The recent decade has seen a rapid adoption and advancement of methods to use and interpret eddy-covariance (EC) flux measurements of carbon dioxide (CO2) in urban environments. Although several studies demonstrate potential for using EC measurements to validate fine-scale emission inventories of CO2, a major challenge remains the source attribution of total measured mass fluxes of CO2 to specific emission sources. The EC-measured fluxes represent the sum of various fossil fuel combustion processes, respiration and photosynthesis. Information on the stable isotope composition of the CO2 could add promising additional information to complement EC flux measurements of CO2 in urban environments.

In this study, we measured the three most abundant CO2-isotopologues at high frequency directly in-situ in the urban atmosphere using a spectroscopy system (TGA 200, Campbell Scientific Inc., Logan UT) and calculated isotopic ratios d13C and d18O at high frequency. Every 10 minutes, the system is calibrated against three reference tanks of known d13C and d18O. Because d13C and d18O vary between representative fuel emission sources (gasoline, diesel, natural gas) and biogenic sources (human and ecosystem respiration) we can use mixing models to attribute sources based on excess concentrations and/or directly complement fluxes with isotopic information at high frequency. While d13C depends on the fuel type and origin (in our study: d13C gasoline 27.2‰; diesel -28.8‰; natural gas -41.6‰), d18O is fractionated in catalytic converters (d18O gasoline vehicles -12.5‰; diesel -18.6‰; natural gas -22.7‰) and exhibits higher variability between samples of biogenic sources to different source mixes of the oxygen.

We sampled air from the top of the 30-m flux tower at Vancouver-Sunset, located in between a busy intersection on one wind sector side, and a residential area characterized by emissions from natural gas furnaces in another sector. We explored how combining isotopic information with eddy covariance data allows us to partition fluxes from a specific wind direction. In a long-term ensemble we can separate between natural gas, gasoline and respiration sources. Results are independently tested against a fine-scale CO2 emission inventory via appropriate source area models, and a statistical-empirical partitioning approach developed for this site by Crawford and Christen (2014, Theor. Appl. Clim.).

Accurate air temperature observations are important for urban meteorology, i.e. to study the urban heat island and adverse effects of high temperatures on human health. Usually, the number of routinely available temperature observations is rather limited. We present a method to derive temperature information for the urban canopy from an alternative source: smartphones. Battery temperature data were collected by users of an Android application for smartphones (opensignal.com). The application automatically sends battery temperature data to a server for storage. A regression model, based on a physical model, is employed to retrieve daily air temperatures from battery temperatures. from a meteorological station of an airport located near the city and from an urban meteorological network in the city. In this study we apply this technique for rural and urban sites in and around Amsterdam (The Netherlands). The evolution of the retrieved air temperatures correspond well with the observations. The mean absolute error of daily air temperatures amounts to 1.4 K, and the bias amounts to 0.4 K. This shows that monitoring air temperatures employing an Android application holds great promise. Finally, we use temperature observations obtained from this technique to validate high resolution WRF mesoscale modeling results over Amsterdam for a warm summer period.

There are a variety models and methods that can characterize and quantify the global increase in carbon dioxide (CO2) mixing ratios in the atmosphere and their links to anthropogenic emissions and terrestrial and oceanic sinks. However, mapping, visualizing and monitoring CO2 emissions and sequestration at policy relevant scales - the local to urban scale - remains a challenge. Emerging modular open source technologies are allowing for the miniaturization, mobilization and increasing accessibility of sensor systems and may be a promising way to observe and map CO2 mixing ratios across heterogenous and complex landscapes, including urban environments. These measurements can serve as either inputs for inverse modelling of emissions and for the visualization of urban emissions in the context of science-communication to public.

We present a system for monitoring CO2 mixing ratios in cities using a network of mobile CO2 sensors deployable on vehicles and bikes combined with geo-spatial analysis and visualization tools. We used components from Arduino (Arduino CC, Italy) coupled with the Li-Cor Li-820 infrared gas analyzer (Licor Inc, lincoln, NB, USA) to prototype a portable CO2 analyzer for possible future CO2 monitoring, and pollution mapping in general. We tested our experimental methodology by deploying a fleet of sensors in the city of Vancouver, Canada to determine excess urban CO2 mixing ratios (i.e. the ’urban CO2 dome’) when compared to values measured at a fixed, upwind, remote background site. Sensors were deployed both in fixed locations and in a mobile campaign using bikes and volunteered vehicles. In the presentation we examine the strengths and weaknesses of such a mobile system to characterize CO2 mixing ratios across a complex urban environment and discuss the spatial biases of the two methods. The presentation will focus on lessons learned from the pilot study and discuss the methodology for processing and analyzing spatially and temporally discontinuous measurements using a web-based visualization tool.

