شاخص‏هاي سنجش از دوري چه‏ اندازه مي‏توانند موجب بهبود برآورد بار معلق شوند؟

How much the Remote Sensing Indices can Improve Suspended Sediment Predictions ?

در اين پژوهش كارايي شاخص‏هاي ماهواره‏اي و پارامترهاي ژيومورفومتري در برآورد بار رسوبي با استفاده از مدل‏هاي مبتني بر هوش مصنوعي و داده‏كاوي به چالش كشيده شده است. بدين منظور، نخست مدل‏ها به كمك پارامترهاي ژيومورفومتري مستخرج از مدل رقومي ارتفاعي و شاخص‏هاي ماهواره‏اي بهينه‏سازي شد و نزديك‏ترين داده‏هاي دبي و رسوب به زمان تصاوير ماهواره‏اي خروجي مدل درنظر گرفته شد. پس از اجراي الگوريتم‏ها، به وزن‏دهي پارامترها و تعيين ميزان تاثيرشان در پيش‏بيني بار رسوبي معلق پرداخته ‏شد. نتايج نشان داد عملكرد مدل‏ها با ورودي‏هاي مختلف گوناگون است. مقادير RMSE مدل‏ها بيانگر آن است كه در صورت استفاده از پارامترهاي ژيومورفومتري به ‏عنوان ورودي مدل مقدار RMSE بيشتر است و در مقابل با استفاده از برخي شاخص‏ها به‏ عنوان ورودي مدل‏ها ميزان RMSE كاهش مي‏يابد؛ به ‏طوري كه در مدل فرايند گوسي با ورودي پارامترهاي ژيومورفومتري مقدار??/?? RMSE= و در صورت ورودي شاخص‏هاي تصاوير ماهواره‏اي مقدار 7/513RMSE= است. با تلفيق پارامترهاي ژيومورفومتري و شاخص‏ها ميزان دقت همه مدل‏ها افزايش يافته و مدل فرايند گوسي با 026/5RMSE= بيشترين دقت را داشته است. نتايج حاصل از وزن‏دهي نيز نشان داد كه شاخص‏هاي Clay index (average) و b5 (average) و NDVI (max) داراي بيشترين وزن بوده و بيشترين تاثير را در پيش‏بيني بار رسوبي معلق داشته‏اند.

Introduction During recent decades in water resources engineering sciences, the prediction of suspended sediment load particularly in flood areas was highly regarded. Nowadays Methods and artificial intelligence techniques to predict hydrologic have become very popular. In recent studies of various parameters such as the spectral reflection bands of satellite images, land use, geology and climatic data have been used. Landsat satellite images according to their high resolution has good spatial resolution. Da Silvia (2015: 53) spectral calibration multispectral satellite images to assess their suspended sediment concentration. Their results showed that the concentration of suspended sediment has been strongly influenced by seasonal rainfall. The yellow river sediment using Landsat satellite images by Zhang et al (2014:136) were evaluated. The results showed that, using the modified algorithm and recovery appropriate climate models, TM / ETM + can be used to quantify the concentration of suspended sediment at the mouth of the yellow river. In this study, mining indices satellite images and watersheds geomorphometry parameters that derive from the characteristics of the basin surface to evaluate and compare the performance of these parameters to predict suspended sediment has been studied. In this study, methods such as artificial neural networks, linear regression, K nearest neighbor, Gaussian processes, support vector machine and evolutionary support vector machine selected and with purpose check the role of these parameters were used to predict suspended sediment load, to detection of the impact of these parameters is to improve the assessment models. Materials and Methods 1-Study Areas There were 68 catchment areas located in the provinces of Gilan and Lorestan from Iran. (Figure 1) Figure 1. The location and studied stations 2-Data processing Data mining geomorphometry After determining the area of study geomorphometry parameters were extracted. Geomorphometry parameters was extracted from 30 meter area digital elevation model (Table 1) Table1. Geomorphometry parameters extracted from DEM Analytical Hillshading MRRTF Aspect MRVBF Catchment Area Plan Curvature Channel Network Base Level Profile Curvature Convergence Index Relative Slope Position Cross-Sectional Curvature Slope Discharge Strahler Order Drainage Density Stream Power Index Flow Accumulation Suspension Load Flow Directions Tangential Curvature General Curvature Topographic Wetness Index Longitudinal Curvature Vertical Distance to Channel Network LS Factor Watershed Basins 3-The modeling process In this study the input parameters in the prediction of suspended sediment load of data mining models such as linear regression, Gaussian processes, neural networks, k-nearest neighbor, support vector machine and evolutionary support vector machine was used. • Linear regression Linear regression to model the value of a quantitative dependent variable that is based on a linear relationship with one or more independent variables used. • Artificial Neural Network Artificial neural networks including computational models that can be used even if the relationship between inputs and outputs of a physical system is complex and nonlinear, with a network of interconnected nodes that are all are joined together. • K-Nearest Neighbor K-Nearest Neighbor algorithm including the selection of a specific number of vector data then randomly from the set for the simulation period following is a given period. Gaussian process A Gaussian process is a stochastic process which consists of random values at any point in space or time domain so that each of the random variables is normally distributed. • Support Vector Machine Support vector machines are a class of supervised learning methods for classification and regression problems applied. • Evolutionary Support Vector Machine Evolutionary vector machine model use of an evolutionary strategy to optimize its. It offers an evolutionary algorithm to solve the problem of dual optimization a support vector machine. 4-Evaluation Model In order to evaluate the algorithms applied to the data, the evaluation criteria Root mean squared error (RMSE), relative error (Re), Correlation coefficient (r), Absolute error (AE) was used. 5-Weighting parameters In this study, for the weighting input parameters of support vector machine algorithm used, this algorithm coefficients a normal vector of linear support machine as the weight of characteristics determines (Sani Abade, 1393: 521). Results and Discussion At first the different algorithms on the data geomorphometry parameters were applied. The results showed that with using geomorphometry parameters Gaussian process model with RMSE = 10.35 and R = 0.986 is the best model to predict suspended sediment load (Figure 3). In the next phase models, were used on the input data indices satellite images. Then index satellite images and geomorphometry parameters as input been together and models were run on them. Also results showed the Gaussian process model RMSE= 5.026 and R=0.99 has highest accuracy in predicting suspended sediment load. Figure 5. The scatter plot of the observed and predicted values with models A: linear regression, B: Artificial Neural Networks, C: nearest neighbor, D: Gaussian process, E: support vector machine, F: evolutionary support vector machine, using a combination of geomorphometry parameters and indicators images satellite. It can be concluded that the models applied in this study than models that their input are climate data have more accuracy. Also the parameters of satellite images have a greater impact on increasing the accuracy of the models. Conclusion The use of indices satellite images and geomorphometry parameters as model input cause increases the accuracy of data mining algorithms to predict suspended sediment load. The results of the study indicated that satellite imagery indices has been more effective in predicting suspended sediment load and using these indicators increase the accuracy of models more effective than geomorphometry parameters. Therefore, considering the indices of satellite images, Gaussian Process Model with RMSE =7.513 and also, if using the geomorphometry parameters of the Gaussian process model with RMSE =10.35 has Highest accuracy. By combining geomorphometry parameters and indicators has increased the accuracy of all models and Gaussian process model with RMSE = 5.026 had the highest accuracy. The results of weighting also showed influence of indices satellite images to predict suspended sediment load.