#Civil Engineering Final-Year Paper #Environmental and Biosystems Engineering Final Project #Biosystems and Agricultural Engineering Final Project #Meteorology Final Project #Hydrological Modeling Stage-discharge relationships in open channels: Practices and Problems. Impacts of land use and land cover change on surface runoff, discharge, and low flows: Evidence from East Africa. Streamflow Evaluations for Watershed Restoration Planning and Design. https://streamflow.engr.oregonstate.edu/ #Engineering Hydrology Final Project Hydrology: Principles, Analyses & Design. Methods of Streamflow Data Analysis.

1. UNIVERSITY OF NAIROBI
DEPARTMENT OF CIVIL AND CONSTRUCTION
ENGINEERING
HYDROLOGICAL STUDY OF THE NYANDO BASIN
By SAKWA IGNATIUS SHIUNDU
F16/1585/2015
A Report submitted as partial fulfilment for the award of the degree of Bachelor of Science
in Civil Engineering in the Department of Civil and Construction Engineering in the
University of Nairobi
2019

2. ii
DECLARATION
I, Sakwa Ignatius Shiundu, declare that this report is my original work and has not been
presented for a degree in any other university.
Signature …………………………… Date ……………………………
APPROVAL
I, Dr. S. O. Dulo, declare that this report has been submitted for examination with my approval
as university supervisor.
Signature …………………………… Date ……………………………

3. iii
DEDICATION
I sincerely dedicate this report to my family, particularly my doting parents – Mr. Vincent N.
Sakwa and Mrs. Caroline W. Sakwa – who continue to till and toil for the family, and to the
Almighty God for the preservation He has bestowed upon us.

4. iv
ACKNOWLEDGEMENTS
Foremost, I am eternally grateful to the Almighty God for, among many mercies and graces,
the mind and fortitude to successfully complete this research project. In the same light, I am
sincerely thankful for the unwavering love and support accorded (to) me by my parents, Mr.
and Mrs. Sakwa; I am all that I am because of the great sacrifices they have made on my
account. I once more humbly acknowledge the resources afforded (to) me by my parents in the
form of sound advice, monetary upkeep, and the meeting of my basic and secondary needs.
Secondly, I am greatly indebted to my supervisor, Dr. Simeon O. Dulo, for the ardent oversight
and valuable insight he rendered and the selfless patience he had from inception to completion
of this study. Dr. Dulo, incumbent Chairman of the Department at the time of writing this
report, generously gave us his time despite his very busy schedule.
I would also like to appreciate Mr. Harrison Lee Baraza of WRA Headquarters, Nairobi for the
cordial welcome he gave me at his office, and for the useful stage-streamflow data he furnished
me with. I would also like to laud the trio of ladies – Madams Pauline, Pascaline and Stella –
of KMD Business Support Services, for making the official requests for rainfall data on my
behalf and promptly sending (to) me the same.
Finally, I am grateful to friends of good will and family for journeying with me during this
milestone. The success of this manuscript is credited in part to the prayers they said for me.

5. v
ABSTRACT
The hydrological response of a catchment, which is controlled by climate, vegetation, drainage,
soils and land tenure, is a crucial factor in operations appertaining flood mitigation and design
of hydraulic structures for water supply. Detailed hydrological assessments are conducted as
part of flood risk and drainage assessments; these use topographical information, site
investigations, hydrological data, flow surveys, rainfall data, geological information and other
historic data to determine the extent of catchment areas contributing runoff, and the flows in
watercourses and drainage systems. This study project sets out to investigate the hydrological
patterns of the zone of the Nyando River Basin traversed by the Pararget tributary.
Conventional methods as well as pertinent software were used to analyze the watershed’s
hydrological data. The study’s findings were corroborated by existing research. Potential
weaknesses in the study were identified and recommendations were proposed for further study.

6. vi
TABLE OF CONTENTS
DECLARATION-----------------------------------------------------------------------------------------------------II
APPROVAL ----------------------------------------------------------------------------------------------------------II
DEDICATION-------------------------------------------------------------------------------------------------------III
ACKNOWLEDGEMENTS ---------------------------------------------------------------------------------------IV
ABSTRACT-----------------------------------------------------------------------------------------------------------V
LIST OF PLATES ----------------------------------------------------------------------------------------------- VIII
TABLE OF FIGURES------------------------------------------------------------------------------------------- VIII
TABLE OF CHARTS ------------------------------------------------------------------------------------------- VIII
LIST OF TABLES --------------------------------------------------------------------------------------------------IX
APPENDICES -------------------------------------------------------------------------------------------------------IX
ACRONYMS ---------------------------------------------------------------------------------------------------------X
DEFINITIONS-------------------------------------------------------------------------------------------------------XI
CHAPTER 1: INTRODUCTION
1.1 Background -------------------------------------------------------------------------------------------- 1 -
1.2 Problem Statement------------------------------------------------------------------------------------ 2 -
1.3 Object Statement-------------------------------------------------------------------------------------- 2 -
1.4 Scope of Study ----------------------------------------------------------------------------------------- 2 -
1.5 Justification of the Study ---------------------------------------------------------------------------- 3 -
CHAPTER 2: LITERATURE REVIEW
2.1 Description of the Nyando River Basin----------------------------------------------------------- 4 -
2.1-1 Location ------------------------------------------------------------------------------------------- 4 -
2.1-2 Physiographic Features and Geology ---------------------------------------------------------- 5 -
2.1-3 Land Use and Land Cover----------------------------------------------------------------------- 6 -
2.2 Climate of the Nyando River Basin---------------------------------------------------------------- 7 -
2.3 Hydrology of the Nyando River Basin ------------------------------------------------------------ 8 -
CHAPTER 3: THEORETICAL FRAMEWORK
3.1 Hydrological Parameters, Measurement and Instrumentation--------------------------- - 10 -
3.1-1 Catchment Characteristics-------------------------------------------------------------------- - 10 -
3.1-2 Components of Streamflow ------------------------------------------------------------------ - 11 -
3.1-3 Measurement of Precipitation---------------------------------------------------------------- - 12 -

7. vii
Rain Gauge Network--------------------------------------------------------------------------- - 13 -
3.1-4 Measurement of Streamflow----------------------------------------------------------------- - 13 -
Measurement of Stage ------------------------------------------------------------------------- - 14 -
Staff Gauge -------------------------------------------------------------------------------------- - 14 -
3.1-5 Stage-Discharge Relationship --------------------------------------------------------------- - 14 -
Rating Curves from Steady Uniform Flow------------------------------------------------ - 15 -
Permanent Control----------------------------------------------------------------------------- - 16 -
Extrapolation of Rating Curve -------------------------------------------------------------- - 17 -
CHAPTER 4: METHOD
4.1 Study Region (scope) ------------------------------------------------------------------------------ - 18 -
4.2 Research Method----------------------------------------------------------------------------------- - 18 -
4.3 Hydrometeorological Data ----------------------------------------------------------------------- - 18 -
4.3-1 Challenges Encountered in Data Collection, and Mitigations Applied ---------------- - 20 -
4.4 Preparation of Data-------------------------------------------------------------------------------- - 20 -
4.4-1 Estimation of Missing Data ------------------------------------------------------------------ - 20 -
4.4-2 Test for Consistency of Record-------------------------------------------------------------- - 21 -
4.5 Presentation of Data ------------------------------------------------------------------------------- - 22 -
4.5-1 Rainfall Data ----------------------------------------------------------------------------------- - 22 -
Point Rainfall------------------------------------------------------------------------------------ - 22 -
Mean Areal Precipitation --------------------------------------------------------------------- - 23 -
Arithmetic-Mean Method--------------------------------------------------------------------- - 23 -
4.5-2 Stage Data -------------------------------------------------------------------------------------- - 23 -
Logarithmic-Plot Method (Excel Solver integrated) ------------------------------------ - 24 -
CHAPTER 5: DISCUSSION OF RESULTS
5.1 Delineation of Study Area ------------------------------------------------------------------------ - 25 -
5.2 Rainfall ----------------------------------------------------------------------------------------------- - 25 -
5.1-1 Consistency Test Results --------------------------------------------------------------------- - 25 -
5.1-2 Monthly and Yearly Rainfall ---------------------------------------------------------------- - 26 -
5.3 Streamflow ------------------------------------------------------------------------------------------ - 30 -
5.2-1 Rating Curve Parameters --------------------------------------------------------------------- - 30 -
5.2-2 Monthly Streamflow Analyses -------------------------------------------------------------- - 32 -
Mean Monthly Flow for Period of Record ------------------------------------------------ - 32 -
5.2-3 Mean Annual Flow for Period of Record -------------------------------------------------- - 36 -
5.2-4 Pattern Analysis ------------------------------------------------------------------------------- - 37 -

8. viii
Variation of Annual Flow around Longer-term Mean Flow for Period of Record - 37 -
CHAPTER 6: CONCLUSIONS AND RECOMMENDATIONS
6.1 Conclusions------------------------------------------------------------------------------------------ - 43 -
6.2 Recommendations---------------------------------------------------------------------------------- - 43 -
6.3 Areas for Further Study -------------------------------------------------------------------------- - 43 -
LIST OF PLATES
Plate 2-1: Nyando Catchment Satellite Image with Elevation Profile (1985) ---------------------- 4 -
Plate 2-2: Nyando Catchment Satellite Image with Elevation Profile (2014)---------------------- 5 -
Plate 5-1: Delimitation of The Study Area and Watersheds of the Basin------------------------ - 25 -
TABLE OF FIGURES
Figure 1-1: Sub-catchments of Nyando River Basin -------------------------------------------------- 3 -
Figure 2-1: LULC map of Nyando River Basin (2003)----------------------------------------------- 6 -
Figure 2-2: Meteorological Network in the Nyando River Basin------------------------------------ 8 -
Figure 2-3: Hydrometric Network in the Nyando River Basin--------------------------------------- 9 -
Figure 3-1: Drainage Basin Characteristics---------------------------------------------------------- - 10 -
Figure 3-2: Disposal of Rain Water------------------------------------------------------------------- - 11 -
Figure 3-3: Bank Storage------------------------------------------------------------------------------- - 12 -
Figure 3-4: Staff Gauge -------------------------------------------------------------------------------- - 14 -
Figure 3-5: Gauge Posts on River Bank-------------------------------------------------------------- - 15 -
Figure 3-6: Stage-Discharge Curve - Logarithmic Plot -------------------------------------------- - 16 -
Figure 4-1: Double-Mass Curve----------------------------------------------------------------------- - 22 -
TABLE OF CHARTS
Chart 5-1: Double-Mass Curve – Ahero------------------------------------------------------------- - 26 -
Chart 5-2: Long-term Rainfall Distribution at Key Watershed Stations------------------------- - 27 -
Chart 5-3: Long-term Watershed Mean Monthly Rainfall ---------------------------------------- - 28 -
Chart 5-4: Variation of Annual Totals around Mean Annual Rainfall--------------------------- - 29 -
Chart 5-5: Upstream Discharge Rating Curve ------------------------------------------------------ - 30 -
Chart 5-6: Upstream Stage-Discharge Curve: Logarithmic Plot --------------------------------- - 31 -
Chart 5-7: Downstream Discharge Rating Curve--------------------------------------------------- - 31 -
Chart 5-8: Downstream Stage-Discharge Curve: Logarithmic Plot------------------------------ - 32 -
Chart 5-9: Mean Monthly Flows---------------------------------------------------------------------- - 34 -
Chart 5-10: 10-year Upstream Monthly Flow Regime--------------------------------------------- - 36 -
Chart 5-11: 10-year Downstream Monthly Flow Regime ----------------------------------------- - 36 -

