Azad Kakehmami 
, Ardavan Ghorbani , Mehdi Moameri , Abazar Esmali Ouri , Zeinab Hazbavi , Sahar Ghafari
, Ardavan Ghorbani , Mehdi Moameri , Abazar Esmali Ouri , Zeinab Hazbavi , Sahar Ghafari
University of Mohaghegh Ardabili
Abstract: (850 Views)
Background and objectives
Remote sensing data, with the comprehensive and accessible nature of big data and the up-to-dateness of spatial data, facilitate the understanding of plant-environment interactions. Integrating such information into species distribution models can lead to the production of a global map of species distribution and be effective in species conservation and restoration. The purpose of this research is to determine the remote sensing predictors including climatic indices, primary and secondary topographic indices and remote sensing indices to predict the distribution of J. excelsa in Khalkhal county of Ardabil province and the northern part of Zanjan province.
Methodology
The studied area includes Khalkhal County in the south of Ardabil province and the northern part of Zanjan province (including the three Counties of Zanjan, Tarem and the northeastern part of Mahenshan) with an area of about 1036742 hectares. At first, by using the land use map and during several stages of study area preliminary field survey, locations with suitable access roads were selected for sample recording. Then, the latitude and longitude coordinates of J. excelsa presence and absence were recorded using Global positioning system device (GPS). A total of 1577 points (presence and absence) were recorded for the species. To avoid the bias caused by sampling during the presence points recording, it was tried to consider areas as presence places that, in addition to the dominance of the species, covered at least one spot with an area of one square kilometer, and also the sampling sites were at least one kilometer away from each other. In order to reduce the autocorrelation and bias of presence data, SDM toolbox and multi-distance method in ArcGIS software were used. The creation of pseudo-absence points was done randomly using the Random Selection tool in ArcGIS software with approximately 1 km intervals. The review of the recorded absence points locations was done by field visit to ensure the absence of the species. Using topographic maps with a scale of 1:25000, a digital elevation model map was prepared using the Spatial Analyst tool in ArcGIS software, and then a slope and aspect map was prepared. In order to prepare the precipitation and temperature layer, the information obtained from satellite images with high temporal resolution (short imaging intervals) was used. Using ENVI software, remote sensing indices (Global Environmental Monitoring Index (GEMI), Leaf Area Index (LAI), Modified Normalized Difference Water Index (MNDWI) and Visible Atmospherically Resistant Index (VARI)) were calculated and rasterized. After preparing the investigated variables layers, Multivariate Adaptive Regression Splines (MARS) and Generalized Linear Model (GLM) were performed in the SAHM software.
Results
Area under curve (AUC) was 0.967 in the GLM and 0.984 in the MARS, which is classified as an excellent level for both models. The most important effective variables in the GLM method are VARI index, slope, temperature, digital elevation model, LAI index, MNDWI index, annual precipitation and GEMI index. The most important effective variables in the MARS method are VARI index, annual precipitation, digital elevation model, slope, MNDWI index, temperature, LAI index, GEMI index and aspect. The suitable habitat for the presence of the species was estimated to be 349569 hectares (33.7% of the total area) in the Generalized Linear Model (GLM) and 340,610 hectares (32.8% of the total area) in the Multivariate Adaptive Regression Splines (MARS) method. The minimum elevation of the species habitat is 450 meters and the maximum elevation is 2800 meters. As the slope percentage increases, the probability of species presence increases and then has a constant trend. There is a probability of the species presence in the precipitation of 300 to 800 mm and in the temperature range of 6 to 18 C. According to the GEMI index, the highest species presence is in 0.18 value and the lowest is in 0.24 value. The highest probability of presence according to the LAI index was observed in 0 and the lowest in 0.05 to 0.15 values. The MNDWI index has a constant trend from -0.5 to -0.1 and after that the probability of presence decreases. The probability of species distribution increases with the increase of VARI index from -0.18 and has an almost constant trend in the range of -0.18 to 0.3.
