Machine learning is a computational technology widely used in regression and classification tasks... more Machine learning is a computational technology widely used in regression and classification tasks. One of the drawbacks of its use in the analysis of spatial variables is that machine learning algorithms are in general, not designed to deal with spatially autocorrelated data. This often causes the residuals to exhibit clustering, in clear violation of the condition of independent and identically distributed random variables. In this work we analyze the performance of some well-established machine learning algorithms and one spatial algorithm in the prediction of the average rent price of certain real estate units in the Miami-Fort Lauderdale-West Palm Beach metropolitan area in Florida, USA. We defined “performance” as the goodness of fit achieved by an algorithm in conjunction with the degree of spatial association of the residuals. We identified significant differences between the machine learning algorithms in their sensitivity to spatial autocorrelation and the achieved goodness...
Social segregation in urban areas is a phenomenon with distinct spatial patterns. It is influence... more Social segregation in urban areas is a phenomenon with distinct spatial patterns. It is influenced by differences in socio-economic (e.g. income) and environmental (e.g. share of green space) factors. Since these influential factors vary in their intensity throughout space it is important to identify and to analyse the spatial patterns of these influencing factors in order to develop adequate coping strategies. Analysis of spatial autocorrelation, cluster analysis and geographically weighted regression may give new insights into spatial patterns and processes of the social segregation context in urban areas (Cahill & Mulligan 2007). In contrast to traditional global statistical regression analyses such as logistic regression or ordinary least square analyses spatial processes are addressed in this study in their very distinct behaviour using geographically weighted regressions (Brunsdon et al. 1998). The aim of this paper is to explore spatial patterns and underlying processes of so...
The development of biogas production in Germany has reshaped agricultural land use and production... more The development of biogas production in Germany has reshaped agricultural land use and production schemes since the implementation of the Renewable Energy Sources Act in 2000. It is associated with a widespread introduction of silage maize as a major substrate. While it contributes significantly to the renewable energy production, intensive maize-for-biogas production comes at the expense of loss of area for food production, is associated with negative ecological effects due to intensive and large-area monoculture, and is also associated with an increase in land prices. However, due to missing data, little is known about the plot-based distribution and development of silage maize for biogas production and, therefore, the local effects remain largely unstudied. This paper aims to identify the plot-level based likelihood of silage maize cultivated for biogas production in the Brandenburg region of Germany from 2008 through 2018. For this ongoing study, we developed and applied a spati...
Brazil's Soy Moratorium solidified the world's largest traders' commitment to stop soybean purcha... more Brazil's Soy Moratorium solidified the world's largest traders' commitment to stop soybean purchases from production areas deforested after July 2006. The aim was to remove deforestation from the soybean supplychain and halt one of the main drivers of forest loss in the Amazon biome. In this study, we investigated changes in deforestation at the property-level for the period 2004 to 2014. The objective was to examine direct and indirect deforestation, defined as on-property displacement and cross-parcel displacement deforestation for soybean expansion in the Amazon region of Mato Grosso, the leading soy-producing state of the Brazilian Amazon. We used publicly available property and land use data to quantify deforestation associated with cropland expansion. Similar to previous studies, we found that direct deforestation for soybean expansion declined following the implementation of the Soy Moratorium. Moreover, our analysis suggest that indirect deforestation occurred already before the implementation of the Soy Moratorium, and decreased following the first period of analyses. However, slight increases of indirect deforestation in the more recent periods, combined with decreasing direct deforestations rendered indirect deforestation to be responsible for more than half of the deforestation associated with soybean expansion. While we acknowledge the overall reduction of deforestation for soybean, our results suggest, given the increasing trends of deforestation in the Brazilian Amazon since 2013, to address indirect deforestation within the Soy Moratorium. This may be achieved by zero-property-deforestation commitments and by strengthening the integration between supply-chain actors, the soybean and beef purchasing companies and the federal policies aiming to control deforestation.
