Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Stedelijke hitte-eilanden ofwel urban heat islands (UHI) zijn letterlijk en figuurlijk een hot topic. In de stad is het warmer dan op het platteland en dit verschil kan in grote steden wel oplopen tot maximaal 10°C. De interesse in dit fenomeen is sterk toegenomen omdat verwacht wordt dat de klimaatsverandering intensere UHI zal generen. Studies specifiek voor België zijn echter schaars. Het stedelijk hitte-eiland is tot nu toe enkel voor de drie grootsteden Gent, Brussel en Antwerpen bepaald. Nochtans kunnen ook kleinere kernen van bebouwing weliswaar kleinere hitte-eilanden generen. Het hoofddoel van deze thesis is dan ook het kwantificeren van stedelijke hitte-eilanden voor zoveel mogelijk plaatsen in België. Morfologische parameters zoals de fractie verstening, de fractie vegetatie en de populatiedensiteit zijn veel gebruikte UHI-indicatoren en geven zo een idee over de sterkte van het UHI. De mogelijke correlatie tussen de gemiddelde minimumtemperatuur en deze parameters wordt dan ook in deze thesis onderzocht. Het UHI wordt bepaald als het verschil in minimumtemperatuur tussen een stedelijk en ruraal station. In totaal werden van 179 stations temperatuurgegevens verkregen, zowel van het KMI als van hobbymeteorologen. Deze gegevens werden na de kwaliteitscontroles onderworpen aan selectiecriteria. Er zijn immers nog andere factoren die de temperatuur beïnvloeden zoals bos, water en topografie. Stations waarbij deze factoren een te sterke invloed hebben op het temperatuursignaal werden verworpen om zo goed mogelijk enkel het signaal van de bebouwing over te houden. In totaal werden zo 82 stations meegenomen in de verdere analyse.Er is ook in België een statistisch significante relatie aangetoond tussen de gemiddelde minimumtemperatuur en de fractie verstening en de fractie vegetatie. Beiden verklaren ze respectievelijk 15% en 18% van de variatie in de gemiddelde minimumtemperatuur. Doordat de fractie verstening en fractie vegetatie onderling sterk gecorreleerd zijn (R²=0.6) moeten deze indicatoren als complementair aanzien worden. De populatiedensiteit verklaart 0% van de variatie en is dus geen goede UHI-indicator. In totaal werden 57 UHI berekend. De bepaalde temperatuurverschillen kunnen echter niet verklaard worden door de bebouwing. Dit maakt dat enkel een schatting kan gemaakt worden van het UHI op basis van de relatie tussen de gemiddelde minimumtemperatuur en verstening. Het UHI tussen een station met 0% en 100% verstening in een 500 m-radius is in België 1,95°C. De zwakke correlatie tussen de verstening en de gemiddelde minimumtemperatuur kan verklaard worden doordat er amper stations in de analyse waren met een hoge versteningsgraad. Weerstations specifiek in kernen van bebouwing plaatsen is aanbevolen om een betere schatting te kunnen maken van het UHI. De huidige KMI-weerstations voldoen niet aan die noden.

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

This study presents an application of a distributed-snow model in order to gain knowledge on the distribution of snow in a mountainous area in the semi-arid Andes of Chile. Data from eleven automatic weather stations is at disposal to drive the SnowModel. Lack of direct observations (for precipitation) and interpolation issues (for wind speed and direction) pose difficulties for obtaining a proper meteorological forcing for the model. This is a common issue in remote environments with complex topography. In order to resolve these problems, alternative input datasets for wind and precipitation were developed and used as forcing for the model. A sensitivity analysis was done in order to gain understandings about the effect of these different wind and precipitation inputs on output of snow distribution and local snow depth. The choice of wind field had a greater impact on the performance of snow distribution simulations than altering the precipitation forcing. The performance on the model to simulate local snow depth magnitude was however more dependent on the chosen precipitation pattern. All the different wind and precipitation inputs were combined into twenty-five possible input combinations. These combinations are evaluated on their model performance by means of different validation techniques. The best overall performance was obtained by combining (a) a precipitation input obtained from snow depth increases during demarcated meteorological conditions and (b) a wind field obtained from the wind speed rescaling of a WRF (Weather Research and Forecasting model) wind grid. Although satisfying results were obtained with certain wind-precipitation combinations there should be sought to improve wind forcing so that fine-scale wind variations can also be reproduced.

