The effect of land use and discharge changes on the coefficients of suspended sediment rating curve in Zagros forest areas

Extended abstract   1- Introduction   The forests of the Zagros vegetation area cover about six million hectares (40%) of Iran's forest area, which is dominated by oak species. These forests are exposed to a number of threats, including land use change, deforestation, over grazing, deliberate and unintentional fires, pest and disease attacks, and climate change. Although accurate and timely identification of changes at the Earth's surface is the basis for understanding the relationship between human functions and natural events and the use of natural resources, it is important to gather information about continuous changes in forest cover. Field operations require a lot of time and money. Therefore, the study of forest cover changes using satellite imagery has become one of the most important sub-branches in forestry science and is a tool for monitoring and controlling various changes in forest ecosystems. On the other hand, soil erosion in forest areas is lower than the other areas such as agricultural and rangeland areas, but extensive deforestation to produce wood and crops in recent years has significantly reduced sediments. Typically, the concentration of suspended sediments is estimated by the method of measuring the scale of sediment and discharge. Such curves are often generated using linear logarithmic regression of sedimentation and discharge data or conventional data curve in which the relationship between discharge and suspended sedimentation is a logarithmic linear equation. In the present study, land use changes in three forest watersheds in Kohgiluyeh and Boyer-Ahmad and Fars provinces were investigated using satellite imagery. Then, the effect of land use changes, especially good forest and rangelands on the values of sediment measurement equations in the study areas were analyzed.   2- Methodology   The studied area with an area of 2400 square kilometers is part of the Zagros forests, which includes three forest watersheds of Seyedabad and Shah Bahram in Kohgiluyeh and Boyer-Ahmad provinces and Shivzohreh catchments in Fars province. At first, forest cover map of the last three decades (1987-2017) was prepared using Landsat satellite images. After initial pre-processing of Landsat images, the Maximum Likelihood (ML) supervised algorithm was used to prepare the land use map and 11 land use classes were separated. An error matrix was used to evaluate the classification accuracy of the land use classes, which includes criteria of producer accuracy, user accuracy, overall accuracy and Kappa index. Based on this, 532 field samples were prepared using a combination of field visits and Google Earth. After preparing the land use change maps in four time periods of 1987, 1997, 2007 and 2017, the annual discharge changes and the sediment rating curve were extracted in three ten-year periods using the data of the hydrometric station at the output of the catchments. For this purpose, rating curves were drawn in the three periods of the mentioned decade and the values of equation coefficients of the curve were compared during the mentioned periods.   3-Results   The results showed that the total accuracy of the land use map of 1987, 1997, 2007 and 2017 was 87.34, 87.62, 87.07 and 90.4%, respectively, and their Kappa coefficient was 0.85, 0.86, 0.85 and 0.89, respectively. The forest land use had a good producer and user accuracy. Also, in Seyedabad area during the first two decades, soil erodibility (coefficient α) increased, but in the period 2007-2017 due to a significant increase in the area of mostly good forest lands (F1), has decreased. In the catchments of Shah Bahram and Shivzohreh, with decreasing forest lands of mainly weak (F3) and pastures of mainly good (R1), during three consecutive decades, soil erodibility has decreased. On the other hand, with the decrease in annual discharge, especially in the period of 2007-2017 in Seyedabad and Shah Bahram catchments, the value of coefficient b has decreased, but in the Shivzohreh catchment, despite the decrease in discharge in three consecutive decades, it seems that due to the hydrological changes of the river as a result of human activities and the availability of new sedimentation resources, the value of the coefficient b of the sediment rating curve has increased.   4- Discussion & Conclusion   Since the rate of erosion in forest lands is lower than rangeland and agricultural lands, so in Seyedabad catchment, despite increasing the area of ​​medium rangeland lands (R2), with decreasing area of ​​good rangeland lands (R1) along with reducing forest lands during the first two decades of the year, soil erodibility (α coefficient) increased and therefore the value of α coefficient of sediment rating curve during the period 1997-2007 compared to the previous period 1987-1997 increased by about 0.05. However, in the period 2007-2017, despite a slight increase in the area of ​​rangeland lands, due to a significant increase in the area of ​​mostly good forest lands (F1), soil erodibility has decreased and therefore the value of the α coefficient of the sediment rating curve compared to the former period has decreased by 0.11. In the Shah Bahram catchment, over the last 30 years, the area of forest lands has steadily decreased by about 43 square kilometers every three decades, while the area of rangelands has increased by about 39 square kilometers. However, the α-coefficient of the sediment rating curve has been steadily decreasing during the studied periods, because on one hand the area of mostly weak forest lands (F3) has decreased over 10-year periods and on the other hand the area of mostly good rangelands (R1) has increased, so soil erodibility (α coefficient) has also decreased. In Shivzohreh catchment, although the area of forest lands has decreased by about 8 square kilometers in the last 30 years, these forest lands are mainly related to weak forest lands (F3). On the other hand, the area of rangeland lands has increased by about 39 square km, which is mainly related to good rangeland lands (R1), so such changes in the coverage of this area have reduced soil erodibility and therefore the value of the α-coefficient of the curve has been steadily declining over three decades.