This thesis investigates the feasibility of implementing a
real time quality control program into a data stream of
hydrometeorlogical data. The vast array of data used in the
forecasting of river levels and avalanches calls for a point of entry
quality control method that is both efficient from a
communications standpoint and practical given the computer
resources available.
The first step in this process is to find a normalization
scheme to enable the direct comparison of precipitation events
between different stations. The normalization scheme derived
uses the climatic database of historical records. The largest set
of historical records available is in the daily time frame.
However, the quick response needed in this type of forecasting
calls for the testing of data in a hourly format. This calls for the
need to develop some sort of transformation between events of
differing time scales.
Once the normalization scheme is in place four tests are used
to analyze the data. These tests compare the incoming data to
what is expected given the climate, forecasted value, previous
weather, and what is occurring at neighboring stations. The
results from these four tests are composited to make a final
opinion of the validity of the incoming data. The data are then
assigned two descriptive parameters. These parameters quantify
the sophistication of the tests performed on the data, and the
believed accuracy of the data. The two scores are then taken into
account to give a final broad description of the program's "opinion"
as to whether the data should be rejected, questioned, screened, or
verified.
Generally the program performs very well. The accuracy and
precision of the tests are left somewhat vague at this point. The
stress in the development of this test was in the modularity and
portability of the program; the testing scheme is not meant to be
limited to the purpose of flood forecasting or even precipitation
data. The threshold parameters, therefore, need to be set by the
end user. These thresholds will be defined by the type of data as
well as the purpose and accuracy of the data checking needed. / Graduation date: 1997
| Identifer | oai:union.ndltd.org:ORGSU/oai:ir.library.oregonstate.edu:1957/28743 |
| Date | 26 November 1996 |
| Creators | Kotwica, Kyle |
| Contributors | Barnes, Jeffrey R. |
| Source Sets | Oregon State University |
| Language | en_US |
| Detected Language | English |
| Type | Thesis/Dissertation |
Page generated in 0.0014 seconds