Abstract Urban heat island (UHI) studies have recognized ten factors as increasing the inner-city... more Abstract Urban heat island (UHI) studies have recognized ten factors as increasing the inner-city temperature compared with that of the surrounding suburbs. The UHI effect is a leading cause of heat-related diseases and mortality in many nations. However, there are still two main shortcomings. First, the effect of UHI is not well recognized in arid and semi-arid regions. Second, the association of multi-dimensional information with surface temperature in urban areas must be examined. This study focuses on the height-related aspects of urban geometry in an arid region. A range of multispectral and spatial vector data were used to derive the surface temperature and two-dimensional (2D) and three-dimensional (3D) information of the study area. All information was aggregated into a grid with common spatial resolution to create a homogeneous dataset. The machine learning statistical model of a boosted regression tree (BRT) was used to reflect the relative influence of 2D and 3D indicators with land surface temperature. Our results showed a cooler surface temperature in the city than in the surrounding area, leading to the question of whether the established UHI definition encompasses all types of cities. In addition, the thermal band was able to distinguish different spatial structures in the study area. The BRT analysis demonstrated that both multi-dimensional 2D and 3D indicators affect the surface temperature. In particular, the 3D indicators play a more important role than 2D indicators in shaping the surface temperature at different urban geometries of the study area. This new method can help urban planners identify the most influential 2D and 3D indicators that affect the surface temperature in different districts of a city.
Our aim with this paper is to present a novel approach for developing story lines and scenarios b... more Our aim with this paper is to present a novel approach for developing story lines and scenarios by combining qualitative knowledge and quantitative data from different disciplines and discussing the results with relevant decision makers. This research strategy offers a solid foundation for perspectives into the future. The "laboratory" is the Brazilian Amazon, one of the hotspots of land-use change where local and global interests both collide and converge: local livelihoods are affected by regional and global climate change and by the loss of biodiversity caused by local and global economic interests in agro-industrial land use; such use contributes, in turn, to climate change. After decades of diverse policy interventions the question arises: What can we learn from past trajectories for a more sustainable development in the future? To answer this question, we combined qualitative story lines for the region, reviewed by local experts, with quantitative land-use scenarios, to study their regional and local manifestations in space. These results were then discussed again with local and national experts. Our findings suggest that in-depth knowledge of the diverging perspectives at a very local level is a fundamental prerequisite for downscaling global scenarios and upscaling local approaches to sustainable land-use management and thus, to producing communicable and applicable results.
Vegetation provides important functions and services in urban areas, and vegetation heights divid... more Vegetation provides important functions and services in urban areas, and vegetation heights divided into vertical and horizontal units can be used as indicators for its assessment. Conversely, detailed area-wide and updated height information is frequently missing for most urban areas. This study sought to assess three vegetation height classes from a globally available TanDEM-X digital elevation model (DEM, 12 × 12 m spatial resolution) for Berlin, Germany. Subsequently, height distribution and its accuracy across biotope classes were derived. For this, a TanDEM-X intermediate DEM, a LiDAR DTM, an UltraCamX vegetation layer, and a biotope map were included. The applied framework comprised techniques of data integration and raster algebra for: Deriving a height model for all of Berlin, masking non-vegetated areas, classifying two canopy height models (CHMs) for bushes/shrubs and trees, deriving vegetation heights for 12 biotope classes and assessing accuracies using validation CHMs. The findings highlighted the possibility of assessing vegetation heights for total vegetation, trees and bushes/shrubs with low and consistent offsets of mean heights (total CHM: −1.56 m; CHM for trees: −2.23 m; CHM bushes/shrubs: 0.60 m). Negative offsets are likely caused by X-band canopy penetrations. Between the biotope classes, large variations of height and area were identified (vegetation height/biotope and area/biotope:~3.50-~16.00 m; 4.44%-96.53%). The framework and results offer a great asset for citywide and spatially explicit assessment of vegetation heights as an input for urban ecology studies, such as investigating habitat diversity based on the vegetation's heterogeneity.
