On 13 and 14 June 2019, the iLeaps workshop on Modelling and observing urban fluxes was held at the Institute for Advanced Metropolitan Solutions (AMS) in Amsterdam (Netherlands). The workshop had 25 attendees ranging from professors to MSc students.
The workshop was started by a presentation by Henk Wolfert as a representative of the host institute (AMS). He summarized the history of the institute as a collaboration between Wageningen University, Delft University of technology an MIT (Boston) that was asked by the municipality of Amsterdam to bring in and develop technology for the current and future challenges Amsterdam faces.
The first keynote speaker was Leena Jarvi from Helsinki University who presented work on the modelling of urban fluxes in cold conditions. Only a single past study on the role of snow on urban fluxes in Denver was published n 1985 before. She showed the role of snow cover on the flux exchange on one hand die to the different albedo and secondly due to the thermal isolating effect of snow. Moreover snow governs the partitioning of the sensible and latent het flux. Urban models appear to give large biases compared to two flux towers in and around Helsinki due to missing parameters like building parameters which is due to special architecture related to cold areas. Moreover she presented a strategy the unravel the CO2 flux in biogenic, building heating and traffic contributions.
Sebastian Schubert (Humboldt Universtat, Berlin) presented the development of the PALM4U large eddy simulation model for the urban environment as developed in the national Urban Climate Under Change research program. The model is covers a building resolved flow and allows for estimating of urban wind speed, pollutant dispersion, human thermal comfort and even Vitamine D production. Simulations were demonstrated for a domain covering Berlin with 2 nested domains (one of 50 x 50 km at 15 m resolution and a nested one with 1 x 1 km at 1 m resolution). This domain can be run for 24 h lead time in 24 h wall clock time at 6048 computing cores. The model has been validated against wind tunnel experiment at Hamburg University for the Hamburg Hafencity area, with very satisfying results. The model is being further developed as a community model and to be driven by ICON and WRF model output.
Bert Heusinkveld from the Wageningen University presented his results of one year of observing fluxes above the city center of Amsterdam, and 5 years of observation from a network of 24 weather stations across the city. Apart from the well known urban heat island effect, the observations also indicate an urban cool island effect in the early morning, and a moisture island also in the early morning. He addressed the question what should be a reasonable averaging time of the fluxes and he showed that at a scale of about 1 h the fluxes start to converge. Surprisingly the observations indicate negative sensible heat fluxes in wintertime, which needs to be further investigated.
Finally Valery Masson from CNRM presented work on the modelling of CO2 fluxes for Toulouse and Helsinki depending on a) meteorology, such as buildings and vegetation and b) other factors such as traffic, peoples respiration. For the latter the CO2 flux rate is modelled as a mean rate scaled with factor representing month day and time of the day. Furthermore he presented work about the updated version of the town energy model in which for building uses was implemented as well as peoples behaviour. The mix of uses substantially improves the performance for Weekdays , Sundays, but not yet for Saturdays. Future research lines will examine the use of urban databases and LCZ, the architecture across France varies substantially, the modelling of the inhabitants behaviour and the modelling of building physics. The conclusion was that also nowadays the data science is more critical than the meteorological knowledge.
Gert-Jan Steeneveld from Wageningen University presented the results of the SUBLIME model intercomparison. After a revision of the modelling recipe model results should look more physical. It was shown that the modelled 50 m air temperature diverges substantially up to 6 K for maximum temperature. Modelled results appears to be in reasonable agreement with observations, especially for wind speed profiles. We also learnt that providing more and more detail concerning the urban morphological parameters helps from stage 0 to stage 1 of the model intercomparison experiment, but not anymore from stage 1 (local climate zone 2 values) to stage 2 (full details). Model results will be further analysed in the frame work of process diagrams that will help to assist to explain the differences between the submitted runs.
Several 20-min presentations were given by the participants:
Jorge Amorim (SMHI): presented UrbanSIS: an C3S contribution for European cities. It contains a downscaling of ERA Interim to 1 km via ALADIN and UrbanSIS, partly online, partly offline modelled. A proof of concept was presented for Stockholm (Sweden).
Umer Alvi presented PhD work on a method to estimate urban air temperatures from satellite data for Turkku (Finland) in comparison with a set of HOBO field observations, and using different land use maps at different resolutions CORINE appears to outperform other map products. The audience raised concerns that LANDSAT data are only available once in 16 days.
Zhaowu Yu presented his postdoc work on the spatial scales of interventions that are relevant for climate services in the city, the threshold value for Efficiency. The questions addressed are where to cool and how to cool (cooling distance and threshold size).
Steven Caluwaerts (U Gent) presented the concept of networks of networks in which urban network of different cities are connected to exchange knowledge and experience about network performance. Also he presented a new project on crowdsourcing at 56 schools across Flanders, and a downsclaing study using ALARO.
Stephan Stockl (UNI innsbruck) extended the SUBLIME model experiment by using the SUBLIME model output to a Langranian dispersion model. First SUBLIME model results has to be used to generate input of turbulent variables. The dispersion parameters differ greatly between the default values of model when compared to the dispersion parameters obtained using SUBLIME model meteorological input.
Mathew Lipson presented his work on the offline urban canopy model testing for Melbourne using different offline contributions for building heating, electricity etc. His talk continued with Conformal-Cubic Atmospheric Model (CSIRO) to feed a column version of the model to estimate effects in different climate scenarios.
Finally Matthias Demuzere gave a update in the field of mapping Local Climate Zones that are now present for Europe. The question was raised how to extend this to the whole globe and make the approach applicable in weather forecast and climate models.
Links to presentations