The Canadian Arctic currently faces changing coastlines due isostatic rebound and climate change-driven sea-levels rising. This thesis seeks to answer how local coastlines will change over time under different Representative Concentration Pathway (RCP) scenarios, where errors in modelled coastlines come from, and how much of an impact isostatic rebound has on sea-levels compared to climate-change driven changes. Maps which show changes in coastline through predicted sea-level changes in 2035, 2065, and 2100 have been produced through ArcGIS Pro’s Forest-based regressions tool, with training data from historical climate variables and projection data under CMIP5 for RCP 2.6, 4.5, and 8.5, where RCP 2.6 refers to a low emissions scenario, RCP 4.5 refers to a moderate emissions scenario, and RCP 8.5 refers to a high emissions scenario. Isostatic rebound data can be sourced and run through the ICE-4G model. Multiple models were examined to ascertain which climate variables are needed to produce a stable model, using training data prior to and including 2000 and testing data post 2000 from 9 spatially diverse locations in the Canadian Arctic. These results suggest that for Cambridge Bay, Resolute, and Alert, there will be little change in coastline under any RCP scenario. However, for Tuktoyaktuk coastlines are projected to advance significantly. While 9 locations total were used in during the process, only 4 were examined in greater detail due to data and time constraints. These finding should be considered when making plans for future coastal infrastructure. (Less)
Anthropogenic climate change has increased the frequency of extreme drought events and leads to “hotter” droughts. Topography controls plant available water and site-specific climatic conditions. Drought sensitivity may therefore vary over short distances between wet and dry locations of the landscape. However, topography is rarely included as a factor in research investigating vegetation dynamics. Due to being located in often complex terrain and growing in unmanaged conditions, primary forests offer the opportunity to study the influence of topography on the impact of droughts on vegetation in natural ecosystems. This study investigated the role of the local environment and topography in controlling drought impact of the 2018 summer drought in primary forests in Sweden, by applying a random forest modelling approach. Drought effects were modelled using summer 2018 Landsat EVI2 anomalies (z-scores) as a function of predictor variables including terrain indices, bioclimatic variables, and forest properties. Topographic predictors included primary terrain indices such as aspect, slope, and the Topographic Position Index (TPI) as well as the more complex compound terrain indices the Topographical Wetness Index (TWI) and Hight Above Nearest Drainage (HAND). The most influential variables were selected with a recursive feature elimination approach, creating a model with 13 predictors. The final model explained 48% (R2 = 0.48) of the variance in an independent subset of EVI2 anomalies. Primary terrain indices described the spatial variability of z-scores better than compound terrain indices. Forest located on steep slopes, high topographic positions, and south facing aspects were associated with negative z-scores, indicative of reduced primary productivity. Valley bottoms and north-facing aspects mostly showed no or positive drought impact. Drought effects, meaning EVI2 anomalies were more negative with increasing distance to wetlands. These topographic and wetland effects on drought impact were seen across all forest types and latitudes, however the severity of negative drought impact differed between forest classes and was more pronounced in the south. The results clearly show that the impact of the 2018 summer drought was strongly controlled by the local terrain in Swedish primary forests, highlighting the importance of incorporating topography in studies aimed at quantifying and predicting drought impact on vegetation. Previous research suggests that primary forests were less affected by the 2018 drought than managed forests were. This analysis suggests that the widespread drainage of wetlands and establishment of monocultures in managed forests may explain this difference. Forest management may therefore exacerbate the future impact of potentially more severe and more frequent droughts. (Less)
Due to climate change, cities are increasingly facing the situation of being exposed to more frequent and prolonged extreme weather events in the future, such as heat and droughts, which are exacerbated by urban heat island (UHI) effects. The implementation of green infrastructure (GI) in the form of trees is seen as a possible adaptation measure to regulate the microclimate. However, there is an increased risk that both existing and future urban tree populations will suffer significantly from changing climate conditions. The aim of this study was to assess to what extent the current tree population in Berlin and Vienna will be threatened by heat and drought periods in the future and to what extent adaptation measures might become necessary. To this end, future climate-related heat and drought threats were investigated based on two emission scenarios (RCP 4.5, RCP 8.5) and selected climate indicators for both the near future (2021-2050) and the distant future (2071-2100). The heat and drought tolerance of trees was assessed based on a literature review. In addition, the exposure of categorised trees in relation to UHI effects was investigated to identify particularly vulnerable locations and recurring patterns. The risk of extreme events (heat and drought) was found to increase in the future for both climate scenarios studied for both cities, especially in the distant future and at the beginning and end of the growing season. Of the urban tree species studied, the majority were classified as moderately sensitive and moderately tolerant. The distribution of highly sensitive trees within the city showed similar distribution patterns in the two cities with single hotspots in parks but different average UHI intensity exposures. It can be concluded that timely, targeted planning of GI adaptation measures is already essential for both cities to be able to provide adequate GI in the near, but especially in the distant future. (Less)