Conclusion
Different models give different results due to having different algorithms. Therefore, it is more reliable to use several prediction methods instead of using one method. The models presented in this study are valid only within habitat conditions of the study area, and for other areas, they should be tested in several areas with similar plant species to measure the compliance of the model.
Remote sensing data, with the comprehensive and accessible nature of big data and the up-to-dateness of spatial data, facilitate the understanding of plant-environment interactions. Integrating such information into species distribution models can lead to the production of a global map of species distribution and be effective in species conservation and restoration. The purpose of this research is to determine the remote sensing predictors including climatic indices, primary and secondary topographic indices and remote sensing indices to predict the distribution of J. excelsa in Khalkhal county of Ardabil province and the northern part of Zanjan province.
Methodology
The studied area includes Khalkhal County in the south of Ardabil province and the northern part of Zanjan province (including the three Counties of Zanjan, Tarem and the northeastern part of Mahenshan) with an area of about 1036742 hectares. At first, by using the land use map and during several stages of study area preliminary field survey, locations with suitable access roads were selected for sample recording. Then, the latitude and longitude coordinates of J. excelsa presence and absence were recorded using Global positioning system device (GPS). A total of 1577 points (presence and absence) were recorded for the species. To avoid the bias caused by sampling during the presence points recording, it was tried to consider areas as presence places that, in addition to the dominance of the species, covered at least one spot with an area of one square kilometer, and also the sampling sites were at least one kilometer away from each other. In order to reduce the autocorrelation and bias of presence data, SDM toolbox and multi-distance method in ArcGIS software were used. The creation of pseudo-absence points was done randomly using the Random Selection tool in ArcGIS software with approximately 1 km intervals. The review of the recorded absence points locations was done by field visit to ensure the absence of the species. Using topographic maps with a scale of 1:25000, a digital elevation model map was prepared using the Spatial Analyst tool in ArcGIS software, and then a slope and aspect map was prepared. In order to prepare the precipitation and temperature layer, the information obtained from satellite images with high temporal resolution (short imaging intervals) was used. Using ENVI software, remote sensing indices (Global Environmental Monitoring Index (GEMI), Leaf Area Index (LAI), Modified Normalized Difference Water Index (MNDWI) and Visible Atmospherically Resistant Index (VARI)) were calculated and rasterized. After preparing the investigated variables layers, Multivariate Adaptive Regression Splines (MARS) and Generalized Linear Model (GLM) were performed in the SAHM software.
Results
Area under curve (AUC) was 0.967 in the GLM and 0.984 in the MARS, which is classified as an excellent level for both models. The most important effective variables in the GLM method are VARI index, slope, temperature, digital elevation model, LAI index, MNDWI index, annual precipitation and GEMI index. The most important effective variables in the MARS method are VARI index, annual precipitation, digital elevation model, slope, MNDWI index, temperature, LAI index, GEMI index and aspect. The suitable habitat for the presence of the species was estimated to be 349569 hectares (33.7% of the total area) in the Generalized Linear Model (GLM) and 340,610 hectares (32.8% of the total area) in the Multivariate Adaptive Regression Splines (MARS) method. The minimum elevation of the species habitat is 450 meters and the maximum elevation is 2800 meters. As the slope percentage increases, the probability of species presence increases and then has a constant trend. There is a probability of the species presence in the precipitation of 300 to 800 mm and in the temperature range of 6 to 18 C. According to the GEMI index, the highest species presence is in 0.18 value and the lowest is in 0.24 value. The highest probability of presence according to the LAI index was observed in 0 and the lowest in 0.05 to 0.15 values. The MNDWI index has a constant trend from -0.5 to -0.1 and after that the probability of presence decreases. The probability of species distribution increases with the increase of VARI index from -0.18 and has an almost constant trend in the range of -0.18 to 0.3.
Conclusion
Different models give different results due to having different algorithms. Therefore, it is more reliable to use several prediction methods instead of using one method. The models presented in this study are valid only within habitat conditions of the study area, and for other areas, they should be tested in several areas with similar plant species to measure the compliance of the model.
Keywords: Habitats suitability, Remote sensing indices, Software for Assisted Habitat Modeling (SAHM)
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