9. ix
Chart 5-12: Mean Annual Flows---------------------------------------------------------------------- - 37 -
Chart 5-13: Upstream Variation of Annual Flow around Mean Flow --------------------------- - 37 -
Chart 5-14: Downstream Variation of Annual Flow around Mean Flow------------------------ - 38 -
Chart 5-15: Flow Duration Curve (Arithmetic Plot) ----------------------------------------------- - 38 -
Chart 5-16: Flow Duration Curve (Log-normal Plot) ---------------------------------------------- - 39 -
Chart 5-17: Mass Curve of Streamflow-------------------------------------------------------------- - 40 -
LIST OF TABLES
Table 4-1: Rainfall Gauge Stations ------------------------------------------------------------------- - 19 -
Table 4-2: Stream Gauging Stations ------------------------------------------------------------------ - 19 -
Table 5-1: Double-Mass Curve Analysis - Ahero Irrigation Research Station------------------ - 26 -
Table 5-2: Long-term Monthly and Annual Rainfall Distribution at Key Stations ------------- - 27 -
Table 5-3: 7-Station Mean Monthly and Mean Annual Rainfall---------------------------------- - 28 -
Table 5-4: Rating Curve Calibration Constants ----------------------------------------------------- - 30 -
Table 5-5: Upstream Monthly and Yearly Streamflow Averages--------------------------------- - 33 -
Table 5-6: Downstream Monthly and Yearly Streamflow Averages ----------------------------- - 33 -
Table 5-7: Upstream Flow Extrema------------------------------------------------------------------- - 41 -
Table 5-8: Downstream Flow Extrema --------------------------------------------------------------- - 41 -
Table 5-9: Summary Streamflow Statistics ---------------------------------------------------------- - 42 -
APPENDICES
Appendix A: Consistency Test Results--------------------------------------------------------------- - 46 -
Appendix B: Rainfall Data Analyses ----------------------------------------------------------------- - 47 -

10. x
ACRONYMS
FAO Food and Agricultural Organisation
GWT Ground Water Table
ITCZ Inter-Tropical Convergence Zone
JICA Japan International Cooperation Agency
KMD Kenya Meteorological Department
LULC Land Use and Land Cover
LULCC Land Use and Land Cover Changes
MSL Mean Sea Level
RMSE Root Mean Square Error
WMO Water Resources Authority
WRA World Meteorological Organisation
WY Water Year

11. xi
DEFINITIONS
Discharge (Q): The volume of water per unit time that passes a specified point on a stream.
Discharge is conventionally measured in cubic feet per second (ft3
/sec or cfs) or cubic meters
per second (cumecs).
Q = w d v,
where w = water width d = mean water depth v = mean water velocity
Rating curve (or stage-discharge curve): A curve relating stage (water height) to water
discharge at a point on the stream channel.
Annual flood: The annual flood on a stream is the highest instantaneous peak discharge of the
water year.
Water year: A 12-month time period for which precipitation totals are measured. Also called
hydrological year.
Flood magnitude: The size of a flood peak in discharge units.
Mean daily discharge: The average discharge of any specified calendar day (midnight to
midnight). It is calculated by taking the total volume of water discharged during that day and
dividing by 86400, the number of second in a day.
Mean annual discharge (Qav): The discharge that would have to flow constantly to equal the
volume of water discharged by that stream over the entire period of years record. Qav is the
total volume of water discharged by the stream during the period of record divided by the
number of seconds in that period. It is more easily calculated by averaging all the individual
mean daily discharges.
Base Flow: The flow in a channel due to soil moisture or ground water.
Drainage Basin: The land zone that contributes water to the runoff past a given point on a
stream.
Sub-catchment: Watershed (see Figure 1-1 and Plate 5-1).
Watershed: The area of land that drains to a single outlet and is separated from other
watersheds by a topographic or subsurface drainage divide.
Watershed Divide: A line or border that defines a watershed topographically.

12. xii
Annual Total Flow: The sum of the daily mean values of discharge for the year.
Annual Mean Flow: The arithmetic mean for the individual daily mean discharges for the year
noted or for the designated period.
Highest Annual Mean Flow: The maximum annual mean discharge occurring for the
designated period.
Lowest Annual Mean Flow: The minimum annual mean discharge occurring for the
designated period.
Highest Daily Mean Flow: The maximum daily mean discharge for the year or for the
designated period.
Lowest Daily Mean Flow: The minimum daily mean discharge for the year or for the
designated period.
Maximum Peak Flow: The maximum instantaneous peak discharge occurring for the water
year or designated period.
Maximum Peak Stage: The maximum instantaneous peak stage occurring for the water year
or designated period.
Instantaneous Low Flow: The minimum instantaneous discharge occurring for the year or for
the designated period.
Annual Runoff: Indicates the total quantity of water in runoff for a drainage area for the year.

13. - 1 -
Chapter 1: INTRODUCTION
Hydrology is the study of the amount and quality of water conveyed on the land surface, and
in soils and rocks near the surface. The hydrological response of a catchment is controlled by
a combination of climate, vegetation, drainage, soils and land use. Detailed hydrological
assessments are conducted as part of flood risk and drainage assessments; these use a
combination of topographical information, site investigations, hydrological data, flow surveys,
rainfall data, geological information and other historic data to determine the extent of
catchment areas contributing runoff, and the flows in watercourses and drainage systems. Key
hydrological studies include catchment analysis, catchment flows, rainfall analysis,
hydrogeology, overland flow/runoff assessment, watercourse hydraulics, flood risk
assessment, and drainage impact assessment (Hydrological Studies-Waterman Group,
https://www.watermangroup.com/, accessed on 28th
May, 2019). As specified in the objectives,
this report will focus on catchment flows and rainfall analysis.
1.1 Background
The sustainable management of the earth’s surface including Land Use and Land Cover
Changes (LULCC) remains a critical environmental challenge that society must address
(Mustard et al., 2004). Besides ecosystem vulnerability, LULCC are major determinants of
global environmental change with potential severe impacts on human livelihoods (Olson et al.,
2008). Such changes manifest in climatological, hydrological and biodiversity responses.
Vitousek et al. (1997) estimated that between 39 and 50% of terrestrial ecosystems have
undergone modification due to anthropogenic influence. The main drivers of LULCC are socio-
economic development, population expansion, and pressures for land for agriculture (Lambin
et al., 2003).
It is commonly argued that forests act both as ‘pumps’ through enhanced evapo-
transpiration (ET) rates and as ‘sponges’ through increased infiltration rates and soil moisture
retention (Bruijnzeel, 2004; Arancibia, 2013). Forested water sheds therefore exhibit smaller
streamflow rates than water sheds dominated by other managed land uses. Forest cover loss
results in changes in albedo, reduction in aerodynamic roughness, reduction of leaf area, and
reduction in rooting depth, consequently causing a reduction in ET which subsequently affects
streamflow (Costa et al., 2003; Farley et al., 2005). The net effect of forest cover loss is
increased water yield (Bosch and Hewlett, 1982). Additionally, a reduction in dry season flow
is often cited as a consequence of deforestation (Bruijnzeel, 1988; Arancibia, 2013; Ogden et
al., 2013; Liu et al., 2015).

14. - 2 -
Even though only approximately 20% of the Lake Victoria water inflows are from
rivers draining into the lake (Awange et al., 2008), the ecosystem health of the lake depends
on the health and flow dynamics of these rivers (Guzha et al., 2018).
1.2 Problem Statement
Owing to adversarial LULCC in the upper basin, water quality has been deteriorating and
quantity fluctuating in the river, with increased peak flows during the rains resulting in
widespread flooding in the lowlands. During the dry season, base flows in the rivers and
streams are considerably reduced, resulting in the drying up of large section of the wetlands
(Raburu et al., 2012). This paper aims at providing information on water resource availability
and hydrological response of the Kano Plains sub-catchment in the Nyando Catchment.
1.3 Object Statement
The main objective of this study was to determine the hydrological characteristics of the
Nyando Catchment.
The specific objectives of this research were:
1. To review previous reports and therein derive the methods suited to this particular
hydrological study.
2. To study the rainfall characteristics of the catchment and find the average annual
rainfall.
3. To study the streamflow of the Nyando Basin, and the high flows and low flows of the
river discharge.
1.4 Scope of Study
The study focused on the hydrological characteristics of the area enclosing the major sub-
catchments (see Figure 1-1) traversed by the Nyando Tributary, particularly its lower reaches
(see Plate 5-1), over ten consecutive water years; rainfall parameters between 1981 and 1990,
and streamflow parameters between 2005 and 2014.

15. - 3 -
Figure 1-1: Sub-catchments of Nyando River Basin
Source: JICA
1.5 Justification of the Study
Forest cover alone is not an accurate predictor of hydrological fluxes in East African
catchments. Variability in results from previous studies supports the need for long-term field
monitoring to better understand catchment responses and to improve the calibration of
currently used simulation models (Guzha et al., 2018). To this end, therefore, this study
couldn’t be timelier. Further, it aims at assessing the efficacy of existing interventions against
land degradation so as to promote welfare of the populace.
The scope chosen (see Method) was ideal owing to the fact that most of the hydrometric
stations within this zone were functional (see Figure 2-3), as opposed to those in individual
sub-catchments (see also Plate 5-1), and because this was downstream the Nyando River.
A ten-year duration of record was chosen so as to capture potential incidences and
effects of variations in the normal hydrological pattern of the catchment caused by climate
change. Specific and separate decades of records (1981-1990 and 2005-2014) were selected
because these were the periods with most complete and continuous records available.
Chronological satellite images (Plates 2-1and 2-2) were used as a sort of ‘time lapse’
to depict LULCC between 1985 and 2014.

