Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → Afrique du Sud → 2016 → A comparative analysis of long-term variations of temperature and rainfall in rural and urban areas, and their effects on the estimation of design storms in Kenya

University of the Western Cape (2016)

A comparative analysis of long-term variations of temperature and rainfall in rural and urban areas, and their effects on the estimation of design storms in Kenya

Gachahi, Lydiah Wangechi

Titre : A comparative analysis of long-term variations of temperature and rainfall in rural and urban areas, and their effects on the estimation of design storms in Kenya

Auteur : Gachahi, Lydiah Wangechi

Université de soutenance : University of the Western Cape

Grade : Philosophiae Doctor - PhD (Earth Science) 2016

Résumé partiel
My Thesis aimed at expanding the current knowledge on how variations of temperature characteristics including the possible existence of urban heat islands (UHI) over urban areas of Kenya could be influencing rainfall characteristics, and to examine if the stationary extreme value distributionis still suitable for modeling urban storm designs in view of the global climate change. My hypothesis was that the floodingoccurring frequently in major urban areas of Kenya are due to increased rainfall caused by global climate change, and the urban heat island (UHI) effect. To put this perception into perspective, temperature and rainfall characteristics and their inter-relationships, of four of the major urban areas in Kenya namely, Nairobi, Mombasa, Kisumu, and Nakuru, were investigated. I obtained data from meteorological stations in and around each urban area, which had at least thirty (30) years of continuous monthly (or daily) temperatures and rainfall values, from the Kenya Meteorological Department. I checked the datasets for quality and missing values and adjusted where necessary before commencing with analysis. I sourced other supporting global dataset from various websites’ data banks.I used various methods of data analysis which included ; i) exploratory data analysis techniques such as the continuous wavelet transform (CWT), geographical information system (GIS) maps, and visual time series plots. In particular and unique in my Thesis was the use of the CWT method as a diagnostic tool to examine non-stationaritiesand variability of temperature and rainfall time series. The use of land surface temperature data to investigate UHIs was also adopted to supplement the air temperature ; ii) statistical methodsincluding parametric (linear and quantile regression) and non-parametric (Mann-Kendal) trend tests, Pearson’s correlation analysis and the extreme value analysis. Statistical hypotheses were stated and testedat the 5% level of significance. Particularly unique in the statistical analysis is the use of the quantile regression which has not been used to investigate trends of the highly variable rainfall of the equatorial East African (EEA) region, and the generalized extreme value analysis (GEV) analysis which has virtually no published literature for the EEA rainfall extreme value analysis. Most of the analysis was carried out in the R environment. I established that ; i) there is warming due to urbanization as well as global warming within and in the neighbourhood of each of the four urban areas especially for the night-time temperature ; ii) there was generally no significant change over time of rainfall atthe monthly, seasonal and annual time scales ; however, there were few exceptions where stations close to urban areas, had trends of monthly and seasonal rainfall. In particularly, I observed thatJune-July-August (JJA) seasonal rainfall was decreasing over the coastal region and increasing over Nairobi.On further analysis of inter relationships between rainfall and temperature I established fairly strong statistical relationship been rainfall and temperature.For instance, I found that JJA seasonal rainfall in Nairobi at the Dagoretti corner station was strongly associated with JJA temperature of the stations to the northeast of Mombasa town (R2 0.7, p-value 0.001). Such relationships were attributed to the changes in local and regional thermal circulations resulting from enhanced temperatures ;iii) Urban heat islands (UHIs) exist in Nairobi and Mombasa which were observable more clearly from the land surface temperature (LST). The UHIs are strongest during the dry season (DJF). For instance,Nairobi urban area hasa strong day-time and weak night-time UHI,particularly within the CBD and heavily built up areas, while Mombasa has a weaker UHI than Nairobi during both daytime and nighttime. However, the UHIsin both cities are more distinguishable in the night-time than in the day-time LST. I found no evidence in general, thaturbanization and/or UHI have directly influenced seasonal rainfall within Mombasabut there was evidence of changes downwind of Nairobi and ;iv) lastly, I carried out the extreme value analysis of rainfall under stationary and non-stationary conditions using the generalized extreme value (GEV) method and established that stationary GEV modelsof the Gumbel type were applicable to produce design storms for each town


Version intégrale (27,0 Mb)

Page publiée le 25 décembre 2018, mise à jour le 29 décembre 2018