Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → États-Unis → 1993 → Development of a streamflow model for a rural catchment in Kenya

Cornell University (1993)

Development of a streamflow model for a rural catchment in Kenya

Thomas, Michael Kariuki

Titre : Development of a streamflow model for a rural catchment in Kenya

Auteur : Thomas, Michael Kariuki

Université de soutenance : Cornell University

Grade : Master of Science (MS) 1993

A streamflow model was developed for use on rural catchments in Kenya. A simple model structure was appropriate given the general availability of data and the intended applications of the model on ungauged catchments. The main features of the streamflow model are : (1) the physical environment is divided into three zones ; the unsaturated zone consists of multiple hydrological response units (HRU), defined by homogeneous land use and soil type, for which a daily soil moisture balance is maintained ; the shallow saturated and deep saturated zones are modelled as regional aquifers, (2) rainfall and evapotranspiration are distributed according altitude, (3) runoff is determined using the US Soil Conservation Service curve number method, as a function of 5 day antecedent precipitation, (4) evapotranspiration varies according to the soil moisture content, and (5) a weather generator permits Monte Carlo simulations over long periods. The data requirements are moderate, consisting chiefly of daily rainfall, mean daily evaporation for each month, and land use/vegetation and soil type information. The ARC-INFO GIS package was used to assemble the topographical, hydrological, land use/vegetation and soil information into different layers which were then combined to produce a map of the different HRUs and a file of their respective characteristics. A system of categorising land use/vegetation was developed which is based on the hydrological charaCteristics and management of different land use/vegetation conditions. Runoff curve numbers were defined for each category. The soil information was mapped using an approach developed by Liniger (1991), based on landforms and climate. The model was tested on the Naro Moru catchment (172 km2) in Kenya, which has climatic conditions that vary from the glaciated peaks of Mount Kenya (5200m) to the semi-arid Laikipia plateau (1800m). The model was calibrated over a two-year period (09/89 - 08/91) and validated over a different two-year period (01/83 - 12/84). Initial parameter values were extracted from relevant literature. A comparison of the observed and simulated streamflow indicated that minimal calibration was required. The simulated streamflow compared reasonably well with observed streamflow values for both the calibration and validation periods (R2 70-85% for 10-day time periods). This indicates that the model is .appropriate for ungauged catchments. Three types of environmental change were considered to demonstrate the use of the model as a tool for water managers ; (1) land use changes, (2) changing streamflow abstractions, and (3) climate change. Two scenarios of land use change were selected, representing extremes of curent trends in land use change. The first consisted of changing the forest reserve to small scale farming. This resulted in greater streamflow, with a 10% increase in mean flow. The second consisted of changing the land use in the savanna zone and lower footzone, mainly a tree/grass cover, to small scale farming. This resulted in a 24% reduction in mean streamflow. The scenario of increasing streamflow abstractions resulted in a 14% reduction in mean streamflow. However, the model results did not properly address the spatial distribution of water supply and demand. Fi ally the climate change scenario, which used the average climatic conditions as predicted by three global climate change models, resulted in a 78% increase in mean streamflow. Streamflow variation increased with a dramatic increase in streamflow over the wet seasons and an extension of the dry season. This produced less abstraction, although the total volume of streamflow increased. The model calibration and validation process, and the examination of different environmental change scenarios, identified refinements to the model that would improve the accuracy and acceptability of the model for use by water managers in Kenya.

Sujets : Models/Rain/Rural areas/Creeks and streams/Hydrology ;

Google books

Page publiée le 18 décembre 2016, mise à jour le 18 octobre 2019