In recent years there has been an increasing interest to explore the potential of low-cost measurement devices and mobile measurements as alternative source of meteorological monitoring data, especially in the urban areas where high-density observations become crucial for appropriate heat load assessment. Non-standardized data collecting procedure, instrument quality, their response-time and design, variable device ventilation and radiation protection influence the reliability of the gathered data. We investigate what accuracy can be expected from the data collected through low-cost mobile measurements and whether the achieved quality of the data is sufficient for validation of the state-of-the-art local-scale climate model. Combining high-resolution urban climate model simulations and dense network of local air temperature measurements may help identify hot spots in urban areas and bring added value in heat warning systems.

We use the dynamical urban climate model MUKLIMO_3 (horizontal resolution of 100 m) to simulate the urban heat load during day-time conditions in the area of Vienna. The model is initiated with the vertical profile of temperature, relative humidity and wind from the operational weather forecast model ALARO-ALADIN (horizontal resolution of 4.8 km). Daily simulations for the entire summer seasons April-October 2011, 2012 and 2013 are performed. Maximum temperatures were evaluated against the monitoring data from the 9 operational weather stations showing an average error below 2°C. The mobile measurements took place on a clear-sky, dry and hot day in July 2011 by bicycles and in July, 2013 by car. Several low-cost devices were tested: Maxim iButton, OnsetHOBO UX100-003 and self-designed solar powered Arduino-based data loggers combined with the Sensirion SHT21 temperature and humidity sensor. The devices were calibrated and tested in stationary mode at the Austrian Weather Service showing accuracy between 0.1◦C and 0.8◦C. In mobile mode, the best response-time was found for self-designed device with Arduino-based data logger and Sensirion SHT21 sensor. The collected data were aggregated on a 100 m horizontal resolution grid and compared with the simulations initialized with the atmospheric conditions for the given day. Both measurement and modelling results show similar spatial gradients for distinct local climate zones. In case of bike measurements the average difference between the modelled and measured temperatures was 1.3°C and 1.1°C when compared to the operational monitoring stations.

The Netherlands has a mild mid-latitude climate. Meteorological records for The Netherlands show that the number of hot summer days has increased, and future climate change projections predict the same trend. Heat stress is the major cause of weather-related urban human mortality. The urban heat island effect is significant for the Netherlands (summertime nocturnal UHI 95% >7 K) (Steeneveld et al., 2011, Heusinkveld et al., 2014) as more than 80% of the Dutch population live in cities and are thus subject to such added stress. For human thermal comfort during heat waves, shading is more important than wind according to Mayer and Höppe, 1987. However, for the Netherlands wind may also be relevant due to the proximity of the sea and large lake bodies. Here, measurements and analysis results are presented using an innovative mobile measurement system and a dense urban weather station network. The mobile measurements were used to assess the spatial variability of human thermal comfort (Heusinkveld, et al., 2010 & 2014). A key feature of the mobile measurement system is the direct measurement of mean radiant temperature and wind speed. To do so, a special cargo bicycle was equipped with 6 pyranometers, 6 pyrgeometers, 2D wind speed/direction, temperature, humidity, bicycle speed and GPS sensors. Mobile measurements can provide great spatial detail from a large set of sensors. However temporal resolution is limited and therefore a dense urban weather station network of temperature/humidity and wind speed was set up. Within a city the lower average wind speed increases the radiation induced temperature error of a thermometer screen. To minimize such errors, all air temperature/humidity sensors used on the mobile and urban weather stations were equipped with aspirated thermometer screens.

References:

Heusinkveld BG, LWA van Hove, CMJ Jacobs, GJ Steeneveld, JA Elbers, EJ Moors, AAM Holtslag (2010) Use of a mobile platform for assessing urban heat stress in Rotterdam, Proceedings of the 7th Conference on Biometeorology. Instituts der Albert-Ludwigs-Universität Freiburg 20 (2010). - ISSN 1435-618X - p. 433–438. Freiburg: 2010.

