Kernel Tuner
Kernel Tuner greatly simplifies the development of highly-optimized and auto-tuned CUDA, OpenCL, and C code, supporting many advanced use-cases and optimization strategies that speed up the auto-tuning process.
Forecasting and mapping human thermal comfort in urban areas
On warm days, a city can become uncomfortably hot. Cities experience a so-called urban heat island effect and are typically warmer than their surroundings. Also, cities experience two parallel developments: increased urbanization and an enhanced frequency of warm episodes due to the changing global climate. These developments may increase human health risks. For example, the 2003 European heat wave led to a health crisis in several European countries and 70.000 heat-related deaths. The situation is especially worrying for vulnerable groups, such as the elderly and people with health issues, but affects the work productivity and well-being of us all. If we want to keep life in the city comfortable in the future, a better understanding of urban weather and climate is essential.
Screencast Demo Summer in the City from Netherlands eScience Center on Vimeo.
Summer in the City develops a novel prototype hourly forecasting system for human thermal comfort in urban areas on street level (~ 100 m scale). The forecasting system can be exploited by weather and health agencies for urban weather forecasting and heat wave warnings. More knowledge will be gained on the spatial variability of microclimates within cities, which can assist governmental agencies in decision making processes concerning public health and the environment. It will lead, for example, to a better understanding of the impact that planting trees would have on a specific microclimate.
This project involves the processing and creation of large data sets of urban physical properties. These data sets come from a variety of sources like aerial photos and Lidar measurements (a remote sensing method that can generate precise, three-dimensional information about the shape of the Earth and its surface characteristics), cadastral maps, and satellite imagery and will be combined in new and innovative ways.
Above: The Summer in the City forecasting application (beta). How much warmer is your street? The map shows the urban heat island effect (UHI). It should be used in combination with a traditional temperature forecast to get an estimation of the actual temperature in your neighborhood. Take the temperature forecast and add the UHI50P for an average day, and the UHI95P for a heatwave. When hovering over the map, a pop-up also shows the population count, and the area fraction of urbanization (houses, streets), vegetation (trees, grass, gardens, parks), and water.
The forecasting of thermal human comfort requires simulations on very high spatial and temporal scales, posing both a meteorological and an eScience challenge. Observational data to validate the weather forecasts at street level are based on a network of weather stations in two cities: Wageningen and Amsterdam. This network has a fine spatial detail compared to other networks in the research field and allows for an evaluation of the forecasted temperature and humidity. Bike traverse measurements of temperature, humidity, wind speed, and radiation are taken regularly. These observations resolve a high degree of spatial detail in urban weather and contribute to an understanding of urban climate.
Next to the proposed research, dissemination of results via the Internet and social media will be explored. Forecasting human thermal comfort on street level is novel and will be of added value for public health and society in general. The first step is to process the urban morphological data into parameters relevant to the weather forecasting model. The mesoscale meteorological model WRF (Weather Research and Forecasting) can then produce fine spatial scale weather forecasts. These forecasts are downscaled to the street scale using knowledge of the available geo-information. Truly a Big Data challenge.
Kernel Tuner greatly simplifies the development of highly-optimized and auto-tuned CUDA, OpenCL, and C code, supporting many advanced use-cases and optimization strategies that speed up the auto-tuning process.