Vegetation has gained importance in respective debates about climate change mitigation and adapta... more Vegetation has gained importance in respective debates about climate change mitigation and adaptation in cities. Although recently developed remote sensing techniques provide necessary city-wide information, a sufficient and consistent city-wide information of relevant urban ecosystem services, such as carbon emissions offset, does not exist. This study uses city-wide, high-resolution, and remotely sensed data to derive individual tree species information and to estimate the above-ground carbon storage of urban forests in Berlin, Germany. The variance of tree biomass was estimated using allometric equations that contained different levels of detail regarding the tree species found in this study of 700 km2, which had a tree canopy of 213 km2. The average tree density was 65 trees/ha per unit of tree cover and a range from 10 to 40 trees/ha for densely urban land cover. City-wide estimates of the above-ground carbon storage ranged between 6.34 and 7.69 tC/ha per unit of land cover, depending on the level of tree species information used. Equations that did not use individually localized tree species information undervalued the total amount of urban forest carbon storage by up to 15 %. Equations using a generalized estimate of dominant tree species information provided rather precise city-wide carbon estimates. Concerning differences within a densely built area per unit of land cover approaches using individually localized tree species information prevented underestimation of mid-range carbon density areas (10–20 tC/ha), which were actually up to 8.4 % higher, and prevented overestimation of very low carbon density areas (0–5 tC/ha), which were actually up to 11.4 % lower. Park-like areas showed 10 to 30 tC/ha, whereas land cover of very high carbon density (40–80 tC/ha) mostly consisted of mixed peri-urban forest stands. Thus, this approach, which uses widely accessible and remotely sensed data, can help to improve the consistency of forest carbon estimates in cities.
Highly dynamic peri-urban areas, particularly in the Global South, face many challenges including... more Highly dynamic peri-urban areas, particularly in the Global South, face many challenges including a lack of infrastructure, ownership conflicts, land degradation, and sustainable food production. This study aims to assess spatial land use characteristics and processes in peri-urban areas using the case of Dar es Salaam, Tanzania. A mixed-method approach was applied, consisting of expert interviews and spatial data analysis, on a local scale along an urban–rural gradient. Expert interviews were conducted during a field study and analyzed regarding the characteristics and processes of peri-urban land development. A GIS-based analysis of land use patterns was applied using satellite imagery and Open Street Map data to identify a number of variables, such as building density and proximity to environmental features. Results show specific patterns of land use indicators, which can be decreasing (e.g., house density), increasing (e.g., tree coverage), static (e.g., house size), or randomly...
Two of the dominant processes shaping today’s European cities are the ageing and diversification ... more Two of the dominant processes shaping today’s European cities are the ageing and diversification of the population. Given that the range of action usually decreases in later life, the living environment around the place of residence plays an important role in the social integration of the older generation. Hence, spatial patterns of residence indicate the extent of opportunities for the older population to engage in urban life and, therefore, need to be addressed by urban planning and policy. The aim of this article is to study the interrelation between diversity in later life—in terms of migrant history, gender, social class, and age—as well as planned and actual (past) movements of elders. We have chosen Berlin as a case study and draw from a quantitative survey with elders (age 60+) from diverse backgrounds (N = 427). Our results from descriptive analysis and statistical hypothesis tests show that age impacts people’s past and planned movement; we observe a peak in the decisions ...
Rapid urban expansion is a significant contributor to land cover change and poses a challenge to ... more Rapid urban expansion is a significant contributor to land cover change and poses a challenge to environmental sustainability, particularly in less developed countries. Insufficient data about urban expansion hinders effective land use planning. Therefore, a high need to collect, process, and disseminate land cover data exists. This study focuses on urban land cover change detection using Geographic Information Systems and remote sensing methods to produce baseline information in support for land use planning. We applied a supervised classification of land cover of LANDSAT data from 1987, 2002, and 2017. We mapped land cover transitions from 1987 to 2017 and computed the net land cover change during this time. Finally, we analyzed the mismatches between the past and current urban land cover and land use plans and quantified the non-urban development area lost to urban/built-up. Our results indicated an increase in urban/built-up and bare land cover types, while vegetation land cover...