Heusinkveld, B.G. , Steeneveld, G.J. , Hove, L.W.A. van , Jacobs, C.M.J. , Holtslag, A.A.M., 2014: Spatial variability of the Rotterdam urban heat island as influenced by urban land use. Journal of Geophysical Research - Atmospheres 119, 677 - 692.

Mayer H, Hoppe P (1987) Thermal comfort of man in different urban environments, Theor Appl Clim 38: 43-49.

Steeneveld, G.J., S. Koopmans, B.G. Heusinkveld, L.W.A. van Hove, and A.A.M. Holtslag, 2011: Quantifying urban heat island effects and human comfort for cities of variable size and urban morphology in The Netherlands., J. Geophys. Res., 116, D20129, doi:10.1029/2011JD015988.

Eddy covariance measurements are increasingly used to quantify carbon dioxide fluxes in urban areas. The net carbon dioxide fluxes are the sum of anthropogenic emissions, biogenic carbon release from plant and soil respiration, and biogenic carbon uptake by plant photosynthesis. In natural environments, such as forests and grasslands, the partitioning of biogenic fluxes is well established. In contrast, the partitioning of net carbon dioxide fluxes in urban environments is more difficult due to the multitude of sources and sinks, and the spatial variability of emissions and uptake. Flux partitioning approaches that have previously been applied in urban environments include modeling fluxes based on surface fractions in the changing turbulent source area, and fusion with bottom up emission inventories.

Here, we present a statistical partitioning approach that uses filtering and empirical mode decomposition to quantify the components of the net carbon dioxide flux in a step-wise fashion. We applied the novel approach to a three year time series of measurements from a tall broadcast tower in a suburban neighborhood of Minneapolis-Saint Paul, Minnesota, USA. During winter, local vehicle counts were significantly correlated (r=0.68–0.91) with weekday net carbon dioxide fluxes over source areas that intersected roads, which enabled us to estimate vehicular emissions throughout the year. Modeled natural gas emissions from space heating based on hourly air temperature differences were significantly correlated (r=0.27) with the net carbon dioxide fluxes throughout the year and subsequently removed from the net fluxes. The above methods will give us a robust estimate of the anthropogenic flux which we can remove from the net carbon dioxide flux. The resulting biogenic flux is then partitioned with common methods of flux partitioning developed for natural environments. Soil respiration fluxes estimated in this way can be compared to modeled fluxes based on observations from a mobile tower over a turfgrass lawn within the footprint of the tall tower. The estimates of anthropogenic carbon emission components are needed to validate building energy and traffic models, while estimation of biogenic carbon release and carbon uptake are useful to identify areas with emission-reduction potentials and establish emission baselines.

Carbon dioxide storage and its governing processes in the urban environment are poorly characterised compared to vertical fluxes, potentially leading to under-estimation of emissions during periods of low turbulence. Some studies apply a flux correction using only data from the flux measurement height (zh), however the assumptions required are not valid during periods where measurements in the inertial sub-layer are likely to be decoupled from the ground. Although CO2 stored below zh can be calculated if the vertical CO2 concentration profile below zh is known, this is often not measured due to cost and access restrictions. In this presentation, guidelines are suggested for the required number and placement of sample points for CO2 storage measurements in a deep urban street canyon, developed using data collected between 2011 and 2014 in Central London, UK. Also considered are the effect of sensor response and sampling interval, and at the processing stage, three different temporal and spatial interpolation methods are evaluated against measured data. A method of independently benchmarking the quality of CO2 storage values using wavelet power spectra is proposed.

Instrumentation consisted of two open path gas analysers measuring at 10 Hz, one within the inertial sub-layer and one at half-canyon height, and a profile system of two closed path gas analysers connected to valve arrays which allowed air to sampled at 16 different locations sequentially. Co-located inlets were used to assess the stability of each closed path gas analyser with each other and the open path gas analysers. The vertical CO2 concentration profile was found to differ substantially from that typically reported in rural environments, requiring few measurements for accurate calculation of the CO2 storage, however calculated CO2 storage was sensitive to sampling interval. The power spectra of the CO2 storage time series varied according to a power law with frequency, and was anti-persistent with the majority of the energy in the signal at high frequencies.