16. - 4 -
Chapter 2: LITERATURE REVIEW
2.1 Description of the Nyando River Basin
2.1-1 Location
The Nyando catchment is situated in the eastern sub-catchments of the Lake Victoria Basin of
the Lake Victoria South Catchment Area in Kenya. It covers an area of 3,600 km2
and is
situated within the Winam Gulf between longitudes 34º47” E and 35º44” E, and latitudes 0º07”
N and 0º20” S. Administratively, the catchment is located in Kenya’s Rift Valley and Nyanza
regions.
The Nyando Wetland is incised within the Kano Plains, which is a floodplain riparian
zone transitional between the surrounding upland areas on one end and the Lake Victoria at the
other extreme end downstream (Raburu et al., 2012).
Plate 2-1: Nyando Catchment Satellite Image with Elevation Profile (1985)

17. - 5 -
2.1-2 Physiographic Features and Geology
The Nyando River system is divided into seven distinct physiographic units: (1) the lowlands
in the west dominated by Kano Plains; (2) the scarp-foot zones of the Nandi and Nyabondo
Escarpments; (3) the broken hill and scarp-foot zones east of Chemelil; (4) the lava plateaus
east of the Kano Plains; (5) the Kendu Escarpment and Nyabondo plateau in the south; (6) the
Nandi Escarpment in the north, and (7) the volcanic hills and plateau comprising the Mau
Forest Complex (North Tinderet Forest, Tinderet Forest, Masaita Forest, South-West Mau
Forest and Londiani Forest in the headwaters) (Raburu et al., 2012).
The relief of the Nyando River catchment ranges from 1134m at Lake Victoria to
approximately 3000m in the Mau ranges. The general slope runs from north-east to south-
west, with lowlands/plateau having a slope range of 0-4º and the escarpments a slope of 19-
43º. The relief of the Kano Plains is subdued, ranging from 1135m to 1170m and the general
slope of the land runs from east-north-east to west-south-west. Over the plains the micro-relief
consists of broad swellings and troughs with the meandering channels of the Nyando River
crossing the lower areas. The highlands are characterized by a fine dendritic drainage pattern
Plate 2-2: Nyando Catchment Satellite Image with Elevation Profile (2014)

18. - 6 -
formed by the two main tributaries of the Nyando River, Pararget (Nyando) and Ainamutua
(Raburu et al., 2012).
The geology of the catchment identical to that of the main Lake Victoria basin, of which
geological formations vary from recent quaternary sediments to old rocks of the Archean age.
The soils in the Nyando basin are predominantly clays but vary greatly in texture, composition
and structure. Soils derived from the quaternary volcanic rocks, which are generally fertile, are
found in the higher rainfall areas on the eastern side of the catchment. Those soils derived from
very ancient granite are reasonably fertile and tend to be in the areas of low rainfall within the
catchment (Opere, 1998).
2.1-3 Land Use and Land Cover
The main land use in the catchment include indigenous and plantation forests, agriculture and
shrub land. On the upper reaches of the catchment, agricultural activities include subsistence
farming of food crops (e.g. maize, millet, sorghum) and cash crops (e.g. tea, coffee and
sugarcane) and dairy farming. This area has suffered extensive deforestation in the past to make
room for human settlement and farming, disregarding best practice in land use, putting severe
environmental strain on the lower reaches (Raburu et al., 2012).
Figure 2-1: LULC map of Nyando River Basin (2003)
Source-World Agroforestry Centre GIS laboratory

19. - 7 -
On the lowlands within the Kano Plains, large-scale sugarcane plantations, rain-fed
food crops and rice production, as well as cattle grazing are common. However, about 75% of
the plains are unsuitable for economically viable small-scale farming (Jaetzold and Schmidt,
1982). The extremely heavy soils combined with a warm climate, relatively low rainfall and
repeated flooding make farming economically unattractive (Raburu et al., 2012). Urban centres
and industries in the catchment include Nandi Hills and Kericho on the upper reaches, Chemelil
and Muhoroni on the middle-reaches and Ahero near the river mouth (Raburu et al., 2012).
2.2 Climate of the Nyando River Basin
Rainfall is normally used as the descriptor of climate within the tropics. The variation of rainfall
from January through December is usually considered a sufficient descriptor of tropical
climate. Other meteorological variables do not exhibit significant variation throughout the year.
The Nyando River Basin is in the equatorial zone of low pressure where winds are generally
light and variable. Two monsoons, Northeast and Southeast, which have contrasting
thermodynamic characteristics, generally prevail over the basin in the course of the year.
Spasmodic outbreaks of westerlies often intrude into the established wind systems to cause
marked changes on the normal rainfall pattern; these westerlies are known to cause above-
normal rainfall whose source is the moisture inflow from the tropical rainforests of the Congo
(Rwigi et al., 2012).
The basin experiences a trimodal rainfall pattern with peaks in April, August and
November with magnitudes declining towards November (Rwigi et al., 2012). The mean
annual rainfall ranges from about 1,100 to 1,600mm with a minimum and maximum mean
monthly rainfall of 72mm and 243mm respectively (JICA, 1992). The amount of rainfall is
greatly influenced by altitude and relief features; the upper reaches of the Nyando catchment
experience higher amounts of rainfall compared to the middle and lower basin. Rainfall
averages 1800mm/y in the highlands and is associated with the south-easterly winds carrying
warm air masses from the Indian Ocean causing orographic rainfall on the highlands. The Kano
Plains experience a sub-humid to semi-arid climate and receives rainfall in the range of 600-
1100mm/y (FAO, 1996). The presence of convective rainfall in the Lake Victoria region is
responsible for most of the rain at the shorelines (Jaetzold and Schmidt, 1982).

20. - 8 -
Figure 2-2: Meteorological Network in the Nyando River Basin
(Raburu et al., 2012)
The relative humidity in the middle and lower basin varies between 55% and 75% in
the dry and rainy seasons, respectively, peaking in May and July with the minimum occurring
in January during the short dry season and October, during the long dry season. The mean
minimum annual temperature peaks are recorded in August through September and ranges
from 14℃ to 18℃. Highest temperatures are recorded in June through July with annual mean
maximum ranging from 27℃ to 32℃. The monthly A-pan evaporation far exceeds monthly
rainfall in the basin throughout the year (JICA, 1992). Annual mean A-pan evaporation ranges
from 1900 to 2200mm while the monthly mean evaporation ranges from 1300 to 2200mm. The
monthly minimum and maximum evaporations are recorded during June/July and March,
respectively. Water stress occurs especially after the second rains which are feeble and
unreliable (Jaetzold and Schmidt, 1982).
2.3 Hydrology of the Nyando River Basin
The hydrology of the catchment is strongly influenced by the north-south movement of the
Inter-Tropical Convergence Zone (ITCZ) and local winds (lake/land breezes), which influence
the spatial and temporal variations of hydro-meteorological parameters (Millman, 1973).
Climate and hydrological data are collected from hydrometric (Figure 2-3) and meteorological
(Figure 2-3) networks, by WRMA, KMD and private institutions (Khisa et al., 2012).
The Nyando River has its headwaters in the Mau Forest complex situated on the eastern
shoulder of the Kenyan Rift Valley and pours its waters into Winam Gulf of Lake Victoria after

21. - 9 -
traversing the Kano Plains. Run-off accumulates in the upper Nyando River and peak discharge
occurs in April or early May. In the last 50 years, annual discharge has averaged 22.22m3
/s
(Nicholson and Yin, 2001). The recorded highest peak was experienced during the disastrous
floods caused by abnormally prolonged ‘Uhuru rains’ in the 1961-62 periods when the entire
Kano Plains were flooded (Millman 1973).
Figure 2-3: Hydrometric Network in the Nyando River Basin
(Raburu et al., 2012)
The arrival of seasonal floods from the upper catchment through the main tributaries of
Ainamutua and Nyando causes a stage rise of up to 8m at Ogilo Bridge in the northern part of
Kano Plains. At this stage the river channel is able to confine water levels of up to 10m high
and therefore the flood wave only inundates the flanking seasonal plains downstream. At the
southern end of the Kano Plains, at Ahero (20km from the river mouth), the lateral confinement
of the flood in the channel ceases and floodwater overtops the river bank (Ongwenyi et al.,
1993). Dykes running 8km downstream from Ahero town were constructed in 1975 to contain
the floods. Consequently, the river commonly overtops at the Gem Rae spreading out in the
Nyando Delta wetland, of which the main sources of water are direct precipitation, runoff from
upland areas, inflow from rivers, recharge from aquifers and backflow from the lake during
flooding; the wetland is incised with a floodplain riparian zone which is the transition between
the surrounding upland areas on one end and the Lake Victoria at the other extreme (Raburu et
al., 2012).

