Abstracts Track 2023

Area 1 - Data Acquisition and Processing

Nr: 13

Characterising Spatiotemporal Variability of South Asia’s Climate Extremes in Past Decades


Yun Chen and Tingbao Xu

Abstract: We systematically examined past spatiotemporal changes in climate variability to gain some cross-regional insights into South Asia’s vulnerability to extreme conditions. Gridded Asian Precipitation-Highly-Resolved Observational Data Integration Towards Evaluation (APHRODITE) precipitation and Princeton Global Meteorological Forcing Dataset (PRINCETON) temperature data from 1975-2004 were used to derive a suite of annual extreme indices. Long term mean and decadal variations of these indices were mapped. Long-term change tendencies were also detected from a suite of ‘slope’ maps composed by the 30 yr change trend at each grid cell in the region. Most precipitation indices indicated a tendency towards drier conditions, whereas all temperature indices marked a steady coherent warming trend. The extremely wet day precipitation index exhibited the largest change, indicating an increase in heavy precipitation in South Asia. The highest maximum temperature extreme showed increases, indicating more unbearable heatwaves in the region. These trends present a previously unrecognised regional picture of the patterns and trends in historical climate extremes, with each grid cell representing spatiotemporal characteristics of changes. The present study is superior to most studies that only summarise an averaged regional trend from tendencies over large areas, and therefore will improve trans-boundary understanding of extreme climates in South Asia. Our study also exemplifies the application of existing gridded regional/global data sets. It provides valuable means of cross-regional information for bridging gaps where gauging observations are unavailable, particularly in data-poor developing countries.

Area 2 - Knowledge Extraction and Management

Nr: 20

Investigation of Efficient Delivery Methods by Survey-based Simulation of Team Delivery


Toshihiro Osaragi and Yuya Taguchi

Abstract: In recent years, with the increase in the use of delivery services, a method for efficient delivery is required. In this paper, we propose a method to shorten the delivery service time of multiple staff (team delivery) by using some techniques developed for the multiple Traveling Salesman Problem. Using the method, a simulation is executed based on the parameters taken from the survey of the actual delivery service. Furthermore, we compare the actual and optimal delivery routes, and demonstrate some new findings.

Area 3 - Domain Applications

Nr: 17

Spatial Analysis of Health Outcomes and Population Exposure to Aviation Noise in the Contiguous United States


Yelena Ogneva-Himmelberger

Abstract: Objective. In this paper, we use GIS to identify areas affected by high aviation noise in the Contiguous United States and to analyze the prevalence of chronic disease in these areas. Specifically, we address two questions: Is there a statistical difference in the prevalence of high blood pressure, coronary heart disease and stroke between areas with different aviation noise levels? Where are the areas of health disparities that are simultaneously experiencing the highest noise exposure? High noise exposure a risk factor for these chronic diseases, so it is important to know where these two conditions are co-occurring. Our findings could inform the development of effective programs and policies related to noise abatement, and help prioritize investment in areas with the biggest health inequities. Data. Aviation noise raster (30-meter) was obtained from the Bureau of Transportation Statistics. It represents the average daily noise level, modeled using flights and meteorological data. Health outcomes data was obtained from PLACES database (Population Level Analysis and Community Estimates) at census tract level. PLACES provides local-level model-based estimates for 29 health measures. We selected 3 measures known to be associated with chronic aviation noise exposure – the prevalence of high blood pressure, coronary heart disease and stroke. Methods. We reclassified noise data into “low noise” (45-65 db) and “high noise” (65db+) categories. 65db is the threshold used by the Federal Aviation Administration to define “significant” aircraft noise exposure. To compare noise-affected areas with non-affected areas, we created circular buffers around each airport, so that the noise-affected area fits inside the buffer. Health outcomes data were clipped to airport buffers, and divided into three groups: those affected by “high-noise”, “low-noise”, and those not affected by noise (30,924 census tracts). Median values were calculated for each exposure group for each health outcome and compared using Kruskal-Wallis Test. To identify areas of the highest health disparities we selected census tracts in the 90-100th percentile for each health outcome and overlaid them with the “high noise”. This resulted in a selection of census tracts where a high proportion of people are estimated to suffer from chronic diseases, and are exposed to a very high noise. Results. The median values for health outcomes were higher in noise-exposed areas compared to non-exposed areas. Statistical test confirmed that these values were statistically different between three exposure groups (<0.05). We identified 126 census tracts (25,623 people) that have the highest prevalence of chronic disease and are also in very high noise areas. The highest number of people exposed to high noise is in Florida, Louisiana, Alabama, Texas, and Georgia. Conclusion. We analyzed two novel, high resolution, nationwide data sets and found a statistically significant association between increasing prevalence of three chronic diseases and elevated noise levels. We also identified areas where extremely high values of health outcomes and noise overlap - areas where a high proportion of the population is estimated to have poor health, and is experiencing chronic exposure to very high airport noise levels. Our findings can support practitioners and policy makers in their efforts to improve the conditions of vulnerable populations in these areas.

