System memory based rainfall-runoff models for the shakkar river watershed of narmada basin
Loading...
Date
2015
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
JNKVV
Abstract
ABSTRACT
The principle objective of this study was to develop and verify the system memory based runoff prediction models on sequential and integrated time scale basis as per the hypothesis, for Shakkar river watershed, comprising an area of 2223 sq. km, of the Narmada basin, M.P., India. The watershed has maximum and minimum elevations respectively 314 m and 1154 m above MSL (mean sea level). The average annual rainfall is about 1245 mm. The daily rainfall data was collected from Land Record Department, Collectorate, Narsinghpur and Land Record Department, Collectorate, Chhindwara for the years 1994 to 2014. The daily runoff data was collected from Central Water Commission, Narmada Division, Paryavas Bhawan, Bhopal. The runoff data were converted into millimeter before subjected to analysis. The dynamic models based on rainfall-runoff processes of a watershed fluvial system was developed in the present study on sequential and integrated (weekly) time scale basis. The qualitative performances of models were ascertained by estimating the values of absolute prediction error (APE), integral square error (ISE) and the coefficient of efficiency (CE). In the present study the permissible limits for APE, ISE and CE were taken respectively as 30%, 10% and 60%, that is the prediction should satisfy the criteria of the APE less than 30%, ISE less than 10% and CE more than 60%.
The runoff prediction models developed on daily and weekly basis for the study area can be summarized as,
Two types of memory based runoff prediction models viz., linear and non-linear were developed by using the daily data series of three consecutive years from 1994 to 1996 of active period (June to September) only. Both the models consider the present rainfall, antecedent precipitation index (API), antecedent runoff index (AQI) as input. The values of coefficient of multiple determination (R2) for the linear and non-linear models were found equal to 0.67 and 0.86 respectively, on the basis of R2 value and prediction performance, the non-linear memory based model was found considered more appropriate than the linear model for the study area. Memory based linear and non-linear weekly runoff prediction models were developed by using only active weeks’ (23rd-39th meteorological weeks) data series of three years, ranging from 1994-1996. The coefficients of multiple determination (R2) values were found equal to 0.86 and 0.96 respectively. The values of APE, ISE and CE for different year used for verification under study, reveals that the weekly non-linear runoff prediction model is better than the linear model.
The following are the salient conclusions obtained from the present study,
1. The fluvial system of Shakkar river watershed exhibits a strong memory on both the sequential and integrated time scale basis, and only past three successive events were found to influence the present event.
2. The first event, immediately preceding the current event was found to have more impact on it, in comparison to other preceding event, and the weights determined for the three successive antecedent events affecting the current event were 44.84%, 32.13% and 23.03% respectively.
3. The non-linear memory based daily runoff prediction model was found more appropriate than the linear model for the Shakkar river watershed of Narmada basin on the basis of coefficient of multiple determinations (R2) and prediction performance.
4. The memory based weekly rainfall-runoff non-linear model was found more appropriate than the weekly linear model for prediction of daily runoff volume for the study area, the inclusion of variable NORW (number of rainy days in a week) was not found to show any significant impact on the value of coefficient of multiple determination (R2) and thereby on the performance of the model.
Description
Keywords
runoff, precipitation, surface water, area, irrigation, fruits, agricultural engineering, hydrology, land resources, soil water