Machine learning is a computational technology widely used in regression and classification tasks... more Machine learning is a computational technology widely used in regression and classification tasks. One of the drawbacks of its use in the analysis of spatial variables is that machine learning algorithms are in general, not designed to deal with spatially autocorrelated data. This often causes the residuals to exhibit clustering, in clear violation of the condition of independent and identically distributed random variables. In this work we analyze the performance of some well-established machine learning algorithms and one spatial algorithm in the prediction of the average rent price of certain real estate units in the Miami-Fort Lauderdale-West Palm Beach metropolitan area in Florida, USA. We defined “performance” as the goodness of fit achieved by an algorithm in conjunction with the degree of spatial association of the residuals. We identified significant differences between the machine learning algorithms in their sensitivity to spatial autocorrelation and the achieved goodness...
Social segregation in urban areas is a phenomenon with distinct spatial patterns. It is influence... more Social segregation in urban areas is a phenomenon with distinct spatial patterns. It is influenced by differences in socio-economic (e.g. income) and environmental (e.g. share of green space) factors. Since these influential factors vary in their intensity throughout space it is important to identify and to analyse the spatial patterns of these influencing factors in order to develop adequate coping strategies. Analysis of spatial autocorrelation, cluster analysis and geographically weighted regression may give new insights into spatial patterns and processes of the social segregation context in urban areas (Cahill & Mulligan 2007). In contrast to traditional global statistical regression analyses such as logistic regression or ordinary least square analyses spatial processes are addressed in this study in their very distinct behaviour using geographically weighted regressions (Brunsdon et al. 1998). The aim of this paper is to explore spatial patterns and underlying processes of so...
The development of biogas production in Germany has reshaped agricultural land use and production... more The development of biogas production in Germany has reshaped agricultural land use and production schemes since the implementation of the Renewable Energy Sources Act in 2000. It is associated with a widespread introduction of silage maize as a major substrate. While it contributes significantly to the renewable energy production, intensive maize-for-biogas production comes at the expense of loss of area for food production, is associated with negative ecological effects due to intensive and large-area monoculture, and is also associated with an increase in land prices. However, due to missing data, little is known about the plot-based distribution and development of silage maize for biogas production and, therefore, the local effects remain largely unstudied. This paper aims to identify the plot-level based likelihood of silage maize cultivated for biogas production in the Brandenburg region of Germany from 2008 through 2018. For this ongoing study, we developed and applied a spati...
Brazil's Soy Moratorium solidified the world's largest traders' commitment to stop soybean purcha... more Brazil's Soy Moratorium solidified the world's largest traders' commitment to stop soybean purchases from production areas deforested after July 2006. The aim was to remove deforestation from the soybean supplychain and halt one of the main drivers of forest loss in the Amazon biome. In this study, we investigated changes in deforestation at the property-level for the period 2004 to 2014. The objective was to examine direct and indirect deforestation, defined as on-property displacement and cross-parcel displacement deforestation for soybean expansion in the Amazon region of Mato Grosso, the leading soy-producing state of the Brazilian Amazon. We used publicly available property and land use data to quantify deforestation associated with cropland expansion. Similar to previous studies, we found that direct deforestation for soybean expansion declined following the implementation of the Soy Moratorium. Moreover, our analysis suggest that indirect deforestation occurred already before the implementation of the Soy Moratorium, and decreased following the first period of analyses. However, slight increases of indirect deforestation in the more recent periods, combined with decreasing direct deforestations rendered indirect deforestation to be responsible for more than half of the deforestation associated with soybean expansion. While we acknowledge the overall reduction of deforestation for soybean, our results suggest, given the increasing trends of deforestation in the Brazilian Amazon since 2013, to address indirect deforestation within the Soy Moratorium. This may be achieved by zero-property-deforestation commitments and by strengthening the integration between supply-chain actors, the soybean and beef purchasing companies and the federal policies aiming to control deforestation.
