Comparison of Digital Methods to Extract the Map of Waterways (Case Study: Khezrabad Region, Yazd Province)

Geomorphological features play an essential role in changing the reactions of a basin and they can have significant effects on the erosion and sedimentation potential of watersheds. Meanwhile, the influence of geomorphological facies is significant in changing the amount of soil erosion, and its effects are manifested in the amount of sediment production. Establishing a relationship between the morphology of waterways and the active processes in them helps to correctly understand and predict their response to natural and human changes. Therefore, it is necessary to identify geomorphological phenomena that can influence the structure of waterways. Today, with the use of computers and the existence of capable remote sensing software and geographic information systems, the necessary calculations can be performed with high speed and accuracy. The purpose of this research is to compare the waterway network extracted from digital models of different heights.

Using the algorithms available in the software, a flow direction grid was created and a cumulative flow grid layer was extracted from it. For each cell, the current accumulation layer determines and ranks the number of cells that direct their current to the mentioned cell according to which the cells with the highest numerical values ​​correspond to the concave lines and the cells with values ​​close to zero and zero correspond to the ridge lines. To extract the waterway network from the studied area, DEMs 10, 30, and obtained DEM were used by ARCHYDRO software, which finally produced three maps. The waterway network was extracted from the three mentioned DEMs. In this study, the drainage network map, extracted from the digital topographic maps provided by the country's mapping organization, was used as a measure close to the ground reality. Based on the constructed waterways, the border of the main closed rivers and the area and perimeter of the main basin were obtained.

According to the results of the total length of the waterways, it can be seen that the cell size of 10 meters has the largest value so that the model calculates the length of the waterways more accurately. According to the drainage density values, it is observed that the amount of drainage density decreases with the increase in cell size. This problem indicates that with the increase in the cell size, the lower ranking waterways are removed, and considering that the drainage density is obtained from formula (1) and in this formula, the number of waterways and their lengths have great effects on the drainage density, so it can be seen that as the cell size of the height digital model increases, the drainage density decreases. The drainage density obtained using a cell size of 10 meters is the closest value to reality. Using DEM10 and Arc hydro software, the waterway network map was extracted and compared with the reference map (waterway network map of the Mapping Organization) of the country.

According to the obtained results, by increasing the cell size of the digital elevation model, small details are not considered and the accuracy of the model decreases; as the pixel size increases, the sub-channels are removed, which affects the output parameters. This model reduces the length of waterways and the density of drainage, which is consistent with the results of Davari and Hack's research. As the cell size increases, the length of the main waterway decreases which confirms the results obtained by Ashurlo et al. (2008). However, the drainage density obtained using a cell size of 10 meters is the closest value to reality. Govan et al. (2001) and Hosseinzadeh and Nadaf Sangani (2013) also achieved similar results in their research. The results showed that with the increase in the cell size, the calculation of the length of the waterways is associated with error to the extent that with the increase of the cell size by more than 10 meters, the error rate becomes more significant and far from the actual amount. The parameters obtained from digital height models with a cell size of 10 meters in the Arc hydro model had the closest results to the real standard. Based on this, Zhang and Montgomery (1994) and Yang et al. (2010) considered 10 meters as an appropriate index in their research. According to the mentioned cases, the correctness of this theorem can also be visually verified in the waterway network maps obtained from the digital models of 10 and 30 meters height, as well as the waterway network obtained from the topographic map 1: 50,000 regions compared to control waterway network map (the waterway network of the country's mapping organization). It was also observed that the smaller the cell size, the more accurate waterway network of the region can be obtained. The results of satellite image processing show that among the three indices of NDVI, NIR and LWC, the NIR index shows a more accurate waterway network pattern. This index showed that 90% of waterways created by this index are consistent with waterway network. Also, in the surveys that were conducted in the plains and mountains separately, it was shown that 94% of the created waterways in the plains and 91% in the mountainous areas are the same as the control waterways, which confirms the high ability of this index to detect waterways in plain areas and to prevent the occurrence of errors in the identification of waterways.