22. - 10 -
In an average year, rain causes localized surface flooding during the rainy season, but
this is short-lived as it evaporates and infiltrates slowly into the waterlogged ground. During
periods of exceptional rain, surface flooding is widespread and may persist until the seasonal
flood arrives. Consequently, Winam Gulf experiences an occurrence of intense sediment
plumes after the flushing of the Nyando Wetland (Raburu et al., 2012).
Chapter 3: THEORETICAL FRAMEWORK
3.1 Hydrological Parameters, Measurement and Instrumentation
3.1-1 Catchment Characteristics
The entire area of a river basin whose runoff (due to a storm) drains into the river in the basin
is considered a hydrologic unit called drainage basin, water shed or catchment area (see Figure
3-1) of the river flowing. The boundary line, along a topographic ridge, separating two adjacent
drainage basins is called a drainage divide. The line of the ground water table from which the
water table slopes downward away from the line on both sides is called the ground water divide.
The single point or location at which all surface drainage from a basin converges or
concentrates as outflow is called the concentration point or measuring point, since the
streamflow is usually measured at this point (Raghunath, 2006).
The time intervening before the rain falling at the most distant point in a drainage area
(i.e. on the fringe of the catchment) reaches the concentration point is called the concentration
time; this is a very significant variable since only such storms of duration greater than the
concentration time are able to produce runoff from the entire catchment and cause high
intensity floods. A fan-shaped catchment produces greater flood intensity since all the
tributaries are nearly of the same length and hence the concentration time is nearly the same
Figure 3-1: Drainage Basin Characteristics
(Raghunath, 2006)

23. - 11 -
and is less, whereas in the fern-shaped catchments, the fringe of the catchment is remote, hence
the time of concentration is longer and the discharge distributed over a long period (Raghunath,
2006).
The features of the drainage net may be physically described by the number of streams,
the length of streams, the stream density and the drainage density. The stream density of a
catchment is the number of streams per square kilometre. The drainage density is the total
length of all stream channels (perennial and intermittent) per unit area of the basin and serves
as an index of the areal channel development of the basin; drainage density varies inversely as
the length of overland flow and indicates the drainage efficiency of the basin. A high value
indicates a well-developed network and torrential runoff causing intense floods, while a low
value indicates moderate runoff and high permeability of the terrain (Raghunath, 2006).
3.1-2 Components of Streamflow
When a storm occurs, a portion of rainfall infiltrates into the ground and some portion may
evaporate. The rest, which flows as a thin sheet over the land surface, is termed as overland
flow. If there is a relatively impermeable stratum in the subsoil, the infiltrating water moves
laterally in the surface soil and joins the streamflow, which is termed as underflow (subsurface
flow) or interflow (see Figure 3-2). If there is no impending layer in the subsoil the infiltrating
water percolates into the ground as deep seepage and builds up the ground water table (GWT)
or phreatic surface. The ground water may also contribute to the streamflow, if the GWT is
Figure 3-2: Disposal of Rain Water
(Raghunath, 2006)

24. - 12 -
higher than the water surface level of the stream, creating a hydraulic gradient towards the
stream. Low soil permeability favours overland flow (Raghunath, 2006).
While all three types of flow contribute to the streamflow, it is the overland flow which
reaches the stream channel first, the interflow being slower reaches after a few hours, and the
ground water being the slowest reaches the stream channel after some days. The term direct
runoff is used to include the overland flow and the interflow (Raghunath, 2006).
Direct surface flow can be analyzed for relatively large drainage areas by the unit
hydrograph method and for smaller areas by overland flow analysis. The direct runoff results
from the occurrence of an immediately preceding storm while the ground water contribution,
which takes days or months to reach the stream, in all probability has no direct relation with
the immediately preceding storm. The ground water flow into the stream would have continued
even if there had been no storm immediately preceding. It is for this reason it is termed as base
flow in hydrograph analysis (Raghunath, 2006).
When the overland flow starts (due to a storm), some water is held in puddles, pits and
small ponds; this water stored is called depression storage. The volume of water in transit in
the overland flow which has not yet reached the stream channel is called surface detention or
detention storage. The portion of runoff in a rising flood in a stream, which is absorbed by the
permeable boundaries of the stream above the normal phreatic surface is called bank storage
(Raghunath, 2006).
Figure 3-3: Bank Storage
(Raghunath, 2006)
3.1-3 Measurement of Precipitation
Precipitation is expressed in terms of the depth to which rain water would stand on an area if
the rain were collected on it. Thus 1 cm of rainfall over a catchment area of 1 km2
represents a
volume of water equal to 104
m3
. The precipitation is collected and measured in a rain gauge.

25. - 13 -
Terms such as pluviometer, udometer, ombrometer and hyetometer are also sometimes used to
designate a rain gauge (Subramanya, 2008).
Rain Gauge Network
Since the catching area of a rain gauge is very small compared to the areal extent of a storm, it
is obvious that to get a representative picture of a storm over a catchment the number of rain
gauges should be as large as possible, i.e. the catchment area per gauge should be small. On
the other hand, economic considerations to a large extent, and such considerations as
topography and accessibility to some extent restrict the number of gauges to be maintained.
Hence one aims at an optimum density of gauges from which reasonably accurate information
about the storms can be obtained. According to WMO recommendations, at least 10% of the
total rain gauges should be of the self-recording type (Subramanya, 2008).
3.1-4 Measurement of Streamflow
Streamflow representing the runoff phase of the hydrologic cycle is the most important basic
data for hydrologic studies. Unlike precipitation, evaporation and evapotranspiration which are
all difficult to measure exactly, and of which the presently adopted methods of measurement
have severely limitations, the measurement of streamflow is amenable to fairly accurate
assessment. Interestingly, streamflow is the only part of the hydrological cycle that be
measured accurately (Subramanya, 2008).
The most satisfactory determination of the runoff from a catchment is by measuring the
discharge of the stream draining it, which is termed as stream gauging. A gauging station is
the place or section on a stream where discharge measurements are made (Raghunath, 2006).
A stream can be defined as a flow channel into which the surface runoff from a specified basin
drains. Generally, there is considerable exchange of water between a stream and underground
water. Streamflow is measured in units of discharge (m3
/s) occurring at a specified time and
constitutes historical data. The measurement of discharge in a stream forms an important
branch of Hydrometry, the science and practice of water measurement (Subramanya, 2008).
Barring a few exceptional cases, continuous measurement of stream discharge is very
difficult. As a rule, direct measurement of discharge is a very time-consuming procedure.
Hence, a two-step procedure is followed; first, the discharge in a given stream is related to the
elevation of the water surface (stage) through a series of careful measurements. In the next step
the stage of the stream is observed routinely in a relatively inexpensive manner and the
discharge is estimated using the previously determined stage-discharge relationship. The
observation of the stage is easy, cheap, and, if desired, continuous readings can also be

26. - 14 -
obtained. This method of discharge determination of streams is adopted universally
(Subramanya, 2008).
Measurement of Stage
The stage of a river is defined as its water surface elevation above a datum. This datum can be
the mean sea level (MSL) or any arbitrary datum connected independently to the MSL. The
gauges used in measuring stage may be manual or automatic. Manual gauges include the Staff
Gauge and the Wire Gauge. The Float-Gauge Recorder and the Bubble Gauge comprise
automatic stage recorders.
Staff Gauge
The simplest of stage measurements are made by noting the elevation of the water surface in
contact with a fixed graduated staff. The staff is made of a durable material with a low
coefficient of expansion with respect to both temperature and moisture. It is fixed rigidly to a
structure, such as an abutment, pier, wall, etc. The staff may be vertical or inclined with clearly
and accurately graduated permanent markings. The markings are distinctive, legible from a
distance, and are similar to those on a surveying staff. Sometimes, it may not be possible to
read the entire range of water surface elevations of a stream by a single gauge and in such cases
the gauge is built in sections at different locations. Such gauges are called sectional gauges
(Figure 3-4). When installing sectional gauges, care must be taken to provide an overlap
between various gauges and to refer all the sections to the same common datum (Subramanya,
2008).
Figure 3-4: Staff Gauge
(Raghunath, 2006)
3.1-5 Stage-Discharge Relationship
As indicated earlier, the measurement of discharge by the direct method involves a two-step
procedure; the development of the stage-discharge relationship which forms the first step is of
utmost importance. Once the stage-discharge (G-Q) relationship is established, the subsequent
procedure consists of measuring the stage (G) and reading the discharge (Q) from the (G-Q)

27. - 15 -
relationship. This second part is a routine operation. Thus the aim of all current-meter and other
direct-discharge measurements is to prepare a stage-discharge relationship for the given
channel gauging section. The stage-discharge relationship is also known as the rating curve
(Subramanya, 2008).
A river is gauged by current meter throughout the rainy season at different stages (water
levels) of the river. The water stage can be read on the enamel painted staff gauges (gauge
posts) erected at different levels at a gauging station (Figure 3-5); it may be noted that
corresponding graduation of gauge posts at two locations are fixed at the same level. From the
plot of stream discharge Q versus gauge height h (rating curve), the stream discharge
corresponding to staff gauge readings taken throughout the year/s can be obtained, provided
the section of the stream at or near the gauging site has not materially altered. Periodical
gauging is conducted to verify the rating curve, or to revise the rating curve if any change in
the section has been noticed (Raghunath, 2006).
Figure 3-5: Gauge Posts on River Bank
(Raghunath, 2006)
The measured value of discharges when plotted against the corresponding stages gives a
relationship that represents the integrated effect of a wide range of channel and flow
parameters. The combined effect of these parameters is termed control. If the (G-Q)
relationship for a gauging section is constant and does not change with time, the control is said
to be permanent. If it varies with time, it is called shifting control (Subramanya, 2008).
Rating Curves from Steady Uniform Flow
The most commonly used stage-discharge ratings treat the discharge as a unique function of
the stage. These ratings typically follow a power curve of the form given by Eq. (3.1) (Herschy,
1995; ISO 1998; Kennedy, 1984; Rantz et al., 1982b):
𝑄 = 𝐶𝑟(𝐺 − 𝑎)𝛽
⋯ (3.1)
where Q is the discharge, G is the stage and a, Cr and β are calibration coefficients.

