Size-based evaluation of TIGGE ensemble systems for precipitation forecasting: an end user perspective

ResearchSpace/Manakin Repository

Show simple item record

dc.contributor.advisor Shamseldin, A en
dc.contributor.advisor Melville, B en
dc.contributor.author Khan, Mudasser en
dc.date.accessioned 2014-12-07T19:29:39Z en
dc.date.issued 2014 en
dc.identifier.citation 2014 en
dc.identifier.uri http://hdl.handle.net/2292/23692 en
dc.description.abstract Precipitation forecasts play a key role in decision making for water resource planning and management. They also influence decisions taken for routine day-to-day operations by the users in various sectors including but not limited to agriculture, transportation, construction, hydropower generation, recreation and so forth. Streamflow forecasting is another major area of application where the quality of precipitation forecasts can greatly affect the overall performance of the system. The errors contained in the precipitation forecasts are introduced to the system at the very beginning. They may also lead to a final result that is far from the actual reality when propagated through different components of a streamflow forecasting system. Improving river flow forecasts for longer lead times by incorporating numerical weather predictions (NWP) into streamflow forecasting systems has attracted the attention of hydrologists in recent years. In order to account for the uncertainties in weather forecasting, meteorologists usually prefer to use an ensemble of NWP forecasts instead of relying on a single result. The process becomes considerably more complex and resource hungry when ensembles of NWP forecasts, known as ensemble prediction systems (EPS) are used to feed the flow forecasting models. In an operational setting, where the use of large weather ensembles may not be feasible due to the computational burden, identification of an objective methodology for optimal selection of smaller subsets becomes crucial. Forecasting of flash flooding demands a quick response and using multiple weather forecasts might not meet the requirements for timely decisions. Furthermore, more might not always result in better; inclusion or exclusion of some forecasts may affect the final forecast product. Hydrologists are therefore constrained to use a limited set of precipitation forecasts. There are very few studies in the literature addressing the issue of how ensemble size may affect the overall quality of the precipitation forecasts. Moreover, most of the previous research in this area is based on verification of ensemble systems against only the intense events. On the other hand, different users of weather forecasts have different needs and all are not always primarily interested in forecasts for the intense events. This study is the first to provide a comprehensive comparison of different ensemble prediction systems for their precipitation forecasts corresponding to users‟ needs in different sectors. The research also presents a unique evaluation of two combination methods for making a multi-model ensemble. This study leads also in comparing three statistical techniques to simplify ensembles. The research aims to provide users with an opportunity to select an ensemble of their choice, from the pool of current operational systems, keeping in view their specific needs and available resources. The target is achieved by presenting a sizebased comparison of multiple ensemble systems in different decision scenarios. Three unimodel ensemble systems operational at the China Meteorological Agency (CMA), UK Met Office (UKMO) and the European Centre for Medium-Range Weather Forecasts (ECMWF) were tested for their precipitation forecasts to study the effect of ensemble size on its performance for a lead time as large as 10 days. Two multimodel ensembles constructed by using two distinct approaches for combining ensembles were also tested in this thesis. The study is based on precipitation forecasts from the above stated five ensemble systems and the precipitation and discharge data observed for the Waikato River in New Zealand. A comprehensive comparison of all the ensembles was made for four different applications of the precipitation forecasts. The deterministic and probabilistic performance of the ensemble forecasts were evaluated separately. Three different forecast attributes, accuracy, reliability and resolution, were evaluated for each ensemble. In attempting to find a suitable strategy for reducing the ensemble size, three statistical techniques were employed to obtain a reduced set of the precipitation ensembles. A river flow forecasting model based on gene expression programming (GEP) was subsequently forced by these reduced ensembles and the resulting ensemble forecasts for the river flow were evaluated against the corresponding observed flow. The results indicate that, in general, the size of an ensemble has small effect on its performance. The Control ensemble, consisting only of the control forecasts (generated using the best available estimate of the current state of the atmosphere) from the participating ensembles, was found to be as good in forecasting occurrence of rainfall as the Grand ensemble which consists of all 90 members of the three unimodel ensemble systems. Similarly, the ensemble forecasts for the most likely precipitation event, probability of exceeding a certain precipitation threshold and the magnitude of precipitation from the smaller ensembles were also comparable with the larger ensemble systems. No significant difference was observed between the flow forecasts driven by the smaller ensembles reduced by applying three stratification techniques and the corresponding full counterparts. In addition to the above findings, this research also presents the framework to evaluate different ensemble systems for their specific use. In this way, this PhD research attempts to develop a deeper understanding of the diverse applications of ensemble precipitation forecasts, as well as adding some case studies of a quantitative nature unlike most of the previous qualitative studies. en
dc.publisher ResearchSpace@Auckland en
dc.relation.ispartof PhD Thesis - University of Auckland en
dc.rights Items in ResearchSpace are protected by copyright, with all rights reserved, unless otherwise indicated. Previously published items are made available in accordance with the copyright policy of the publisher. en
dc.rights.uri https://researchspace.auckland.ac.nz/docs/uoa-docs/rights.htm en
dc.rights.uri http://creativecommons.org/licenses/by-nc-sa/3.0/nz/ en
dc.title Size-based evaluation of TIGGE ensemble systems for precipitation forecasting: an end user perspective en
dc.type Thesis en
thesis.degree.grantor The University of Auckland en
thesis.degree.level Doctoral en
thesis.degree.name PhD en
dc.rights.holder Copyright: The Author en
dc.rights.accessrights http://purl.org/eprint/accessRights/OpenAccess en
pubs.elements-id 468152 en
pubs.record-created-at-source-date 2014-12-08 en


Full text options

This item appears in the following Collection(s)

Show simple item record

http://creativecommons.org/licenses/by-nc-sa/3.0/nz/ Except where otherwise noted, this item's license is described as http://creativecommons.org/licenses/by-nc-sa/3.0/nz/

Share

Search ResearchSpace


Advanced Search

Browse