Application of dilation mathematical morphology algorithm to detect transition zone of ecosystem

Introduction
According to the regime shift and extension of the desert boundary in last decades, it is important to identifying of transition zones, which are the most likely area to crossing over into the desert state in arid ecosystems. Shifts in relationships between climatic variables especially soil moisture and productivity relationships are not easily traced, mainly because the responses of biological processes to variation in rainfall and soil moisture are characterized by several temporal and spatial scales. Between-seasons differences in magnitude and frequency of rainfall events, and in seasonal rainfall amounts and distribution, add to the difficulties in defining threshold values of ecosystem responses to changes in rainfall characteristics. Thus, detection of changes in ecosystem productivity should be established through a long-term study. This is especially true in the case of annual vegetation, which exhibits no carry-over effects from previous seasons, i.e., the productivity of each growing season reflects only that specific season’s weather conditions. The impact of changes in climatic conditions on productivity is, therefore, complex and combines the effects of several driving factors. Identifying and mapping such thresholds is difficult because of the high diversity of vegetation, soil and bare rock patterns and ecogeomorphic instabilities in these regions. Embedded recovery and erodibility potentials in these patterns convert together by erosion vegetated and dilation of complementary phase or conversely. The adjacency of ecological (vegetation) and morphologic (soil and rocks) parameters in transition zones is illustrative high potential to cross over into an irreversible threshold. Therefore, early detection of transition zones can be effective in controlling and preventing of desert borders extension. For this purpose, application of mathematical morphological algorithm like dilation provides the tools for implementing spatial domination of this patterns. The application of morphological analysis is a new technique in environmental sciences spatially in Iran and one of the most basic morphological operations is dilation. Dilation algorithms effects on shape and structure of a feature and adds pixels to the boundaries of features in an image. It performances in binary images. In a morphological algorithm such as Dilation, the value of each pixel in the output image is based on a comparison of the corresponding pixel in the input image with its neighbors. By choosing the size and shape of the structure element, it is possible to construct a morphological operation that is sensitive to specific shapes in the input image. We assessed the application of dilation algorithm to detection of transition zones in part of Khorasan Razavi province.
Materials and methods
In this research, we applied a mathematical morphological algorithm to detection of the transition zone between arid and semi-arid regions in the part of Khorasan Razavi province, because of an altitude and climatic gradient. The altitude difference in this region is more than 800 m from the west to the southeast which effects on rainfall fluctuation about 200 mm through this gradient. As a result, vegetation covers as an ecological stabilizing factor is under the influence of such gradient. An image processing was applied for detecting desert areas based spatial differences and geomorphic properties considering lithology, soil and vegetation relationships. The required data in this survey was surface reflectance images of MODIS for June 2004. These data sets are known as "MOD 09 Surface Reflectance 8-day L3 global" product with spatial resolutions 500 m and are computed from the MODIS Level 1B land bands 1,2,3,4,5,6, and 7 (centered at 648 nm, 858 nm, 470 nm, 555 nm, 1240 nm, 1640 nm, and 2130 nm, respectively). The product is an estimate of the surface spectral reflectance for each band as it would have been measured at ground level if there were no atmospheric scattering or absorption. In the first step, binary images as inputs for implementing of dilation algorithm was obtained by implementing of spectral angle mapper algorithm for spectral un-mixing on surface reflectance images. In this step, it was necessary to determine surface conditions as a function of the relative cover of vegetation, bare soil, and rocks. Radiometric and geometric enhancements were applied before image classification and processing. However, in order to the accessibility of normal state of ecological conditions, the monthly image of June was produced by using a weighted average. Such parameterization may be obtained by using a spectral angle mapper (SAM), which classifies pixels based on the spectral distance between the pixel integrated reflectance and the representative spectra of these three surface cover types. The morphological algorithm of dilation was performed by laying the structuring element on the image. Since the size and shape of structuring element (kernel) are important in morphological operations, in the second step, choosing a proper structuring elements was required for implementing dilation technique. So, three kinds of 3×3 kernels of two-dimensions were examined and the most appropriate kernel was chosen for this purpose. Where the origin of the structuring element coincided with a pixel with 1 value (60 in this research), there was no change and moved to the next pixel. But where the origin coincided with a pixel with 0 value, made black all pixels from the image covered by the structuring element. Finally, the fringes of dilated phases were extracted by subtracting the dilated cover fractions from the main image. Transition zones are then identified as boundaries between erodible and recovery area by a combination of two dilated layers.

Results and Discussion
In result of implementing of the spectral algorithm, the surface conditions were determined as a function of the relative cover of bare soil and rocks, and green shrubs. In this context, we have two complementary phases of vegetation and bare soil and rocks, which recovery potential concerns the domination of shrub and other green vegetation and erodibility refers to the domination of bare soil and rock surfaces. Bare soil fraction and shrub cover change inversely along the altitude and climatic gradient in the case study. Implementing of dilation by a structuring elements of two dimensions which all of the elements involves 1 value, provided a proper covers which surrounded of each pixel completely. The implementing morphological algorithm of dilation on both cover types by a 3*3 kernel resulted in a significant extension of cores of maximal cover proportions. In the other words, the operation of dilation set the maximum value of all pixels in the input pixel's neighborhood for the value of the output pixel which in this study the maximum value was 60 (or 1). Subtracting the dilated cover fraction revealed a most distinctive narrow boundary zone characterized by the convergence of differences between the inverse extension potentials of shrubs and bare rock cover. The results have proved that there is a significant variation in shrub and soil fraction distribution along these climatic gradients.

Conclusion
In this research, a new methodology was developed for mentioned approach based on the analysis of potential inverse trends of erosion and recovery embedded in heterogeneous patterns of vegetation, soil and rocks cover in transition zones. Also, we hypothesized that this heterogeneity in itself contains important information regarding the formation of desert thresholds in transition zones. This information includes the mutual trends of inverse recovery and erodibility potentials in spatial patterns. The methodology applies spectral mixing analysis to map surface conditions and uses mathematical morphology algorithms of dilation to detection of transition zones, where two complementary phases are located adjacent each other. Accordingly, we could separate area with a high rate of erodibility potentials into the southeast of the case study. Although, the desert boundary were detected, but the result indicates an expansion of desert boundary into the North West. The new methodology has proved application of morphological algorithms in the detection of arid ecosystems fringe. Also, it may be implemented in wide regions of semi-arid to arid transitions, providing information that is instrumental in identifying eco-geomorphic changes under global climate change and changes in human disturbance regimes.