Three-dimensional (3D) city models are increasingly utilized among stakeholders of different sectors, particularly those dealing with building information. 3D city models are a three-dimensional digital representation of urban structures, most prominently buildings. They offer diverse applications and have gained wide acceptance as a support tool in planning practices. The push to achieve greater interoperability in the GIS sector has been a motivating factor for standardized practices for 3D city models. National frameworks for 3D data have been established in several countries to facilitate information sharing between different actors such as municipalities or other administrative bodies. A national specification (NS) pertaining to 3D data is in development in Sweden, along with a standard aimed explicitly at 3D building models, denoted as NS building. Municipalities will have to adjust their modeling practices to adhere to the standard in the future realization of their 3D city models. Proprietary software, such as Esri, has been common to establish 3D city models in the context of municipalities in Sweden. The interoperability of pre-existing 3D building models derived in an Esri environment with NS building has not yet been addressed. This thesis aimed to propose a workflow to transform Esri-derived building models and conform them to the NS building standard to assess the applicability of the Swedish standard for building models in Esri MultiPatch format. As a case study, a 3D city model from the municipality of Karlskrona was used to develop a method of transformation between the formats. The project results were that a script that streamlines the process of converting building models in Esri MultiPatch format to NS building could be established. For this, a spatial Extract, Transform, and Load (ETL) process was developed and realized using Safe Software's Feature Manipulation Engine (FME). The process centered around making the geometries and attributes of the test dataset compatible with NS building models of different levels of detail (LoDs) through the process of de-aggregation, restructuring, and defining surfaces according to their types (bottom, roof, or wall surfaces). Each LoD was represented as a single feature class, and the attribute table was stored in a stand-alone manner, sharing a common identifier with the building models. The output models and the attribute table could be stored in the initial Esri file geodatabase (File GDB). The script that this thesis presents manages to capture all purely Esri-derived models and model them according to the Swedish standard, with separated surfaces of the bottom, roof, and wall parts relevant to the different LoDs required of the Swedish standard. This excludes the creation of roof surfaces for one of the LoDs represented in the standard. Moreover, as the standard was still in development at the time of writing, the presented workflow may need to be subjected to changes in the future and edited if it were to be better inclusive of a multi-source dataset, as opposed to solely Esri derived. The externally sourced models present in the test data introduced more nuance, which was difficult to account for on all fronts.
Nowadays it has become relevant for scientists to understand the impact of tropical forest structure on the Global Carbon Cycle. Dynamic vegetation models have been developed to pursue this issue, specifically by improving algorithms that could analyze the allometry, biomass content and mortality rate of each singular simulated plant individual and how this responds to climate change, management and rising CO2 concentrations. In these studies, I use and test the LPJ-GUESS vegetation model, which is an ecosystem framework for modelling the structure and dynamics of terrestrial ecosystems at landscape, regional and global scales (Smith. et al., 2001). It was used in order to simulate a better reproduction of structural tropical tree dynamics through the use of allometric equations for the scaling of the tropical tree growth in the Amazon forest. Specifically, in four locations from the Amazon forest, stand structure and developments were well reproduced by the model in comparison to observations of tropical forest structure available from the TEAM network, which provides observational information on the number of individual trees and their sizes on a range of plots across the tropics. Specifically, allometries were both investigated by adjusting standard settings in the model to reach better performances and assess how these settings impact size structure and observations. My results show that the model is not fitting the observation by not simulating accurately directly the size structure of the trees. Improper algorithms and the use of inappropriate parameter values were identified as possible explanations for the lack of agreement of the model with the TEAM network observations. (Less)