Abstract Urban heat island (UHI) studies have recognized ten factors as increasing the inner-city... more Abstract Urban heat island (UHI) studies have recognized ten factors as increasing the inner-city temperature compared with that of the surrounding suburbs. The UHI effect is a leading cause of heat-related diseases and mortality in many nations. However, there are still two main shortcomings. First, the effect of UHI is not well recognized in arid and semi-arid regions. Second, the association of multi-dimensional information with surface temperature in urban areas must be examined. This study focuses on the height-related aspects of urban geometry in an arid region. A range of multispectral and spatial vector data were used to derive the surface temperature and two-dimensional (2D) and three-dimensional (3D) information of the study area. All information was aggregated into a grid with common spatial resolution to create a homogeneous dataset. The machine learning statistical model of a boosted regression tree (BRT) was used to reflect the relative influence of 2D and 3D indicators with land surface temperature. Our results showed a cooler surface temperature in the city than in the surrounding area, leading to the question of whether the established UHI definition encompasses all types of cities. In addition, the thermal band was able to distinguish different spatial structures in the study area. The BRT analysis demonstrated that both multi-dimensional 2D and 3D indicators affect the surface temperature. In particular, the 3D indicators play a more important role than 2D indicators in shaping the surface temperature at different urban geometries of the study area. This new method can help urban planners identify the most influential 2D and 3D indicators that affect the surface temperature in different districts of a city.
Our aim with this paper is to present a novel approach for developing story lines and scenarios b... more Our aim with this paper is to present a novel approach for developing story lines and scenarios by combining qualitative knowledge and quantitative data from different disciplines and discussing the results with relevant decision makers. This research strategy offers a solid foundation for perspectives into the future. The "laboratory" is the Brazilian Amazon, one of the hotspots of land-use change where local and global interests both collide and converge: local livelihoods are affected by regional and global climate change and by the loss of biodiversity caused by local and global economic interests in agro-industrial land use; such use contributes, in turn, to climate change. After decades of diverse policy interventions the question arises: What can we learn from past trajectories for a more sustainable development in the future? To answer this question, we combined qualitative story lines for the region, reviewed by local experts, with quantitative land-use scenarios, to study their regional and local manifestations in space. These results were then discussed again with local and national experts. Our findings suggest that in-depth knowledge of the diverging perspectives at a very local level is a fundamental prerequisite for downscaling global scenarios and upscaling local approaches to sustainable land-use management and thus, to producing communicable and applicable results.
Vegetation provides important functions and services in urban areas, and vegetation heights divid... more Vegetation provides important functions and services in urban areas, and vegetation heights divided into vertical and horizontal units can be used as indicators for its assessment. Conversely, detailed area-wide and updated height information is frequently missing for most urban areas. This study sought to assess three vegetation height classes from a globally available TanDEM-X digital elevation model (DEM, 12 × 12 m spatial resolution) for Berlin, Germany. Subsequently, height distribution and its accuracy across biotope classes were derived. For this, a TanDEM-X intermediate DEM, a LiDAR DTM, an UltraCamX vegetation layer, and a biotope map were included. The applied framework comprised techniques of data integration and raster algebra for: Deriving a height model for all of Berlin, masking non-vegetated areas, classifying two canopy height models (CHMs) for bushes/shrubs and trees, deriving vegetation heights for 12 biotope classes and assessing accuracies using validation CHMs. The findings highlighted the possibility of assessing vegetation heights for total vegetation, trees and bushes/shrubs with low and consistent offsets of mean heights (total CHM: −1.56 m; CHM for trees: −2.23 m; CHM bushes/shrubs: 0.60 m). Negative offsets are likely caused by X-band canopy penetrations. Between the biotope classes, large variations of height and area were identified (vegetation height/biotope and area/biotope:~3.50-~16.00 m; 4.44%-96.53%). The framework and results offer a great asset for citywide and spatially explicit assessment of vegetation heights as an input for urban ecology studies, such as investigating habitat diversity based on the vegetation's heterogeneity.