28. - 16 -
Cr is the discharge when the effective depth of flow (G - a) is equal to 1; a is the gauge
height of zero flow; β is the slope of the rating curve (on logarithmic paper); (G - a) is the
effective depth of water on the control. When the exponent β approaches to 1.5, the rating is
also known as a Guglielmini rating curve (Ufficio Idrografico del Magistrato di Venezia, 1914).
The rating equation is based on the Manning equation, which frequently is used as the
governing equation for steady uniform flow problems (Braca, 2008).
Permanent Control
A majority of streams and rivers, especially nonalluvial rivers exhibit permanent control, in
which case the relationship between stage and discharge is a single-valued relation expressed
as: 𝑄 = 𝐶𝑟(𝐺 − 𝑎)𝛽
, in which Q = stream discharge, G = gauge height (stage), a = a constant
representing the gauge reading corresponding to zero discharge, Cr and β are rating curve
constants. This relationship can be expressed graphically by plotting the observed relative stage
(G - a) against the corresponding discharge values in an arithmetic or logarithmic plot.
Logarithmic plotting (Figure 3-6) is advantageous as Eq. (3.1) plots as a straight line in
logarithmic coordinates. The straight line is drawn to best represent the data plotted as Q vs (G
- a). Coefficients Cr and β need not be the same for the full range of stages. The constant a
representing the stage (gauge height) for zero discharge in the stream is a hypothetical
parameter and cannot be measured in the field (Subramanya, 2008).
The best values of Cr and β in Eq. (3.1) for a given range of stage are obtained by the
least-square-error regression (Subramanya, 2008). However, conventional methods of
determining the three rating curve constants a, Cr and β not as efficient as the Excel Solver,
which has a lower Root Mean Square Error (RMSE) and a higher correlation coefficient than
Figure 3-6: Stage-Discharge Curve - Logarithmic Plot
(Raghunath, 2006)

29. - 17 -
conventional models. Excel solver available in Microsoft Excel is a nonlinear optimization
code, and its specific implementations have been proven in use over many years as one of the
most robust and reliable approaches to solve difficult and highly nonlinear programming
problems (Singh et al., 2018).
Excel solver has the capability to optimize linear as well as nonlinear equations by
changing specified parameters. It consists of linear programming solver (LPS) to optimize
linear equations (simplex LP), generalized reduced gradient (GRG) solver, and evolutionary
solver to optimize nonlinear equations. Rating curve equations are basically of nonlinear form;
therefore, GRG nonlinear solver and evolutionary solver are used to obtain the optimum values
of rating curve parameters (Singh et al., 2018).
Equation (3.1), 𝑄 = 𝐶𝑟(𝐺 − 𝑎)𝛽
, is called the rating equation of the stream and can be
used for estimating the discharge Q of the stream for a given gauge reading G within range of
data used in its derivation (Subramanya, 2008).
Extrapolation of Rating Curve
Most hydrological designs consider extreme flood flows. As an example, in the design of
hydraulic structures, such as barrages, dams and bridges one needs maximum flood discharges
as well as maximum flood levels. While the design flood discharge magnitude can be estimated
from other considerations, the stage-discharge relationship at the project site will have to be
used to predict the stage corresponding to design-flood discharges. Rarely will the available
stage-discharge data include the design-flood range and hence the need for extrapolation of the
rating curve (Subramanya, 2008).
Before attempting extrapolation, it is necessary to examine the site and collect relevant
data on changes in the river cross-section due to flood plains, roughness and backwater effects.
The reliability of the extrapolated value depends on the stability of the gauging section control.
A stable control at all stages leads to reliable results. Extrapolation of the rating curve in an
alluvial river subjected to aggradation and degradation is unreliable and the results should
always be confirmed by alternate methods. There are many techniques of extending the rating
curve and two well-known methods are the Conveyance Method and the Logarithmic Plot
Method (Subramanya, 2008). The latter technique is elaborated in the next chapter of this
report.

30. - 18 -
Chapter 4: METHOD
4.1 Study Region (scope)
The study of the Nyando Catchment was narrowed down to the major sub-catchments found
within and/or overlapping the Kano Plains and traversed by the Pararget Tributary. It was in
this region that the rainfall gauging stations skirting the Kano floodplain and chosen for
procurement of hydrometeorological data were located, being in part the source of the
floodplain’s inlets.
Further, the downstream gauging station was located within this sub-catchment, the
upstream one being in Kipkelion on the Pararget (Nyando) tributary. Kipkelion Railway
Station, the ‘uppermost’ rainfall gauge station used in the study, was included so as to
corroborate findings of upstream stage and discharge if need arose.
The lower reaches of the Nyando River just after the confluence of its two tributaries
(Ainamutua and Pararget) were delineated (see Plate 5-1) using commands and procedures on
Google Earth Pro program, which delineation facilitated illustration of the downstream
elevation profile/cross-section. Prior demarcation was also done to obtain Plates 2-1 and 2-2.
4.2 Research Method
This study depended upon a literature survey of available research published in electronic
databases. The search was conducted using the search engines Google Scholar and Scopus.
The principal method employed for literature search was screening for peer-reviewed
journals and reports based on hydrological studies. Key words and phrases used included
“Nyando Catchment”, “streamflow analyses”, “hydrological study” and “land use and land
cover”.
Further, more data in the form of satellite images of the area of interest and its pertinent
characteristics was abstracted courtesy of such online resources as Google Earth Pro and
Google Maps. Existing maps of the Nyando basin were overlain onto satellite images not only
to allow for cartography/delineation of the boundaries of the sub-catchment, but also to
corroborate existing research.
4.3 Hydrometeorological Data
The hydrological variables considered in this study were: mean monthly rainfall and surface
runoff i.e. mean annual discharge, peak discharge, water yield and low flows. Annual rainfall
data (monthly totals of rainfall) sets were obtained from the Kenya Meteorological Department
(KMD), while streamflow data sets were sourced from the Water Resources Authority (WRA).

31. - 19 -
Seven rainfall gauging stations (Table 4-1) and two streamflow gauging stations – one
upstream, the other downstream – (see Figure 2-3 and Table 4-2) were found appropriate for
the hydrological assessment of the sub-catchment; the criteria used for selection of these was
based on the fact that majority of the stations were closed down for providing erroneous and
inconsistent data, i.e. data lacking accuracy and precision.
Table 4-1: Rainfall Gauge Stations
Station
Code1 Station Name Latitude Longitude
Altitude
(m)
Years of
Records
Missing
Data
(%)
9034081
KIBOS NATIONAL FIBRE
RESEARCH CENTRE
0° 06’ S 34° 81’ E 1173 1980 – 2016 22.07 %
9034086
AHERO IRRIGATION
RESEARCH STATION
0° 13’ S 34° 93’ E 1219 1980 – 2009 24.17 %
9035020
KIPKELION RAILWAY
STATION
0° 02’ S 35° 46’ E 1931 1980 – 2003 15.65 %
9035046 CHEMELIL PLANTATION 0° 06’ S 35° 15’ E 1229 1980 – 2016 27.25 %
9035148 KORU BIBLE SCHOOL 0° 02’ S 35° 26’ E 1707 1980 – 2010 10.75 %
9035199
AINAMOI CHIEF'S CAMP
–KERICHO
0° 03’ S 35° 26’ E 1981 1980 – 1994 22.22 %
9035269
KIPSITET CHIEF'S OFFICE
–KERICHO
0° 21’ S 35° 16’ E 1864 1980 – 1999 12.08 %
Homogeneity tests were carried out on rainfall data using each station as a problem
station while the rest of the rainfall stations served as base stations. This was done with the
assumption that the sub-catchment was meteorologically homogenous.
Table 4-2: Stream Gauging Stations
Station Code Tributary Gauging Station Years of Records
Missing
Data (%)
ICC06 Nyando (upstream) at Kipkelion 1967 – 2018 25.5%
IGDO1 Nyando (downstream) at Ahero Bridge
1948 – 1962
& 2005 – 2018
34%
Care was taken not to exceed the instrument accuracy of the data as supplied by the
sources/institutions. To this end, an accuracy of one decimal place was maintained for rainfall
data, whereas stage data had an accuracy of two decimal places. Flow values were calibrated,
and therefore given an accuracy of three decimal places.
1
The Station Coding, Gi, is used here only for the purpose of convenience.

32. - 20 -
4.3-1 Challenges Encountered in Data Collection, and Mitigations Applied
As aforementioned herein, one of the drawbacks encountered was that some of the stations
were decommissioned for supplying inaccurate and imprecise hydrometeorological data.
Moreover, only limited data was available from some of the operational stations, leading to
further screening of stations.
Availability and adequacy of data were the main limitations, and filling in the missing
data records for all data sets had to be done using suitable methods in order to secure continuity
of data which is a vital requirement in any research. The number of years with data varied for
the two data sets. As such, Ten years of data, (1981-1990 precipitation, and 2006-2015
streamflow), were used as this was found to be period with the highest quality data.
Microsoft Excel’s ‘TREND’ formula, based on non-linear regression, and having
accounted for seasonality, was used to model missing values of stage upstream provided there
were values for the corresponding stage downstream, and vice versa. In the unfortunate event
where both upstream and downstream stage values were missing for the same day, estimates
could not be made and therefore the array was left blank. The stage-discharge relationship was
then used to calibrate missing values of discharge.
4.4 Preparation of Data
Before using the rainfall records of a station, it was necessary to first check the data for
continuity and consistency. The continuity of a record may be broken with missing data due to
many reasons such as damage or fault in a rain gauge during a period. The missing data can be
estimated using the data from the neighbouring station. In these calculations the 2
normal
rainfall may be used as a standard comparison (Subramanya, 2008).
4.4-1 Estimation of Missing Data
Given the annual precipitation values, 𝑃1, 𝑃2, 𝑃3, ⋯ , 𝑃
𝑚 at neighbouring M stations
1, 2, 3, ⋯ , 𝑀 respectively, it is required to find the missing annual precipitation 𝑃
𝑥 at a station
X not included in the above M stations. Further, the normal annual precipitations 𝑁1, 𝑁2, ⋯ , 𝑁𝑖
at each of the above (M + 1) stations including station X are known (Subramanya, 2008).
If the normal annual precipitations at various stations are within about 10% of the
normal annual precipitation at station X, then a simple arithmetic average procedure is followed
to estimate 𝑃
𝑥. Thus
2
The normal rainfall is the average value of rainfall at a particular date, month or year over a specified 30-year
period. The 30-year normal are recomputed every decade. Thus the term normal annual precipitation at station
A means the average annual precipitation at A based on a specified 30 years of record (Subramanya, 2008).