Nr: 18

Investigation of Volunteered Geographic Information Data for Near-Real-Time Spatial Analysis of Fire Hazards


Janine Florath, Jocelyn Chanussot and Sina Keller

Abstract: In recent years forceful natural hazards and disasters have posed a severe threat to humans [1, 2]. First responders, like rescue services, play a crucial role in reducing any threats and need real-time information about the location and possible extent of the hazard to mitigate and cope with potentially hazardous effects [3]. The mapping of approximate extents of wildfires can be achieved with remote sensing (RS) data with sufficient accuracies [4], but rarely in (near) real-time for non-commercial satellites due to the satellites’ overpass time. As alternative data, volunteered geographic information (VGI) data, such as Twitter data, can be used to identify potential hazard areas or affected areas in almost near-real time [5, 6]. In our study, we aim to investigate and evaluate the possibilities and limitations of tweets for the spatial analysis of fire hazards in the absence of RS data. As a case study, we focus on the Bobcat fire in the Los Angeles National Forest in California, Unites States (U.S.), in 2020. We extract tweets with relevant keywords and their geographic location for the time of the fire from 06/09/2020 to 10/09/2020. Besides, as reference data, we rely on the given fire area generated with Sentinel-2 RS data (acc. to [4]) on 10th September 2020. When focusing on the methodological part, we consider different approaches. For the geographic cluster point of all the tweets, we first calculate the weighted median location of the tweet points, taking into account influencing factors like population density. Next, we estimate approximate hazard locations or affected areas with several approaches according to the availability of the respective data in the tweet texts: viewing angle to fire (from location point to mentioned target point), distance to fire, and road segment blocking information. Finally, the approaches’ results are combined, and the possibly affected minimal and maximal hazard areas of the fire are estimated with a confidence interval. The findings from the Bobcat fire study indicate that the calculation of the weighted median center from tweets aligns well with the fire area, as determined from RS imagery. Our approach also demonstrates that the estimated minimal and maximal affected area corresponds closely to the area detected from RS. However, it should be noted that the estimation of the affected area from VGI may yield slightly different results, with the center of the estimated area located in an area where the population is more affected by the wildfire. Using geospatial analysis approaches on VGI obtained from sources such as Twitter can assist in identifying the region impacted by natural disasters, such as wildfires. This usage is particularly beneficial when RS data are unavailable or inadequate for determining the extent of potential impacts on populations. Despite limitations in precision when compared to other forms of data, VGI can provide approximate and timely estimates of affected areas, enabling emergency responders to plan their operations, allocate resources, and prioritize areas of concern. [1] Karnjana Songwathana. 2018 [2] Ben Wisner et al. 2014 [3] Haiyan Hao et al. 2020 [4] Janine Florath et al. 2022 [5] André Dittrich et al. 2014 [6] Yandong Wang et al. 2015.

Nr: 12

Small Infill Potentials: An Automated Approach to Identify Vacant Lots Based on Cadastral Data


Denise Ehrhardt, Martin Behnisch and Mathias Jehling

Abstract: Infill development policies have been widely adopted as strategies to reduce urban sprawl and to promote sustainable urban transformation. However, little empirical data is available to analyse infill processes and to facilitate future mobilisation of infill potentials. This is especially true for small-scale residential infill, which often takes place on vacant or underused lots as soft densification. We address this issue by introducing a geospatial approach that enables automatic detection of vacant lots for large areas based on cadastral data. Based on the definition of vacant lots, we derive parcel and neighbourhood characteristics to delimit them from built-up parcels and other infill development potentials and employ a hierarchical decision tree to decide whether a parcel can be considered a vacant lot or not. The workflow consists of five successive processing steps. First, we delimit built-up areas and reduce the dataset to land use type residential and mixed use. Second, a preliminary dataset of non-built-up parcels is generated. As many of the identified parcels do not match the definition of vacant lots, in a third step the preliminary data is revised to eliminate too small, odd shaped or not accessible parcels. To distinguish persistent from temporary vacant lots, we combine the final dataset of vacant lots with information on new development areas. Finally, the accuracy of the results is evaluated by comparing it to a reference dataset. The approach proves to be robust regarding its precision, showing an accuracy of 95.5 %. The method is applied in a study area in Germany, covering 11.000 km² and containing large cites as well as rural municipalities. Based on the derived vacant lots dataset, we analyse spatio-temporal development for the period of 2011-2021. Our results show that every fourth vacant lots was mobilised since 2011. Yet, at the same time, additional vacant lots emerge, as new residential development areas are not fully built-up, which, in rural areas, results in a net increase of vacant lots. Although the quantity of vacant lot area in 2021 suggests high potential for residential infill, parcel- and neighbourhood characteristics indicate that the main development on these infill potentials is expected to promote additional single-family housing, rather than more dense structures. We argue that more active and strategic planning of soft densification processes is needed to realise the full potential of infill on vacant lots. In practice, automatic identification and monitoring of infill potentials and processes are important both, for policy-making and for local planning practitioners. For small municipalities with little financial capacities, the approach can provide an overview regarding their vacant lots and can serve as a basis for strategic planning decisions. For the regional or national level, a yearly monitoring can be established at little cost. Although the approach proves to be robust regarding its precision and is promising for a nation-wide application, the soon expected data availability for whole Germany has to be awaited and some optimisation in the method have to be made to implement the workflow in practice.