Evaporation Water Balance In Arid Area Anbar Governorate – Iraq
Estimation of water balance for ungauged basin in arid area environment is a major challenge. The main problem is there are no precise equations to estimate evaporation and surface runoff in arid area due to lack of data in these regions. Multiple linear regression (MLR) stepwise and backward regres...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | http://eprints.usm.my/56386/1/Evaporation%20Water%20Balance%20In%20Arid%20Area%20Anbar%20Governorate%20%E2%80%93%20Iraq_Ahmed%20Saud%20Mohammed.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Estimation of water balance for ungauged basin in arid area environment is a major challenge. The main problem is there are no precise equations to estimate evaporation and surface runoff in arid area due to lack of data in these regions. Multiple linear regression (MLR) stepwise and backward regression methods were used to develop evaporation and surface runoff models. The results showed that the evaporation developed model in linear regression method has proven its efficiency and its ability to predict evaporation and the superiority against most important models that used for estimating the evaporation. The results for evaporation developed model were R2(0.923), NAE (0.134) and NSE (0.91) better than Thornthwaite and Blanny Criddle models with results of R2(0.884), NAE (0.583) and, NSE (0.278) and R2 (0.91), NAE (0.324) and NSE (0.611) respectivily. The significant influence factors are temperaure, wind speed and sunshine. To identify the parameters for surface runoff and to select the significant groups for main factors of runoff prediction model in catchments, three groups of independent variables have been established for MLR analysis.The results showed that the best surface runoff model for Group 2 backward regression method with R2 (0.744) and NAE (0.146) and NSE (0.722) where the significant influence factors were rainfall, catchment slope, catchment area and runoff coefficient. To improve the accuracy of runoff prediction model, similiar three groups of MLR surface runoff model were analysis for two ANNs models and AI techniques(SVM).The results indicate that MLP showed better results compare to RBF and SVM methods for the predictive process, where the surface runoff MLP Group 2 produced the best results compared to other models. The results of surface runoff MLP Group 2 were in Training Phase (R2= 0.846, RMSE 0.160, NAE = 1.251, NSE = 0.846 ) while in Testing Phase (R2= 0.788, RMSE 0.182, NAE = 0.628, NSE = 0.775). Regression equation for evaporation model was integrated in GIS software (ArcGIS 10) to map the spatial distribution for monthly and seasonal evaporation, water surplu for whole catchment study area. Runoff regression equation was used to estimate the sub-catchments runoff in the study area using transposition approach. Transposition of surface runoff data process was carried out to estimate runoff volume in sub-catchment study area (ungauged area). The runoff volume ranged between 1,321,732 m3 to 2,488,979 m3. Spatial distribution for runoff volume were carried out using GIS environment on the entire study area. |
|---|