33. - 21 -
𝑃
𝑥 =
1
𝑀
[𝑃1 + 𝑃2 + ⋯ + 𝑃
𝑚] ⋯ (4.1)
If the normal precipitations vary considerably, then 𝑃𝑥 is estimated by weighing the
precipitation at the various stations by the ratios of normal annual precipitations. This method,
known as the normal ratio method, gives 𝑃
𝑥 as
𝑃𝑥 =
𝑁𝑥
𝑀
[
𝑃1
𝑁1
+
𝑃2
𝑁2
+ ⋯ +
𝑃
𝑚
𝑁𝑚
] ⋯ (4.2)
(Subramanya, 2008)
4.4-2 Test for Consistency of Record
Ideally, many statistical analyses require that data being used be homogenous in order that the
research results be considered satisfactory. Homogeneity of data was tested using and mass
curve analyses, one of two most popular methods in hydrology (Ogallo, 1981).
If the conditions relevant to the recording of a rain gauge station have undergone a
significant change during the period of record, inconsistency would arise in the rainfall data of
that station. This inconsistency would be felt from the time the significant change took place.
Some of the common causes for the inconsistency of record are:
a) shifting of a rain gauge station to a new location,
b) change in the ecosystem due to such calamities as forest fires, landslides,
c) the neighbourhood of the station undergoing marked change, and,
d) occurrence of observational error from a certain date. (Subramanya, 2008)
The checking for consistency of a record was done by the double-mass curve technique;
this technique is based on the principle that when each recorded data comes from the same
parent population, they are consistent. A group of 5 – standard is 5 to 10 – base stations in the
neighbourhood of the problem station X were selected. The data of the annual (or monthly or
seasonal mean) rainfall of the station X and the average rainfall of the group of base stations
covering a long period was arranged in the reverse chronological order (i.e. the latest record as
the first entry and the oldest record as the last entry in the list) (Subramanya, 2008).
The cumulative precipitation of the station X (i.e. Σ𝑃
𝑥) and the cumulative values of the
average of the group of base stations (i.e. Σ𝑃
𝑎𝑣) were then calculated starting from the latest
record. Values of Σ𝑃
𝑥 are plotted against Σ𝑃
𝑎𝑣 for various consecutive time periods (Figure 4-
1). A decided break in the slope of the resulting plot indicates a change in the precipitation
regime of station X. The precipitation values at station X beyond the period of change of regime
(point 63 in Figure 4-1) was corrected by using the relation: 𝑃
𝑐𝑥 = 𝑃
𝑥
𝑀𝑐
𝑀𝑎
⋯ (4.3)

34. - 22 -
where 𝑃𝑐𝑥 = corrected precipitation at any time period 𝑡1 at station X
𝑃𝑥 = original recorded precipitation at the time period 𝑡1 at station X
𝑀𝑐 = corrected slope of the double-mass curve
𝑀𝑎 = original slope of the double-mass curve (Subramanya, 2008)
In this way, the older records were brought to the new flow regime of the station. It is apparent
that the more homogenous the base station records are, the more accurate will be the corrected
values at station X. A change in the slope is normally taken as significant only where it persists
for more than five years. The double-mass curve is also helpful in checking systematic
arithmetic errors in transferring rainfall data from one record to another (Subramanya, 2008).
4.5 Presentation of Data
4.5-1 Rainfall Data
Point Rainfall
Point rainfall, also known as station rainfall, refers to the rainfall data of a station. Depending
on the need, data can be listed as daily, weekly, monthly, seasonal or annual values for various
periods. Graphically, these data are represented as plots of magnitude vs chronological time in
the form of a bar diagram. Such a plot, however, is not convenient for discerning a trend in the
rainfall as there will be considerable variations in the rainfall values leading to rapid changes
in the plot. The trend is often discerned by the method of moving averages, also known as
moving means (Subramanya, 2008).
Figure 4-1: Double-Mass Curve
(Subramanya, 2008)

35. - 23 -
Mean Areal Precipitation
As indicated earlier, rain gauges represent only point sampling of the areal distribution of a
storm. In practice, however, hydrological analysis requires a knowledge of the rainfall over an
area, such as over a catchment (Subramanya, 2008).
To convert the point rainfall values at various stations into an average value over a
catchment, the Arithmetic-mean method was adopted.
Arithmetic-Mean Method
Rainfall measurements, obtained from rainfall stations as point measurements, were converted
to areal rainfall using the arithmetic mean method to represent the average annual depth of
rainfall in the catchment, which representation is considered better for catchment rainfall as
opposed to point observations (Rwigi et al., 2012).
When the rainfall measured at various stations in a catchment shows little variation, the
average precipitation over the catchment area is taken as the arithmetic mean of the station
values. Thus if 𝑃1, 𝑃2, ⋯ , 𝑃𝑖, ⋯ 𝑃
𝑛 are the rainfall values in a given period in N stations within
a catchment, then the values of the mean precipitation 𝑃
̅ over the catchment by the arithmetic-
mean method is
𝑃
̅ =
𝑃1 + 𝑃2 + ⋯ + 𝑃𝑖 + ⋯ + 𝑃
𝑛
𝑁
=
1
𝑁
∑ 𝑃𝑖
𝑁
𝑖=1
⋯ (4.4)
(Subramanya, 2008)
The consistency tests conducted indicated that the study region had meteorological
homogeneity, thereby justifying the use of the arithmetic-mean method for zonal rainfall
analysis.
Data was presented in both tabular and graphical formats: plots of long-term mean
monthly rainfall for 10-year period, mean annual rainfall and of consistency tests (double-mass
curves) at all seven rain gauge stations were defined. Tables of monthly totals and yearly
averages were included in the next chapter, whilst other tables were included in the appendices.
4.5-2 Stage Data
The stage data is often presented in the form of a plot of stage against chronological time known
as stage hydrograph. In addition to its use in the determination of stream discharge, stage data
itself is of importance in design of hydraulic structures (e.g. reservoirs), flood warning and
flood protection works. Reliable long-term stage data corresponding to peak floods can be
analyzed statistically to estimate the design peak river stages for use in the design of hydraulic
structures, such as bridges, weirs, etc. Historic flood stages are invaluable in the indirect

36. - 24 -
estimation of corresponding flood discharges. In view of these multifarious uses, the river stage
forms an important hydrologic parameter chosen for regular observation and recording
(Subramanya, 2008).
Logarithmic-Plot Method (Excel Solver integrated)
In this technique the stage-discharge relationship given by Eq. (3.1), i.e., 𝑄 = 𝐶𝑟(𝐺 − 𝑎)𝛽
, in
the previous chapter was made use of; calibration coefficients a, Cr and β of Eq. (3.1) were
obtained by the least-square-error method (least-sum-of-squared-residuals in Excel Solver) by
regressing (𝐺 − 𝑎) on 𝑄. (𝐺 − 𝑎) is considered the predictor/explanatory variable or
independent variable whereas 𝑄 is the response variable/dependent variable of the bivariate
data. The values with the highest correlation coefficient 3
(R2
↦ 1) i.e. constants with the
highest predictive power were chosen as the rating curve constants.
The stage was then plotted against the discharge on a log-log paper. A best-fit linear
relationship was obtained for data points lying in the high-stage range and the line (was)
extended to cover the range of extrapolation. A straight line (R2
= 1) implied the rating curve
constants had absolute predicting power.
By use of these values of the rating equation given by Eq. (3.1), the values of discharge
corresponding to modelled values of stage were estimated so as to fill in the missing data.
Data was presented through tabular and graphical means: tables of monthly and annual
stage and streamflow averages upstream and downstream for all ten years of record, as well as
tables of high and low flows, were included. Other tables were attached in the appendices for
reference, comparison and corroboration.
Graphs included rating curves, logarithmic plots, mean monthly flow plots, mean
annual flow plots, 3- and 5-year moving mean plots, mass curves and flow-duration curves. In
addition, plots of maximum, minimum and one-standard-deviation for long-term mean
monthly flow were made. Consideration was made by featuring both upstream and downstream
plots.
3
R2
is the Determination Coefficient that depicts the goodness of fit of a model.

37. - 25 -
Chapter 5: DISCUSSION OF RESULTS
5.1 Delineation of Study Area
The study area (see Scope) was marked out using Google Earth Pro, a popular geospatial
software application; Figure 1-1 was overlain onto a satellite image (dated November, 2019)
of the Nyando River Basin not only to cross-check and authenticate theory, but also to bring
the schematic into context. The lower reach of the Nyando River just after the confluence of
the Ainamutua and Pararget tributaries was also delineated and its elevation profile obtained.
5.2 Rainfall
5.1-1 Consistency Test Results
Using double-mass curve analysis (see Table 5-1), consistency tests were performed for all
seven raingauge stations. Plots generated from the consistency tests show that the watershed
has meteorological homogeneity, which plots are included in the appendices. A plot of Ahero’s
consistency results (Chart 5-1) is included in this chapter to illustrate homogeneity.
Plate 5-1: Delimitation of The Study Area and Watersheds of the Basin

38. - 26 -
Chart 5-1: Double-Mass Curve – Ahero
Meteorological consistency in the watershed justifies the use of the Arithmetic mean method,
which is used when rainfall values at gauge stations of interest vary by 10% at most.
Table 5-1: Double-Mass Curve Analysis - Ahero Irrigation Research Station
5.1-2 Monthly and Yearly Rainfall
Monthly and yearly rainfall were both computed using the arithmetic mean method, which
method was justified by the meteorological homogeneity of the watershed depicted by the
consistency test results. Table 5-2 overleaf is a summary of the outcome of these computations.

39. - 27 -
Table 5-2: Long-term Monthly and Annual Rainfall Distribution at Key Stations
Chart 5-2 below is derived from Table 5-2. Chart 5-2 (see also Chart 5-3) illustrates that the
sub-basin generally experiences a bimodal rainfall pattern, with highs in the region of 200mm
in April and highs approaching 150mm in August, and lows around 67mm between December
of one year (say, 1980) and January of the next (1981, say).
The greater Nyando River Basin experiences an average annual rainfall of about
1400mm ranging from well below 1100mm/y around the lakeshores to over 1800mm towards
the eastern highlands (Rwigi et al., 2012). Table 5-2 provides a good comparison against
existing research; a mean annual rainfall of 1382.1mm, and boundaries of 1110.2mm/y and
1811mm/y.
It is also evident from Chart 5-2 and Table 5-2 that Ainamoi (in Kericho) receives the
most precipitation in the watershed. The rainfall magnitude is greatly influenced by altitude
and relief features; Ainamoi in Kericho (see Figure 2-1) is situated somewhere between the
middle reaches of the Nyando River Basin and Lake Victoria, at an elevation of about 1880m
Chart 5-2: Long-term Rainfall Distribution at Key Watershed Stations

40. - 28 -
above sea level (see Plate 5-1). It is therefore most likely bound to experience both orographic
and convectional rainfall.
Table 5-3 below shows the mean monthly rainfall and mean annual rainfall for the
watershed under study for all ten years of the period of record, as well as the collective monthly
averages and annual average for the ten-year period.
Table 5-3: 7-Station Mean Monthly and Mean Annual Rainfall
Chart 5-3 is a derivative of the last row of Table 5-3 (see also Table 5-2). Unlike Chart
5-2 which shows individual trends in the bimodal rainfall pattern, Chart 5-3 illustrates the
collective bimodal rainfall trend of the watershed under study.
Rainfall is normally used as the descriptor of climate within the tropics. The variation of rainfall
from January through December is usually considered a sufficient descriptor of climate within
the tropics. Chart 5-3 arguably demonstrates the bimodal rainfall pattern of the sub-catchment;
Chart 5-3: Long-term Watershed Mean Monthly Rainfall