Vegetation has gained importance in respective debates about climate change mitigation and adapta... more Vegetation has gained importance in respective debates about climate change mitigation and adaptation in cities. Although recently developed remote sensing techniques provide necessary city-wide information, a sufficient and consistent city-wide information of relevant urban ecosystem services, such as carbon emissions offset, does not exist. This study uses city-wide, high-resolution, and remotely sensed data to derive individual tree species information and to estimate the above-ground carbon storage of urban forests in Berlin, Germany. The variance of tree biomass was estimated using allometric equations that contained different levels of detail regarding the tree species found in this study of 700 km2, which had a tree canopy of 213 km2. The average tree density was 65 trees/ha per unit of tree cover and a range from 10 to 40 trees/ha for densely urban land cover. City-wide estimates of the above-ground carbon storage ranged between 6.34 and 7.69 tC/ha per unit of land cover, depending on the level of tree species information used. Equations that did not use individually localized tree species information undervalued the total amount of urban forest carbon storage by up to 15 %. Equations using a generalized estimate of dominant tree species information provided rather precise city-wide carbon estimates. Concerning differences within a densely built area per unit of land cover approaches using individually localized tree species information prevented underestimation of mid-range carbon density areas (10–20 tC/ha), which were actually up to 8.4 % higher, and prevented overestimation of very low carbon density areas (0–5 tC/ha), which were actually up to 11.4 % lower. Park-like areas showed 10 to 30 tC/ha, whereas land cover of very high carbon density (40–80 tC/ha) mostly consisted of mixed peri-urban forest stands. Thus, this approach, which uses widely accessible and remotely sensed data, can help to improve the consistency of forest carbon estimates in cities.
Highly dynamic peri-urban areas, particularly in the Global South, face many challenges including... more Highly dynamic peri-urban areas, particularly in the Global South, face many challenges including a lack of infrastructure, ownership conflicts, land degradation, and sustainable food production. This study aims to assess spatial land use characteristics and processes in peri-urban areas using the case of Dar es Salaam, Tanzania. A mixed-method approach was applied, consisting of expert interviews and spatial data analysis, on a local scale along an urban–rural gradient. Expert interviews were conducted during a field study and analyzed regarding the characteristics and processes of peri-urban land development. A GIS-based analysis of land use patterns was applied using satellite imagery and Open Street Map data to identify a number of variables, such as building density and proximity to environmental features. Results show specific patterns of land use indicators, which can be decreasing (e.g., house density), increasing (e.g., tree coverage), static (e.g., house size), or randomly...
Two of the dominant processes shaping today’s European cities are the ageing and diversification ... more Two of the dominant processes shaping today’s European cities are the ageing and diversification of the population. Given that the range of action usually decreases in later life, the living environment around the place of residence plays an important role in the social integration of the older generation. Hence, spatial patterns of residence indicate the extent of opportunities for the older population to engage in urban life and, therefore, need to be addressed by urban planning and policy. The aim of this article is to study the interrelation between diversity in later life—in terms of migrant history, gender, social class, and age—as well as planned and actual (past) movements of elders. We have chosen Berlin as a case study and draw from a quantitative survey with elders (age 60+) from diverse backgrounds (N = 427). Our results from descriptive analysis and statistical hypothesis tests show that age impacts people’s past and planned movement; we observe a peak in the decisions ...
Rapid urban expansion is a significant contributor to land cover change and poses a challenge to ... more Rapid urban expansion is a significant contributor to land cover change and poses a challenge to environmental sustainability, particularly in less developed countries. Insufficient data about urban expansion hinders effective land use planning. Therefore, a high need to collect, process, and disseminate land cover data exists. This study focuses on urban land cover change detection using Geographic Information Systems and remote sensing methods to produce baseline information in support for land use planning. We applied a supervised classification of land cover of LANDSAT data from 1987, 2002, and 2017. We mapped land cover transitions from 1987 to 2017 and computed the net land cover change during this time. Finally, we analyzed the mismatches between the past and current urban land cover and land use plans and quantified the non-urban development area lost to urban/built-up. Our results indicated an increase in urban/built-up and bare land cover types, while vegetation land cover...
Uploads
Papers by Tobia Lakes