Feasibility study on changing in the structure of the land degradation model based on landscape ecological metrics and artificial neural networks (case study: Arak)

In recent decades, the population of urban areas has increased dramatically and many vegetations have fallen prey to the onslaught of urbanization. Human beings, regardless of the main factors of the environment and its tolerance, have caused excessive destruction of the environment and endangered the survival, growth and health of current and future generations. On the other hand, in recent years, the ecological approach of Landscape in assessing environmental impacts has been greatly welcomed due to the use of the concepts of spatial integrity as the main basis in environmental planning. The processes and phenomena that exist in environmental systems are often dependent on many variables and there are very complex relationships between components that make it very difficult to analyze. This problem always causes errors in the accuracy and precision of model predictions, so examining the performance of networks and finding the best type of neural network to achieve acceptable and more reliable results is one of the priorities and modeling concerns. As a result of using these concepts to evaluate the appearance of the land, while saving time, the evaluation of the outcome of the activity can be determined cumulatively and in the shortest time. Land degradation model is one of the environmental impact assessment methods that quantitatively represents the effects of human activities in region or watershed scale. In fact, the degradation coefficients that are obtained by this model is an index of the damage done by human activity in a region. the main problem is that this model at some stage of implementation, relying on expert judgment and in addition need to extensive field work. so in this study investigated the relationship between land degradation and landscape ecological metrics that are descriptive index of landscape. Understanding this relationship and identifying metrics that have a strong correlation with degradation coefficients, is very effective to investigation feasibility use of these metrics in degradation model in order to future objectivity. The area of Arak city, using the political border map and in ArcGIS 9.3 software environment was 315 networks (networks 5 km 5 km and 25 km2) on the ground, some of which were full network and some incomplete networks. The degradation, ecological vulnerability and physiological density and then, degradation coefficient were determined. In the next stage, landscape metrics using Patch Analyst software based on the land use map was calculated. Finally, the correlation between the degradation coefficient and landscape metrics using the SPSS 17 software was analyzed statistically. After reviewing the described results, metrics with correlation with destruction numbers that were common to all 4 levels of study were determined in each user. The TLA measure was omitted because all networks are the same in all networks and applications because of the same size. Agriculture: NumP, PScov, TE, AWMSI Range: NumP, TE, AWMSI City: NumP, PScov, TE CA, Barren lands: NumP, PScov, TE, MPE, AWMSI CA, In forest use, there were no measured water areas that had a significant correlation with degradation numbers and were common to all 4 levels studied. The results showed that the metrics that were surveyed, two metrics: total edge (TE) and the number of patches (NumP) had the highest correlation coefficient with land degradation. By changing the scale (network size), the result will not change, so it can replace the degradation model. The findings of the degradation model indicate the destructive effects of industrial and transportation activities in Arak. In this research, artificial neural network was also used. ‌Multi-layered non stero peryspector, in comparison with other methods, has a high accuracy and predictive power and modeling of the factors influencing it. Network estimates were evaluated by MAE, RMSE and R2 criteria. In natural resources, due to the confusion and scatter in the data and the flexibility of the feed-through neural network with two hidden layers, data processing usually begins with two hidden layers. Therefore, network training was performed with four neurons in the input layer, three neurons in the first secret layer, one neuron in the second hidden layer and one neuron in the output layer. The network estimates were evaluated by MAE, RMSE and R2 criteria. Comparing the actual volume output with the network output predicted by the network is another way to evaluate the model. As can be seen, the black dots are the measured dots that indicate how much volume has been added. Based on the results obtained from the implementation of the demolition model in Arak city, it was found that destructive activities such as pollution of factories and various industries, "motor vehicles" and domestic, commercial, etc. heating sources, which were observed in most work units, as well as climatic and geographical conditions of the city. Arak (being surrounded on three sides to the mountain, towards the west and southwest winds and adjacent to Miqan desert) are the main causes of destruction. According to the results obtained in Arak, out of a total of 94 work networks, 37 networks need to be reconstructed due to their location within the boundaries of protected areas, and 22 networks due to high physiological density and high intensity of destructive factors. 21 networks are undevelopable due to faults. Existence of powerful and active faults such as Kooshak, Ensrat, Tafresh, Indes, Tabarteh, Talkhab, and Leopard spring are the most important faults in the province, which indicates the potential for earthquake risk in this province. 35 networks have the first to third development priorities. These networks, which are mainly located in the western and southwestern parts of the city, have low physiological densities and are located farther away from the city of Arak, where the severity of destruction is still high. There is no crisis and the future development of the city should be directed to these areas. Industrial growth has created many job opportunities as a result of the migration of large numbers of people from the cities, villages and villages of Markazi province and other provinces to meet the needs of the growing population. Urban and industrial areas are in this direction. Finally, considering what has been said about the natural sensitivity of the region's ecosystem and the results of the destruction model in this city, it seems that with the continuation of the current trend, there is a possibility that the intensity of development and economic activities will exceed the natural capacity of the region. In which case it will be very difficult and even impossible to restore it to its original state; For this reason, it is necessary to implement large-scale development projects that have a great destructive effect.