41. - 29 -
two monsoons, North-East and South-East, which have contrasting thermo-dynamic
characteristics, generally prevail over the basin in the course of the year (Rwigi et al., 2012),
and these should be responsible for the bimodal rainfall pattern in the sub-catchment. Now, it
has been argued before that the pattern is in fact trimodal. Nonetheless, from Chart 5-3, the
margins between precipitation in September, October and November are rather fine. This
remains an area for further research since several more studies vouch for a bimodal pattern.
The mean annual rainfall for the 10-year period of record for the sub-catchment was
computed by finding the average of the total annual rainfall for the ten years (see column with
blue data bars in Table 5-3), i.e., by dividing the sum of annual totals by ten. The 7-station total
annual rainfall for respective years was computed by first finding the averages of the respective
monthly totals of all seven stations for the corresponding 10 years, then summing up these
averages (see Table 5-3). For instance, monthly totals of all seven Januarys were tallied and
divided by seven. This computation was repeated for all twelve months, and for every
corresponding year. The monthly totals were supplied by KMD, and were computed by adding
up the mean daily rainfall for respective months. This 7-station breakdown (arithmetic mean
method) was used instead of point rainfall analyses, and was justified by the watershed’s
meteorological homogeneity.
From theoretical support (FAO, 1996 and Rwigi et al., 2012), rainfall averages
1800mm/y in the highlands (see Ainamoi in Table 5-2) and 1200mm/y, give or take, in the
floodplains. From Chart 5-4 below and Table 5-3, the mean annual rainfall for the ten years of
record – 1382.1mm/y – is also well within the range of 1100mm to 1600mm postulated by
JICA (1992).
Chart 5-4: Variation of Annual Totals around Mean Annual Rainfall

42. - 30 -
Spasmodic outbreaks of westerlies often intrude into the established wind systems to
cause marked changes on the normal rainfall pattern; these may be responsible for the variation
demonstrated by both Chart 5-4 and Table 5-3.
The Nyando River Basin has no distinct dry period, and thus it may be regarded as a
moderately humid catchment (Rwigi et al., 2012).
5.3 Streamflow
5.2-1 Rating Curve Parameters
The rating equation 𝑄 = 𝐶𝑟(𝐺 − 𝑎)𝛽
was used in two ways: (i) to calibrate missing flow values
for corresponding values of modelled stage, and (ii) to validate data values in continuous
stretches of record. The rating curve parameters a, Cr and β used upstream were different from
those used downstream:
Table 5-4: Rating Curve Calibration Constants
Parameters Upstream Downstream
a 5.2 x 10-5
≃ 0 0
Cr 8.508 19.96
β 1.410 1.412
4
Control Permanent Shifting
Arithmetic plots (Charts 5-5 and 5-7) were used to obtain the upstream and downstream
rating curves.
4
The combined effect of these parameters is termed control. If the (G-Q) relationship for a gauging section is
constant and does not change with time, the control is said to be permanent. If it varies with time, it is called
shifting control (Subramanya, 2008).
Chart 5-5: Upstream Discharge Rating Curve

43. - 31 -
Chart 5-6: Upstream Stage-Discharge Curve: Logarithmic Plot
Logarithmic plots (Charts 5-6 and 5-8) were preferred to normal plots for the stage-
discharge curves as the former plotted as a straight line in logarithmic coordinates to better
illustrate the predictive power of the rating curve constants.
Chart 5-7: Downstream Discharge Rating Curve
The near-unity values of the determination coefficient (R2
) are the first and best
descriptors of the rating curves’ predictive power. Curve-fitting was done using polynomial
curves that had either the highest value of the determination coefficient (R2
) or the most
reasonable fit, or both.

44. - 32 -
Chart 5-8: Downstream Stage-Discharge Curve: Logarithmic Plot
The upstream discharge rating curve parameters have absolute forecasting power (see
Charts 5-5 and 5-6), and are stronger calibrators than their downstream equivalents (see Charts
5-7 and 5-8). This is due to the fact that the upstream rating curve had permanent control, i.e.,
every year in the 10-year period of record had the same values of rating curve constants.
Shifting control was exhibited by the downstream rating curve in that each year with
continuous streamflow record had different rating curve constants. The shifting control may be
due to scour-and-fill in the sand-bed channel, aquatic vegetation, variable backwater or change
to channel cross-section; dykes running 8km downstream from Ahero town, which were
constructed in 1975 to contain the floods, may be responsible for the change in the channel’s
cross-section. Moreover, backflow from the lake during flooding is a source of water for the
Nyando Delta Wetland, and may be a contributor to the shifting control. The values listed in
Table 5-4 were used as a compromise by using all the data for the 10-year period of record to
find the rating curve constants.
5.2-2 Monthly Streamflow Analyses
Mean Monthly Flow for Period of Record
Mean monthly flow is the average flow for a given month of the year. In addition, a mean
monthly flow may be calculated, by month, for the full period of record. For example, the
average streamflow for January 2005 is based on the daily flows for that month, whereas, the
long-term mean monthly flow for January is based on the average for all Januarys in the period
of record at a particular gauge (Oregon State University, 2005).

45. - 33 -
Table 5-5: Upstream Monthly and Yearly Streamflow Averages
To obtain these values, the mean flow for each month for the entire period of record
was calculated as follows; First, the mean monthly flow for each month for every year in the
period of record was computed. The sum of daily flows for each month, when divided by the
number of days in the month, gives the mean flow for that month. Each year's monthly data for
a particular month was the tallied and this tally divided by the number of years to obtain the
mean monthly flow for that month for the period of record (see Tables 5-5 and 5-6).
Table 5-6: Downstream Monthly and Yearly Streamflow Averages
Tables 5-5 and 5-6 were used to plot Charts 5-9, 5-10, 5-11, 5-12, 5-13 and 5-14.

46. - 34 -
Chart 5-9: Mean Monthly Flows
Plots of monthly discharge provide a visualization of the annual cycle, depicting the
months that contain high flows, low flows or average flows. From Chart 5-9, it is clear that
peak discharge occurs in May and September, whereas base flows occur in February and July.
It is also apparent that the watershed generally has a bimodal flow pattern. The high flows of
May and September correspond to and succeed the heavy rains of April and August (see Chart
5-3), hence the bimodal trend. The build-up of storm runoff that peaks in May and September
begins in the respective immediate precursors, i.e., April and August, during the short and long
rains respectively.
April and August represent the overland flow and interflow components of streamflow,
which components comprise direct runoff; overland flow reaches the stream channel first,
while interflow takes a few hours. The peaks of May and September represent the combined
effect of direct runoff and ground water flow, which takes days or months (days in this
scenario) to reach the stream. The ground water flow into the stream continues even with no
storm immediately preceding, and is represented by the low flows of February and July as well
as the rest of the months. In other words, ground water flow is present all year round in the
watershed.
The flows observed in the months after the peak flows, i.e., June, July, October,
December and January may be due to recharge from bank storage; bank storage is the portion
of runoff in a rising flood in a stream, which is absorbed by the permeable boundaries of the
stream above the normal phreatic surface (GWT).

47. - 35 -
It is thus implicit that, taking Ahero downstream as the 5
concentration point, and
Kipkelion upstream as the fringe, the watershed has a rather short 6
concentration time, since
only storms of duration greater than the concentration time are able to produce runoff from the
entire catchment and cause high intensity floods (Raghunath, 2006). This means that even the
short rains last longer than the concentration time.
It also follows that the Nyando watershed under study is a fan-shaped catchment (see
Plate 5-1), and may be made up of fern-shaped drainage units; a fan-shaped catchment produces
greater flood intensity since all the tributaries are nearly of the same length, and hence the
concentration time is for every tributary is nearly identical and is less, whereas fern-shaped
catchments have a fringe that is remote, hence the time of concentration is longer and the
discharge distributed over a long period (Raghunath, 2006), hardly resulting in flooding.
Now, upstream flow is observed to be much less than corresponding downstream flow
for the same respective months and therefore under the same storm conditions. The low
upstream flow measured at Kipkelion concentration point indicates that the drainage sub-unit
upstream is fern-shaped, and has low drainage density, whereas the high downstream flow
measured at Ahero points to a fan-shaped drainage unit.
The features of the drainage net may be physically described by the number of streams,
the length of streams, the stream density and the drainage density (see Figures 2-2 and 2-3).
The stream density of a catchment is the number of streams per square kilometre. The drainage
density is the total length of all stream channels (perennial and intermittent) per unit area of the
basin and serves as an index of the areal channel development of the basin; drainage density
varies inversely as the length of overland flow and indicates the drainage efficiency of the
basin. A high value indicates a well-developed network and torrential runoff causing intense
floods (fan-shaped catchment), whereas a low value indicates moderate runoff and high
permeability of the terrain (fern-shaped catchment) (Raghunath, 2006). This information infers
that the Nyando sub-catchment under study has a high stream density and drainage density.
5
The single point or location at which all surface drainage from a basin converges or concentrates as outflow is
called the concentration point or measuring point, since the streamflow is usually measured at this point.
6
The time intervening before the rain falling at the most distant point in a drainage area (i.e. on the fringe of the
catchment) reaches the concentration point is called the concentration time (Raghunath, 2006).

