Affiliation(s)
1. Department of Agricultural Engineering, Bolgatanga Polytechnic, Bolgatanga 767, Ghana
2. Department of Statistics, Bolgatanga Polytechnic, Bolgatanga 767, Ghana
3. Department of Civil Engineering, Bolgatanga Polytechnic, Bolgatanga 767, Ghana
ABSTRACT
The statistical characteristics of a hydrological data for the purposes
of decision making in water resource planning and management is only
justifiable if the data has the right attributes. This requires that the data
being analysed are consistent, free of trend and being part of a stochastic
process whose random characteristics is described by an appropriate
distribution hypothesis. The data available for statistical analysis had a lot
of missing values which could not be ordinarily filled but required a more
comprehensive approach to fill these missing values. The KSOM (Kohonen Self-organising Map) was
used to fill the missing runoff data from the Jidere-Bonde, Lokoja and Makundi
river sites in the Niger basin. Results from
the studies have shown that KSOM is the best tool for filling hydrological data
with high number of missing values. After the data had been processed, some
statistical applications were used to establish the runoff time-series
characteristics of the three river sites of the Niger River basin. The results
showed good attributes for all three river sites, except that Jidere River’s
data exhibited inconsistency. The presence of trend was also established for
all three river sites; Jidere River was modelled based on 3-parameter
lognormal, the other two river sites were modelled based on normal distribution
probability. The presence of trend and other attributes require that a more
stochastic modelling process be carried out. However, the results established give reference for water resource planning and
management.
KEYWORDS
Kohonen Self-organising Map,
water, resources, planning, management.
Cite this paper
References