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

An increasingly higher proportion of regional climate model simulations are being performed at convection permitting scales (CPS), i.e., a horizontal grid scale below 4 km. At this scale, deep convection becomes a grid-scale process and is resolved explicitly, rather than parameterized. Several recent studies have investigated the added value of CPS simulations, focusing mainly on the benefits for the representation of summertime convective precipitation. However, some research gaps or areas of disagreement continue to exist. For one, the added value of CPS in studies that use regional climate models to quantify the biogeophysical impact of land-use change (LUC) on climate has not yet been investigated. Given the fact that more and more LUC impact studies that use regional climate models are opting for CPS simulations, such an investigation is timely. Second, considerable disagreement exists in existing literature on the added value of CPS for extreme precipitation on the sub-daily timescale. Specifically, studies that compare the future change signal in extreme precipitation on the hourly timescale of CPS to conventional, non-convection permitting scale (nCPS) simulations have produced mixed results, with some reporting that CPS simulations project a higher increase in extreme hourly accumulations, while others do not. Further research into this topic is therefore needed, especially given the considerable societal impact of short-duration extreme precipitation. Therefore, the goal of this dissertation is to investigate the added value of regional climate model simulations at CPS in the areas of a) LUC impact assessment and b) greenhouse gas emission induced future changes in extreme precipitation.We start our analysis by presenting a new methodology to evaluate the impact of LUC in a regional climate model. The goal in developing this methodology is to address several deficiencies in current LUC impact evaluations. The methodology uses site pair observations from across Europe, and is applied to the regional climate model COSMO-CLM in nCPS configuration (0.22°, ~25 km horizontal resolution). Also, we choose to limit the investigation to one specific LUC, namely deforestation. The observational dataset shows that a considerable seasonal and diurnal cycle exists in the temperature difference between forest and nearby open land sites. COSMO-CLM simulations performed at nCPS resolution are able to reproduce the sign and magnitude of this temperature signal during winter daytime only. An analysis of the difference in the various surface energy budget components shows that the main summer daytime model process related bias is an unrealistic decrease in low and mid-level cloud cover and associated increase in incoming shortwave radiation (δSWin) over open land, triggering additional biases in the latent heat flux and ground heat flux. Additionally, the model’s inability to reproduce the observed nighttime cooling is attributed mainly to a bias in the difference in incoming longwave radiation (δLWin).We then apply our evaluation methodology to both nCPS (0.22° or ~25 km horizontal resolution) and CPS (0.025° or ~2.8 km horizontal resolution) COSMO-CLM simulations. These runs are evaluated against one high quality forest/open land site pair located in eastern Germany. For summer daytime, results show that CPS simulations improve upon nCPS simulations in several aspects. First, δSWin is significantly reduced compared to nCPS simulations, an improvement that can be linked to a better representation of afternoon convective clouds. Second, improvements are seen in the simulated difference in sensible heat flux (δH) and ground flux (δG), which can be linked to a better representation of surface roughness, as well as surface energy budget and atmospheric feedbacks resulting from the improvements in both δSWin and roughness. For nighttime, (smaller) improvements are seen as well. Most notably, the bias in nighttime δLWin is reduced, which could linked to the increase in vertical resolution, and an associated improvement in the difference in lower boundary layer vertical mixing. Thus, we conclude that, overall, CPS simulations improve considerably upon nCPS simulations in simulating the local biogeophysical climate impact of deforestation.Next, we investigate the added value of CPS in simulating both present day extreme precipitation and the future change in extreme precipitation driven by greenhouse gas emissions. Six multi-decadal COSMO-CLM regional climate simulations are performed, namely one set of evaluation, control and future simulations at nCPS (0.11° or ~12 km horizontal resolution), and another set of simulations at CPS (0.025° or ~2.8 km horizontal resolution). The study area is Belgium and surroundings. Present day evaluation results show that CPS simulations are vastly superior in simulating the frequency of extreme precipitation on the hourly timescale, but not on the daily timescale. CPS biases on the daily timescale are linked to small convective cells still being under-resolved at 0.025° resolution.Finally, the difference in the future precipitation change signal is analyzed. When comparing the projected future change in extreme precipitation of nCPS to CPS simulations, considerable differences exist. On the daily timescale, both nCPS and CPS simulations project that extreme precipitation will increase. However, the projected increase is significantly higher at CPS, especially for the most extreme daily precipitation events. With regards to the spatial pattern of the future increase on the daily timescale, no clear link to topography or any other geographic feature can be discerned for either nCPS or CPS. On the hourly timescale, CPS simulations project large increases in extreme precipitation for both Flanders and Ardennes sub-regions, with the highest increases seen over the latter region. nCPS simulations are able to match this increase in the hilly Ardennes region, but not in the relatively flat Flanders regions, where they project a decrease in extreme hourly precipitation for most of the intensity range. We can therefore conclude that on the hourly timescale, the added value of CPS depends on topography. This result explains the discrepancy in available literature, as the studies reporting little difference in the future increase of hourly extreme precipitation events between nCPS and CPS have all been performed for hilly or mountainous terrain.

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Clouds have a significant effect on air quality, among others by affecti ng photolysis (through their effect on shortwave radiation), and because of their role in chemical reactions that occur at the interface of clou d droplets. Yet, atmospheric models experience great difficulty in simul ating cloud-related variables correctly, which regularly leads to poorly simulated concentrations of atmospheric pollutants such as tropospheric ozone and fine particulate matter. Considering this, the main goal of t he proposed research is to improve the simulation of cloud in atmospheri c models by exploiting satellite data in a data assimilation procedure, and to demonstrate the impact of an improved representation of cloud on simulated air quality at the regional scale. The methodology will be as follows. In a first step, relevant satellite data will be evaluated and selected. Criteria for selection will be based, among others, on the type of geophysic al parameter the instrument can provide (e.g., cloud top temperature, op tical depth, ...), as well as other observation characteristics (e.g., s patial and temporal resolution). Next, the selected data will be process ed to yield the required information (e.g., vertical profiles of cloud l iquid/ice water). Resulting profiles will be compared with data from an independent source, e.g. from recently launched satellite instruments de veloped for experimental cloud studies. Retrieved cloud parameter profil es will then be inserted into the mesoscale meteorological model Advance d Regional Prediction System (ARPS) using data assimilation techniques. Subsequently, the impact of assimilating cloud data into ARPS will be ev aluated, by confronting simulated values of relevant quantities (e.g., s urface shortwave radiation, cloud liquid water content) with observed va lues. Following the succesful validation of ARPS, the urban/regional air quality model AURORA will be enhanced to fully benefit from the im proved cloud simulations. This will be achieved by improving parameteris ations that calculate photolysis coefficients. Finally, the accuracy of regional air quality simulated with the improved AURORA model will be ev aluated, focusing on ground-level pollutant concentrations of ozone.&nbs p; It is expected that a successfull a chievement of the objectives will lead to significantly improved air qua lity simulations with the AURORA model, especially for ozone.