48. - 36 -
Charts 5-10 and 5-11 portray the seasonality in the watershed’s flow regime, upstream
(at Kipkelion) and downstream (at Ahero).
Chart 5-10: 10-year Upstream Monthly Flow Regime
Chart 5-11: 10-year Downstream Monthly Flow Regime
5.2-3 Mean Annual Flow for Period of Record
Mean annual flow is the average flow for the individual year or multi-year period of interest.
When working with hydrologic data it is customary to view the data by water years rather than
by calendar years (Oregon State University, 2005).
Mean annual flow is obtained by dividing the sum of all the individual daily flows by
the number of daily flows recorded for the year. The mean flow for the period of record is
calculated by dividing the sum of all the individual flows by the number of flows recorded for
the period of record. If mean annual flows are available for each year of the record, their sum

49. - 37 -
may be divided by the number of years of record to obtain the long-term mean annual flow for
the period of record (Oregon State University, 2005).
Chart 5-12: Mean Annual Flows
5.2-4 Pattern Analysis
Variation of Annual Flow around Longer-term Mean Flow for Period of Record
The plot of variation of annual flow around long-term mean annual flow for period of record
is a plot of the mean annual flow for each year in the period of record as compared to the overall
mean flow for the entire period of record (Oregon State University, 2005).
Chart 5-13: Upstream Variation of Annual Flow around Mean Flow

50. - 38 -
Chart 5-14: Downstream Variation of Annual Flow around Mean Flow
From these plots, it is possible to identify patterns of wet and dry years.
Flow Duration Analyses
The flow duration curve is a plot that shows the percentage of time that flow in a stream is
likely to equal or exceed some specified value of interest, such as a design flow of some
specified value, or to show the discharge of the stream that occurs or is exceeded some percent
of the time (e.g., 80% of the time) (Oregon State University, 2005).
Chart 5-15: Flow Duration Curve (Arithmetic Plot)
A flow duration curve is a plot of discharge vs. percent of time that a particular
discharge was equaled or exceeded. The area under the flow duration curve (with arithmetic

51. - 39 -
scales) gives the average daily flow, and the median daily flow is the 50% value (Oregon State
University, 2005).
Chart 5-16: Flow Duration Curve (Log-normal Plot)
A flow duration curve characterizes the ability of the basin to provide flows of various
magnitudes. Information concerning the relative amount of time that flows past a site are likely
to equal or exceed a specified value of interest is extremely useful for the design of structures
on a stream. For example, a structure can be designed to perform well within some range of
flows, such as flows that occur between 20 and 80% of the time (or some other selected
interval) (Oregon State University, 2005).
The shape of a flow-duration curve in its upper and lower regions is particularly
significant in evaluating the stream and basin characteristics. The shape of the curve in the
high-flow region indicates the type of flood regime the basin is likely to have, whereas, the
shape of the low-flow region characterizes the ability of the basin to sustain low flows during
dry seasons. A very steep curve (high flows for short periods) would be expected for rain-
caused floods on small watersheds. Regulation of floods with reservoir storage, will generally
result in a much flatter curve near the upper limit. In the low-flow region, an intermittent stream
would exhibit periods of no flow, whereas, a very flat curve indicates that moderate flows are
sustained throughout the year due to natural or artificial streamflow regulation, or due to a large
groundwater capacity which sustains the base flow to the stream (Oregon State University,
2005).

52. - 40 -
Mass Curve Method
The mass curve is a graphical method that is often used in calculations of reservoir storage
capacities. The cumulative quantity of streamflow for the period of record is calculated
beginning at the start of the period. A continuous running sum is then developed that is
calculated to the end of the period of record. Typically, the mean monthly or annual streamflow
values are used for this analysis (Oregon State University, 2005).
The mass curve shows the cumulative runoff volume of the stream. When mean
monthly streamflow values are used (see Chart 5-17), seasonal availability of water and storage
requirements can be identified. When annual values are used, patterns of wet and dry years
Chart 5-17: Mass Curve of Streamflow

53. - 41 -
may be revealed, which patterns may aid in planning reservoir carry-over storage (Oregon State
University, 2005).
Flow Extrema (High Flows and Low Flows)
Tables 5-7 and 5-8 illustrate the instances of high flows and low flows upstream and
downstream respectively, using instantaneous and mean values.
Table 5-7: Upstream Flow Extrema
Table 5-8: Downstream Flow Extrema
Table 5-8 lists the annual floods of every water year; the flows in Table 5-7 are
relatively too small to be considered as floods when compared to those downstream.

54. - 42 -
Table 5-9 summarizes the important output expounded in previous tables and charts in this
chapter, namely mean annual discharge, peak discharge, water yield and low flows.
Table 5-9: Summary Streamflow Statistics
Table 5-9 demonstrates that the highest 7
annual flood during the decade of record
(approximately 130m3
/s) occurred on 11th
May, 2010.
The high water yield generated by the watershed may be due to lack of adequate vegetal
cover following adversarial land tenure, which results in greater overland flow and lesser
infiltration, hence greater floods. It may also be due to relatively heavy downpour compared to
small size of the watershed.
7
The annual flood on a stream is the highest instantaneous peak discharge of the water year.

55. - 43 -
Chapter 6: CONCLUSIONS and RECOMMENDATIONS
6.1 Conclusions
From these studies, it can be concluded that the watershed has bimodal rainfall pattern, with
short rains in May and long rains in August, and with a mean annual rainfall of 1382.1mm.
The watershed also has a bimodal flow pattern, and is flood-prone around April and
September. It has a mean annual flow of 14m3
. Water yield of the sub-catchment can be
inferred from the rating curve once the relation between stage and discharge is established.
6.2 Recommendations
In hydrological studies, great emphasis should be laid on streamflow (as has been done in this
study) because it is the only component of the hydrological cycle that can be measured
accurately.
Of extreme importance is the capability of the stage-discharge relation to be applicable
for extreme flow conditions. Discharge measurements are usually missing in the definition of
the upper and lower end of the rating curve. The extrapolation of these data is subject to serious
errors that can have significant implications for flood management (upper curve) and for water
resources planning (lower curve). Note that the uncertainties related to extrapolation can be
reduced if indirect methods of determining unmeasured peak discharge (for example a rainfall-
runoff model) are used (Braca, 2008).
6.3 Areas for Further Study
The discharge rating curve transforms the continuous stage data to a continuous record of
stream discharge, but it is also used to transform model forecasted flow hydrographs into stage
hydrographs. This is needed, for instance, to estimate the inundated areas during a flood (Braca,
2008).
The rating curve is derived from steady uniform flow. In natural channels, however, the
water surface slope varies for unsteady flow, the cross section changes with sediment
deposition and erosion, and the resistance coefficient changes with bed and flow conditions
(Braca, 2008).
The relation between stage and discharge can be modified by a great number of factors
that result in changes in the shape and position of the rating curve, or in loops in the rating
curve. Principal factors that affect the rating curve include (Herschy, 1995; Kennedy, 1984;
Rantz et al., 1982b): changes to the channel cross section due mainly to scour and fill; growth
and decay of aquatic vegetation; log or debris jams (an accumulation of logs and other organic

56. - 44 -
debris which blocks the flow of a stream of water); variable backwater; rapidly changing
discharge; discharge to or from overbank areas (Braca, 2008).
Backwater effects occur when disturbances tend to propagate upstream. For example,
the effect of a lake (slowing the flow down) or a cataract (speeding the flow up) is felt upstream.
A further very important source of complexity, peculiar of some streams during unsteady flow,
is hysteresis (also known as loop rating) which results when the water surface slope changes
due to either rapidly rising or rapidly falling water levels in a channel control reach. Hysteresis
is most pronounced in flat sloped streams. On rising stages the water surface slope is
significantly steeper than for steady flow conditions, resulting in greater discharge than
indicated by the steady flow rating. The reverse is true for falling stages (Braca, 2008).
Variable backwater, rapidly changing discharge, and flow to or from overbank areas all
result in looped or non-unique ratings and are typically addressed through including additional
parameters, such as an estimate of the water surface slope or the rate of change of the water
surface at the gauge. So, when the type of flow departs significantly from the steady flow state,
the simple stage-discharge relation is no longer sufficient to define the discharge. Another
parameter should be included, i.e. the slope of water surface. Essentially, in these conditions
the ordinary approach, i.e. using the single-valued stage-discharge rating for the computation
of discharge records, is not applicable: the discharge rating under conditions of variable
backwater and for highly unsteady flow cannot be defined by stage alone (Braca, 2008).
Last but not least, it is important to end the debate and finally establish whether the
Kano sub-catchment – and the greater Nyando catchment – has a bimodal or a trimodal rainfall
pattern. Recent events, such as the torrential downpour experienced in late November and early
December, 2019, point to changes in the bimodal trend. Of equal import is research on the
effect of global warming on potential changes in established rainfall and flow regimes.

57. - 45 -
REFERENCES
Braca, G. (2008). Stage-discharge relationships in open channels: Practices and Problems.
FORALPS, Dipartimento di Ingegneria Civile e Ambientale. Trento, Italy: Universita degli
Studi di Trento.
Guzha, A. C., Rufino, M. C., Okoth, S., Jacobs, S., & Nobrega, R. B. (2017). Impacts of land use and
land cover change on surface runoff, discharge and low flows: Evidence from East Africa.
Journal of Hydrology: Regional Studies, p49-51. Retrieved from
https://doi.org/10.1016/j.ejrh.2017.11.005
Lehre, A. (n.d.). Methods of Streamflow Data Analysis. Dept. of Geology, Humboldt State
University. Retrieved from
http://www.humboldt.edu/~geodept/geology531/531_handouts/streamflow_data_analysis.pdf
Oregon State University. (2002 - 2005). Streamflow Evaluations for Watershed Restoration Planning
and Design. (P. Klingeman, Ed.) Retrieved from OSU:
http://water.oregonstate.edu/streamflow/
Raburu, P. O., Okeyo-Owuor, J. B., & Kwena, F. (2012). Community Based Approach to the
Management of Nyando Wetland, Lake Victoria Basin, Kenya (1st ed.). Nairobi, Kenya:
McPowl Media Ltd.
Raghunath, H. M. (2006). Hydrology: Principles, Analysis & Design (2nd ed.). New Delhi, India:
New Age International (P) Ltd.
Rwigi, S. K., Opere, A. O., & Mutua, F. M. (2012). Comparative Case Study of Rainfall-Runoff
Models Over the Nyando River Basin. J. Meteorol. Relat Sci., vol. 6, p37-39. Retrieved from
http://dx.doi.org/10.20987/jmrs.08.2012.604
Subramanya, K. (2008). Engineering Hydrology (3rd ed.). New Delhi, India: Tata McGraw-Hill
Publishing Co. Ltd.
United States Department of Agriculture. (2015). Part 630 National Engineering Handbook. In
Chapter 5: Streamflow Data.
V.P. Singh et al. (eds). (2018). Water Science and Technology Library. Hydrologic Modeling: Select
Proceedings of ICWEES-2016. vol 81, pp. p524-532. Singapore: Springer Nature Singapore
Pte Ltd. Retrieved from https://doi.org/10.1007/978-981-10-5801-1_36

58. - 46 -
APPENDICES
Appendix A: Consistency Test Results

59. - 47 -
Appendix B: Rainfall Data Analyses