Agricultural Meteorology
S. Bayati; Kh. Abdollahi
Abstract
Introduction: Rainfall data are required for planning, designing, developing and managing water resources projects as well as hydrological studies. Some previous studies have suggested increasing the density of the rain gauge network to reduce the estimation error. However, more operational stations ...
Read More
Introduction: Rainfall data are required for planning, designing, developing and managing water resources projects as well as hydrological studies. Some previous studies have suggested increasing the density of the rain gauge network to reduce the estimation error. However, more operational stations require more installation costs and monitoring. Some common techniques including statistical methods, spatial interpolation, information-based theory and combination are used to evaluate and design the network. Chaharmahal va Bakhtiari province is a mountainous region; hence, a denser rainfall network is expected in this mountainous environment. The aim of this study was to evaluate the condition of rain gauge stations in Chaharmahal va Bakhtiari province using two approaches, i.e. geostatistical methods and entropy theory.
Materials and Methods: The main required data set for this study is a time series of rainfall data. These data were collected on a daily scale from the Regional Water Company of Chaharmahal va Bakhtiari. After performing statistical tests, the annual data series was prepared for 46 rain gauge stations. A statistical period of 2000 to 2016 was used. The homogeneity of data was investigated by double mass test and histogram drawing methods using Excel and SPSS software, and the existence of trend in the time series of data was investigated by applying a Spearman test. Then, the adequacy of rain gauges in the gauging network was investigated. Annual rainfall interpolation maps and their standard error maps were prepared using the kriging method. Contribution of each station in reducing or increasing the error in the rain gauge network was investigated by removing each station in a cross validation procedure. The efficiency of the rain gauge network was evaluated using the concept of discrete entropy and the values of entropy indices. The value of keeping the rain gauge stations was determined using the net exchange information index.
Results and Discussion: There was no homogeneity problem and significant trend in the data series. Considering the permissible error percentage of 5%, there is a need to add 15 new rain gauge stations to the network. To apply the geostatistical method, we applied it once without deleting any station; then, the kriging interpolation error was calculated for the precipitation data. Then, only one station was removed at each stage, and both the error and the contribution of each station in increasing or decreasing the error compared to the case without Station deletion were obtained. The results indicated that Ab-Turki, Shahrekord, Borujen and Barez stations were more important than other stations. Two stations namely Chaman-Goli and Ben stations can also be considered as the influential stations in error due to the density of stations in the region and error maps. Similarly, the results of the entropy theory method were found effective in evaluating the design of the rain gauge network. The highest value of H(x) was observed in the data of Armand station (3.26) and the lowest value was observed in Abbasabad station (2.28). Since H(x) shows the uncertainty of measuring data, the maximum and minimum uncertainty were found for Armand and Abbasabad sites, respectively. Based on the Net Exchange Information Index, Bardeh, Bareh Mardeh and Dezkabad stations were ranked 1 to 3, respectively, indicating that they transmit and receive more information than other stations. On the other hand, a number of stations including Dorak anari, Abtorki and Chelo stations had the lowest values.
Conclusion: Due to the vast extent of the area and also considering the permissible error percentage of 5%, the number of the stations in this area was found to be insufficient. Thus, although calculating the kriging error maps showed that some stations do not have a significant share in increasing the error, removing the stations is not recommendable. Regarding the new stations, new 15 rain gauge stations are needed to check out the error maps. According to the field observations, the higher priority should be given to the northwestern area (which had the largest interpolation error) in the first place. For the regions with lower error, such as northeast, east, southeast, west and southwest that do not have rain gauge stations, additional rain gauge stations should be constructed.
Khodayar Abdollahi; Somayeh Bayati
Abstract
Introduction: Curve number (CN) is a hydrologic parameter used to predict the direct runoff depth or the excessive rainfall that infiltrates into the soil. This parameter, which indicates surface water retention, is very important in the processes relating to flooding. Vegetation of the region is a major ...
Read More
Introduction: Curve number (CN) is a hydrologic parameter used to predict the direct runoff depth or the excessive rainfall that infiltrates into the soil. This parameter, which indicates surface water retention, is very important in the processes relating to flooding. Vegetation of the region is a major factor affecting peak flow and flood volume. The peak flow is highly influenced by the land surface characteristics, for example at the time that vegetation coverage is naturally low or while vegetated areas are decreasing, the peak discharges increase as well. In this study, the flood hydrograph of Kareh-Bas Basin was simulated using the HEC-HMS model. The simulation was used to estimate the values of the annual curve number in the basin of interest.
Materials and Methods: Model data requirements for this study were temperature, precipitation, and evapotranspiration and discharge time series. The model was calibrated for the period 2000-2010. Then, the model was implemented independently for simulating of rainfall-runoff for each year without any change in the optimized parameters. The model was calibrated only by changing curve number. The average curve number of the basin for each year was computed using the weighted mean method. The MODIS leaf area index raster maps were downloaded from the Modis site. The maps were converted into ASCII format for spatial statistics and calculating the monthly spatial average. The correlation between the curve number and leaf area index was investigated by a nonlinear curve fitting. This lead to the development of a curve number as a function of the vegetation cover for each year. Finally, the accuracy of the developed relationship was investigated using the Nash-Sutcliffe efficiency coefficient by comparing the curve number obtained from the HEC-HMS model and the simulated values from the new relationship.
Results and Discussion: The obtained Nash-Sutcliff coefficient of 0.58 showed that the HEC-HMS model was capable to simulate the flood hydrograph with relatively good accuracy. The sub-basin spatial mean showed that the sub-basins 1 and 2 take the highest curve number values. This indicates that surface water retention in these sub-basins is less than the other sub-basins, which may lead to a sharper hydrological response or flood. In sub-basins 3 and 4, where vegetation density is higher thus land use acts as a predominant factor in hydrologicalbehavior of these sub-basins, the curve number was lower. The study shows the hydrological response depends on the temporal variation of the land cover, for instance in 2010, when the leaf area index increased by a factor of 1.4, the curve number has decreased to 47. As it is predictable with decreasing vegetation the peak discharge and flood volume was increasing. We found a direct nonlinear relationship between basin scale Leaf Area Index and Curve Number with a correlation coefficient of 0.7, indicating that the variation of the curve number is a function of the leaf area index. The developed model allows calculating curve number values based on the remotely sensed leaf area index. This relationship can be used as an auxiliary function for capturing the vegetation changes and dynamics. The accuracy of the derived equation was evaluated in terms of Nash-Sutcliffe's efficiency coefficient. A value of Nash-Sutcliff coefficient of 0.72 showed that this relationship is good enough for calculating basin or sub-basin curve number values capturing the dynamics of leaf area index.
Conclusions: The obtained Nash-Sutcliff efficiency coefficient from HEC-HMS showed that the model was able to simulate the flood hydrograph of Kareh-bas basin with relatively good accuracy. However, the visual interpretation shows there is a weakness in the simulation of the falling limb of the simulated hydrographs. This may be an indication that the drainage of stored water at the basin was not well-simulated by the model. In general, it can be said that peak discharge and flood volume were under-estimated. By increasing the curve number, the peak discharge values also were increasing. The pair data for spatially weighted values for curve number and averaged annual leaf area index showed that an increase in leaf area index leads to a lower value in obtained curve number. This may result in lower peak discharge and volume of the flood. Such relationships may be taken as a measure for flood control. Meanwhile remotely sensed leaf area index products may be considered as an opportunity to capture the dynamics of the land cover.