نوع مقاله : پژوهشی
نویسندگان
1 استادیار بخش تحقیقات حفاظت خاک و آبخیزداری، مرکز تحقیقات و آموزش کشاورزی و منابع‌طبیعی استان ایلام، سازمان تحقیقات، آموزش و ترویج کشاورزی، ایلام، ایران
2 دانشیار پژوهشکدة حفاظت خاک و آبخیزداری، سازمان تحقیقات، آموزش و ترویج کشاورزی تهران، ایران
چکیده
کلیدواژه‌ها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
Introduction and Goal
In general, the modeling process for any phenomenon or dependent variable using effective environmental factors as independent variables can be generalized to the future in addition to determining the status of current conditions. Therefore, the basic assumption in all environmental modeling is based on the principle of uniformity or uniformitarianism, which can be generalized for the subject of this research, and topographic metrics are one of the main factors in the creation and expansion of gully erosion. This research was conducted with the aim of determining the accuracy of water erosion susceptibility maps based on Wight of Evidence and Dempster-Schiffer weight models using relative operator or receiver characteristic curve (AUC and ROC). Also, the thresholds of effective factors (including low-elevation indices and considered environmental factors) on the occurrence of gully erosion in Ilam watershed were determined.
Materials and Methods
The Ilam watershed is in Ilam province in the southwest of Iran. In this research, Dempster-Shiffer and maximum weight models were used to determine the thresholds and zoning of gully erosion sensitivity. This research was done in seven main steps. In the first step, the research area was selected and the flood event distribution map (dependent variable) was prepared, and the data were randomly divided into two training or calibration groups (70%) and experimental or prediction groups (30%). In the second step, maps of 22 effective factors (independent variables) were prepared. In the third step, the effective factors were selected using the alignment test between the effective factors and the occurrence of floods. In the fourth step, Wight of Evidence and Dempster-Schiffer weight models were implemented and zoning maps were prepared. In the fifth step of validation, evaluation, comparison of zoning maps and prediction of sensitivity obtained from the mentioned models were done in two stages of recalibration and prediction. In the sixth step, thresholds and importance of effective factors were determined based on the best model. In the seventh step, the zoning map and prediction of sensitivity to the erosion event of gully was prepared.
Results and Discussion
In this research, after field surveys, interpretation of aerial photographs and processing of satellite images, 331 gullyes were identified and investigated. The distribution map of gully erosion has been prepared and digitized for 331 gullyes, of which 70% of the gullyes were used to train Wight of Evidence and Dempstershafer models, which includes 232 gullyes. The remaining 30%, which includes 99 gullyes, was used to validate the models from the characteristic curve of the ROC system and the surface area under the AUC diagram. First, the data obtained from the gully erosion zoning maps were fitted separately to the points of 99 gullyes and 232 gullyes using both Wight of Evidence and Dempster Shaffer models, and after transferring the data to the data excel file were analyzed in SPSS software. According to obtained from calibration points (30%), both models had an acceptable percentage of the area under the curve, which indicates the high performance of both models in the region.
Conclusion and Suggestions
Accurate field identification of gully erosion points using GPS device was one of the results obtained from this research. As a result, the erosion distribution map of 331 gulles was prepared and it was checked and recorded for the first time in the researched watershed. One of the advantages and innovations of this research was the use of Alus satellite images and GIS and Saga GIS software. The results showed that the most effective independent variables were extracted on the dependent variable of watershed erosion in the watershed system and 22 grid layers were prepared. Finally, the results of the statistical analysis provide a ranking table of the most important indicators in creating gully erosion in the region, which is the land use index of 37.1%, waterway density with 17.7%, lithology with 13.2%, climate with 9.4%, precipitation 3 5.5% and NDVI with 3.5% and other indicators respectively affect the erosion, respectively of the gully. The results of this research can be promoted and taught, and executive bodies can use the results of this research to take necessary measures to control the gully erosion.
کلیدواژه‌ها [English]
)