precipitation extremes over the netherlands: changes due to climate and urbanization
Vahid Rahimpour Golroudbary is a PhD student in the Department of Water Resources. His supervisor is prof.dr. Z. Su from Faculty of Geo-information Science and Earth Observation.
The detection of changes in weather is important for the preparedness of societies for extreme events due to climate and urbanisation. Knowledge of the behaviour of extreme events is needed for many practical problems, as most infrastructures were designed with the assumption of a stationary climate (i.e., constant properties over time) and are sensitive to extreme weather events such as extreme precipitation. Precipitation extremes affect runoff volumes, infrastructures, aquatic ecosystems and species and, most notably, human life. Current trends in extreme precipitation due to, for example, climate and urbanisation, could violate the time-invariant assumptions for design criteria, which might lead to more uncertainty in return level estimations and a consequential increase in the failure probability of infrastructures. Although variations in meteorological parameters are widely accepted, stationary assumptions are commonly used to design infrastructure. This study aims to add to the knowledge of changes in precipitation and the nonstationary behaviour of precipitation due to climate and urbanisation.
The main objective of this research was to understand and quantify the effects of urbanisation and climate change on precipitation in the Netherlands. Consideration was given to the impacts of urbanisation on precipitation extremes via the comparison of observations over urban and non-urban areas using trend detection and attribution analysis approaches. The EVT offers statistical properties for the distribution of extremes in sufficiently large time series and further provides a probabilistic distribution of extreme events for quantifying the return levels of a given return period. In this respect, the block maxima approach was applied to characterize the behaviour of extreme precipitation, and the probability of the maxima was determined for each block. The GEV function fitted to the obtained maxima of blocks, and subsequently, the return levels for different return periods were estimated. A common challenge in nonstationary distributions is the selection of suitable covariates to explain variations, here variations in extreme precipitation. Therefore, seasonal models based on the sinusoidal pattern of precipitation occurrences and the correlation of these occurrences with other climatological parameters were studied. Although the nonstationary estimations had less uncertainty than the stationary estimations, some uncertainty remained due to the influences of other forcing factors and spatial dependencies.
The study attempted to characterize and attribute the seasonal variation of daily extreme precipitation events in the Netherlands (Chapter 2). Statistical models for extreme values were used to fit daily rainfall maxima for all months during the period of 1961-2014 using data from 231 rain gauges distributed across the country. The climatological spatial variability of extreme precipitation was illustrated on a daily basis. A harmonic model for all monthly maxima was adopted instead of individual models. The fitted distributions show the inaccuracy of stationary assumptions for estimating return levels. The nonstationary model estimated parameters with less uncertainty and with smaller CIs than the stationary model, thus permitting a more accurate representation of extreme precipitation in the Netherlands. The spatial pattern of the annual mean location and scale of the GEV parameters was compatible with the coastal land cover (such as the wooded and heathland areas of the Veluwe region of Gelderland province) and orography (in the southeast of the country). The regional differences for the location parameter peaked over the west coast, especially over the central west coast during the summer half-year (between June and November), while the centre and east of the country had the highest regional differences during the winter half-year (between December and May). The scale parameter peaked in the centre of the country during the summer and was highest in the east in the early summer and along the west coast in the spring. The spatial distribution of the extreme event probability clearly reflects regional differences in the Netherlands.
In Chapter 3, index analysis performed on the precipitation data for one 54-year period and two independent multi-decadal periods displayed positive signals for the frequency and intensity of extreme events. An analysis of extreme precipitation trends for each index revealed the existence of spatial differences in the number and magnitude of extreme events over the Netherlands. The increases in extreme precipitation were high in comparison to the increases in annual total precipitation. Significant differences were found between an earlier multi-decadal period and a recent multi-decadal period in the Netherlands. The differences in the monthly maximum daily precipitation amounts and trends between the two multi-decadal periods were reflected by higher precipitation values in the late summer and autumn. The significant changes in different indices indicate that severe precipitation events were not distributed homogeneously across the study area. The possible effects of land use on extreme precipitation were assessed by quantifying the differences between urban and rural rain gauge stations using a spatial gridding method. The data from all the categorized stations show that urban areas receive more intense extreme precipitation than rural areas, but this discrepancy is rarely significant and depends on the investigated region. Relative to other areas in the Netherlands, urban areas in the western populated regions of the country exhibit prominent urban land use influences on extreme precipitation patterns. Five years of basic weather observations (2011-2015) from the automatic and amateur networks received much attention during an analysis of urban effects on meteorological parameters, with a focus on temperature and precipitation (Chapter 4). The hourly analysis indicates that the UHI effect is a nocturnal phenomenon in the Netherlands. The role of UHIs was more prominent after sunset, when the effect had a magnitude of over 2˚C. The seasonal analysis revealed that the UHI occurred in all seasons during the year, not only during summer. However, the UHI intensities in summer were higher than those in other seasons. The significant linear relationship between UHI intensity and population density suggest that a highly dense population influences the UHI intensity. Although there was no clear relationship between precipitation and other basic meteorological parameters, the hourly precipitation increased up to 7% more in urban areas than in rural areas. Furthermore, the precipitation difference between urban and rural areas was the highest after sunrise in the morning. The maximum precipitation occurrences were more likely to be greater in the urban stations than in the paired rural stations. The distribution of hourly precipitation showed the highest location parameter value for August (with an 11% increment in urban areas). In addition, a distinct seasonal cycle for the precipitation based on the UHI demonstrated the maximum UHI and precipitation that occurred in summer. Faster warming occurred in spring, and quicker cooling occurred in autumn. The findings indicate a distinct seasonal cycle for temperature and precipitation in Dutch residential areas and are in agreement with results obtained by amateur weather stations and the scientific literature.
The scaling of extreme precipitation with temperature was performed using the statistical quantile regression and binning methods described in Chapter 5. Positive 3-7% scaling rates at most stations were found between extreme precipitation and dew point and air temperature from 1985 to 2014 throughout the country. The stationary model was improved by modelling the nonstationary behaviour of extreme precipitation associated with the dew point and atmospheric air temperatures, as well as the NAO index. A higher precipitation for urban than for non-urban areas was found by considering the land cover upwind from the stations. The return levels for the monthly maximum daily precipitation were estimated to be 5-7% higher in urban areas than in non-urban areas in August. Likewise, the IDF curves were improved by nonstationary estimations associated with the dew point and atmospheric air temperatures of urban and non-urban areas. The results showed a higher frequency and intensity of extreme precipitation events in urban areas than in non-urban areas for short durations. The study concluded that nonstationary models refine IDF curves and should be used for analysing extreme precipitation by considering the probable covariates due to external forces (such as large-scale modes, circulation types and temperature changes). This study raises the need for a better estimation of changes in the frequency and intensity of extreme precipitation and the acceptance of nonstationary behaviour in the context of urbanisation and climate variabilities. Overall, although the effects of urban areas on extreme precipitation in the Netherlands seems not to be statistically significant, a consistent picture regarding the sign (i.e. positive) was presented throughout this study. While the differences are not very large they are persistent and could be important. It should be noted that for drawing firm conclusion, new observations in the case study with similar local conditions (e.g., some measurements in rural areas and a few in the urban-influenced neighbours and others far away) are needed to better quantify the instrumental errors and uncertainties.
