Hydrological Update Of The Watershed Resources Management Model And Gis-Based Application To The Humid Tropical Environment Of South-East Nigeria

Abstract

Hydrological models are useful tools for prediction and understanding of hydrological

phenomena which underpin watershed resources utilization. A watershed model simulates

hydrological processes in a more holistic manner across the entire watershed area. Most

watershed models are fairly complex; interactions among watershed hydrological components

are essentially nonlinear and take place on a wide range of temporal and spatial scales. The

Watershed Resources Management (WRM) model was developed in Canada in 1991 and

published by Mbajiorgu (1995a & b). The objectives of this study were to: (i) update the

hydrological component models of evapotranspiration and canopy interception, (ii) develop a

MAPWINDOW GIS Input Application for WRM model input files preparation and data

processing, (iii) employ the Shuffled Complex Evolution – University of Arizona (SCE-UA)

optimization scheme for automatic calibration of WRM model, and (iv) validate the WRM

model for application to the humid tropical environment of South-East Nigeria.

The spatial structure and numerical solution scheme of the original WRM model were retained

for distribution of hydrological responses and parameter specification. Recent advances in

physically-based modelling of hydrological processes with particular regard to canopy

interception and evapotranspiration were implemented. Modular components of the model are as

follows: Initialization routine; Timing routine; Rainfall-event routine; Ponded-infiltration

routine; Runoff routine; Saturation-runoff routine; Kinematic-f1ow routine; Conservationstructures

(terraces) routine; Grassed-waterways routine; Hydraulic-structures (reservoirs)

routine; Culvert routine; Evapotranspiration-event routine; Baseflow routine; Soil-moisture

accounting and Subsurface-lateral flow routines. WRM input application was developed as a

plug-in to MAPWINDOW GIS using Visual Basic programming language. The input application

facilitates analyses of geospatial data, extraction of the watershed drainage network, and creation

of WRM model input files. Model parameters often do not represent directly measurable entities

but must be estimated using other means as well as model calibration. Automatic calibration was

implemented by the SCE-UA method. The model was calibrated and validated using observed

runoff and sediment yield data from the Upper Ebonyi River Watershed, South-East Nigeria.

Three statistical evaluation techniques, namely, Coefficient of determination (R2), Nash-Sutcliffe

Efficiency (NSE), and Percent bias (PBIAS) were employed to evaluate the WRM model

application, in addition to graphical comparisons of simulated/predicted and observed time

series.

An input application software plug-in to Mapwindow GIS was developed for WRM model.

Optimum values for six parameters of WRM model were generated from the calibration

technique. Simulated runoff rates and sediment yield data were generated for the Upper Ebonyi

River Watershed and compared/matched with observed time series from June to October, 2013.

From the results obtained during calibration and validation phases of the model application, the

coefficient of determination values were 0.83 to 0.99 for runoff prediction and 0.99 for sediment

yield. NSE values of 0.62 and above were obtained for runoff prediction and of 0.57 and above

were obtained for sediment yield prediction. PBIAS values of calibration and validation phases

of runoff prediction were less than ± 12% which is considered a very satisfactory model

performance. However, for sediment yield simulations, the model did not perform as well since

PBIAS values ranged from 42% to 77% with the acceptable limits being ± 55%. The WRM

model as hydrologically updated has thus been satisfactorily applied to a humid tropical

environment quite distinct and different from the Atlantic Canadian environment of the model’s

original development.