University of Twente Student Theses


Triple Sensor Approach for Monitoring Water and Climate

Ishimwe, Jean de la Croix (2020) Triple Sensor Approach for Monitoring Water and Climate.

[img] PDF
Abstract:Despite the proliferation of sophisticated climate variables measurement techniques, the resulting datasets remain subject to errors due to the complex nature of the variables being estimated. This raises a formidable challenge in water management domains where proper decisions depend heavily on the reliability of the datasets. In this study, the triple sensor approach for the monitoring of water and climate was investigated by testing climatic variables from mutually independent sources on the extended triple collocation method to establish the conditions for its effective applicability for the identification of reliable climatic datasets. The factors affecting rainfall measurement systems (as the main research variable) were described and the differences leading to reliability concerns about the resulting datasets were appreciated. Further, datasets of rainfall, short-wave incoming solar radiation, and 2m air temperature were statistically engaged in different set-ups using the extended triple collocation covariance analysis and the results were used to judge the most reliable at several locations in and around the Lake Naivasha basin in Kenya. Some of the used datasets were obtained from independent citizen observers in the study area, to emphasize the potential of their involvement in data collection especially in data scarce areas. It was concluded that the triple sensor approach allows a more absolute standpoint to assess the reliability of datasets of many climatic variables at a given location compared to the usual dual dataset comparison used for validation except for situations where the triple collocation initial assumptions are violated. Data ranges, variability and size of the time-series were noted as factors contributing to a potential violation of the assumptions and consequently resulting in biased outcomes which cannot be used decisively about the corresponding datasets reliability. The ability of the triple sensor approach to penalize datasets with outliers, as a rather a frequently observed characteristic in climatic datasets, was demonstrated for rainfall datasets. While the triple sensor approach has potential, it cannot always be used as a substitute for other data filtering techniques such as the identification of missing data values before any statistical analysis for dataset validation. It was also noted that the data requirements of the triple sensor approach also limit its application. Lastly, the triple sensor toolbox implemented in ILWIS, as an important communicative tool to stakeholders in scientific research, was tested and suggestions for the improvement of its usability, presentation, and interpretation of its results were outlined.
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page