed in terms of changes in the variances of climate variables. The covariance response has received much less attention, despite the existence of large-scale modes of variability that induce covariations in climate variables over a wide range of spatial scales. Addressing this, the framework facilitiates a unified assessment of forced changes in the regional variances and covariances of climate variables.

With more than half of the world's population living in urban areas, it is important that the relationships between the urban environment and climate are better understood. The current research aims to continue the effort in assessing and understanding the urban environment through the use of a global climate model (GCM). Given the relative newness of the presence of an urban land type and model in a GCM, there are many more facets of the urban-climate relationship to be investigated. By comparing thirty-year ensembles of CAM4 coupled with CLM4 both with (U) and without (U_n) the inclusion of the urban land type, the sensitivity of the atmospheric model to urban land cover is assessed. As expected, largest differences tend to be in the Northern Hemisphere due to the location of most of the globe's densest and expansive cities. Significant differences in the basic climate variables of temperature and precipitation are present at annual, seasonal, and monthly scales in some regions. Seasonality to the urban influence also exists with the transition months of Spring and Fall having the largest difference in temperatures. Of the eleven regions defined by Oleson (2012), three were most impacted by the presence of urban land cover in the model—Europe, Central Asia, and East Asia.

Since urban attributes can vary greatly within one world continent, the sensitivity of regional climates to the urban type parameters is also explored. By setting all urban land cover to only one urban density type, the importance of city composition on climate, even within the same city, is highlighted. While preserving the distinct urban regional characteristics and the geographical distribution of urbanized areas, the model is run with homogeneous urban types: high density and tall building district. As with the default urban and excluded urban runs, a strong seasonality to the differences between the solo-high-density simulation and default urban (U_HD – U) and solo-tall-building-district-density simulation and default urban (U_TBD – U) exists. Overall, the transition and winter months are most sensitive to changes in urban density type.

In order to make informed decisions in response to future climate change, researchers, policy-makers, and the public need climate projections at the scale of few kilometers, rather than the scales provided by Global Climate Models. The North American Regional Climate Change Assessment Program (NARCCAP) is such a recent effort that addresses this necessity. As the climate models contain various levels of uncertainty, it is essential to evaluate the performance of such models and their representativeness of regional climate characteristics. When assessing climate change impacts, precipitation is a crucial variable, due to its direct influence on many aspects of our natural-human ecosystems such as freshwater resources, agriculture and energy production, and health and infrastructure. The current study performs an evaluation analysis of precipitation simulations produced by a set of dynamically downscaled climate models provided by the NARCCAP program. The Assessment analysis is implemented for a period that covers 20 to 30 years (1970-1999), depending on joint availability of both the observational and the NARCCAP datasets. In addition to direct comparison versus observations, the hindcast NARCCAP simulations are used within a hydrologic modeling analysis for a regional ecosystem in coastal Louisiana (Chenier Plain). The study concludes the NARCCAP simulations have systematic biases in representing average precipitation amounts, but are successful at capturing some of the characteristics on spatial and temporal variability. The study also reveals the effect of precipitation on salinity concentrations in the Chenier Plain as a result of using different precipitation forcing fields. In the future, special efforts should be made to reduce biases in the NARCCAP simulations, which can then lead to a better presentation of regional climate scenarios for use by decision makers and resource managers.

Severe drought plays a critical role in altering the magnitude and interannual variability of the net terrestrial carbon sink. Drought events immediately decrease net primary production (NPP), and drought length and magnitude tend to enhance this negative impact. However, satellite and in-situ measurements have also indicated that ecosystem recovery from extreme drought can extend several years beyond the return to normal climate conditions. If an ecosystem’s drought recovery time exceeds the time interval between successive droughts, these legacy effects may reinforce the impact of future drought. Since the frequency and severity of extreme climate events are expected to increase with climate change, both the immediate and prolonged impact of drought may contribute to amplified climate warming by decreasing the strength of the land carbon sink. However, it is unknown whether terrestrial biosphere models capture the impact of drought legacy effects on carbon stocks and cycling. Using a suite of twelve land surface models from the Multi-scale Synthesis and Terrestrial Model Intercomparison Project (MsTMIP), we assessed model ability to simulate drought legacy effects by analyzing the modeled NPP response to drought events across forested regions of the US and Europe. We found that modeled drought legacy effects last about one year (2% reduction in NPP), with complete NPP recovery in the second post-drought year. Since observations suggest that legacy effects extend up to four years post-drought, with a 9% growth reduction in the first post-drought year, models appear to underestimate both the timescales and magnitude of drought legacy effects. We further explored vegetation sensitivity to climate anomalies through global, time-lagged correlation analysis of NPP and climatic water deficit. Regional differences in the lag time between climate anomaly and NPP response are prevalent, but low sensitivities (correlations) characterize the entire region. Significant correlations coincided with characteristic lag times of 0 to 6 months, indicating relatively immediate NPP response to moisture anomalies. Model ability to accurately simulate vegetation’s response to drought and sensitivity to climate anomalies is necessary in order to produce reliable forecasts of land carbon sink strength and, consequently, to predict the rate at which climate change will progress in the future. Thus, the discrepancies between observed and simulated vegetation recovery from drought points to a potential critical model deficiency.

The Urban Heat Island (UHI) was investigated using satellite data, ground observations, and simulations with an Urban Canopy Parameterization in a numerical weather prediction model. Satellite-observed surface skin temperatures at Xi'an City and Oklahoma City (OKC) were analyzed to compare the UHI intensity for the two inland cities. A larger population density and larger building density in Xi'an City creates a stronger skin-level UHI effect. However, ground observed 2-m surface air temperature (Tair) data showed an urban cooling island (UCI) effect that occurred over an urban region in OKC during the daytime of July 19, 2003.

The sensitivity and accuracy of an Urban Canopy Model were evaluated by comparing simulation results between the urban and rural areas of OKC. The model reproduced skin temperature differences between the rural and urban area and reproduced a UCI effect in OKC. Furthermore, the Weather Research and Forecasting (WRF)/Noah/Single-Layer Urban Canopy Model (SLUCM) simulations were also compared with ground observations, including wind speeds, wind directions, and energy fluxes. Although the WRF/SLCUM model failed to simulate these variables accurately, it reproduced the diurnal variations of surface temperatures, wind speeds, wind directions and energy fluxes reasonably well.