فصلنامه سنجش از دور و سامانه اطلاعات جغرافیایی در منابع طبیعی
سال یازدهم شماره 3 (تابستان 1399)

Predicting the trend of land use/land cover chenges in natural range ecosystem via remote sensing techniques and evaluating their potentials by modeling, plays an important role in decision making. The goal of this research is monitoring and predicting land use/land cover changes in Nour-rood basin by CA-Markov in a 60 year periods (1988-2048). Landsat TM (1988, 1998, 2008) and OLI (2018) imagery of similar months (in July) were classified by maximum likelihood method algorithm. Terrestrial reality derived from topographic at scale 1:25000 and aerial photos available in the (GDNR) and (WMM) during 1988-2008 and field visits (2018) were evaluated for accuracy. The accuracy of the production maps calculated with Kappa coefficient. So that the highest and lowest ratio were related to the images of 1998 and 1988, respectively with the values of 0.86 and 0.81. The results were compared with field ground truth to determine the accuracy of results. Random matric used to convert land use classes and the map of land cover of Nour-rud basin predicted, in (2018-2028). The results showed that in (1988-2018), forests and rangelands with excellent and fair cover conditions had decreasing and ranges with good condition, rocks and residential areas had increasing trend. Total area of rangelands decreased from 116206 hectares in 1988 to 106336 hectares in 2018. Moreover, the results of Markov model with more than 85% precision showed the same trend of land use changes from 2018-2048. Excellent rangeland cover conditions, showed decreasing trend, rocky and residential areas will also have an increasing trend until 2048. Markov's prediction model also shows an accuracy of more than 85%. The trend of land use changes during 2018-2048 will be the same as in previous. In whitch case, excellent range condition will have decreasing trend; rocky and residential areas will have an increasing trend until 2048.

Population growth and urbanization have caused many changes in land use and land cover that has been greatly affected by the structure, function and service of the ecosystem. Since 2008, more than half of the world's population has lived in urban areas, which, according to the United Nations, 80 percent of the world's population will be urban by 2050, and this continuous increase will lead to the rapid expansion of urban areas. Most of these cities are located in developing countries, especially in Asia and Africa, where a considerable part of them are arid and semi-arid countries. Estimates show that 85 percent of total Iran's lands are under arid and semiarid conditions climate. Arid areas with dry climate, poor vegetation, lack of water, limited rainfall and very fragile environment are identified. These characteristics cause the vulnerability of urban ecosystems. Considering that the function and performance of the environment depended on the composition and distribution of their structural elements, to better understand the dynamics of land, it is necessary to study the changes in spatial patterns. Information on land use changes over time can predict future changes and also be used to identify land a principled and sustainable design and planning,  also to determine the improper process of land change and prevent its spread. Landscape metrics can be used to identify this spatial pattern and it's change.  Applicability of landscape metrics is for the Quantifying of the landscape change and also it's used for analysis and planning of land uses. Metrics is a useful tool for designing and finding exact relationships between the structure and function of landscape functions. Accordingly, this study has tried to study and measure changes in landscape structure of Yazd city by using landscape metrics for three decades in this region from destruction and fragmentation of natural patterns and help planners and policymakers for the orientation of sustainable urban development. The aim of this study is to quantify and measure changes in spatial patterns of land use in Yazd city by using landscape metrics in two levels of class and landscape during the period of 27 years. Also, the distribution and composition of spatial patterns of land use have been investigated in two levels of green and built-up patches that have a major role in ecosystem function.

In this study, Landsat 5 at 1991 and Landsat 8 at 2017 satellite images have been used. After performing the necessary corrections on the satellite images, the classification was done using the maximum likelihood method. according to the diversity of vegetation in the area, three categories including bare land, vegetation and built-up were identified and classified. In order to investigate the accuracy of classification, error matrix and statistical parameters of the kappa coefficient and overall accuracy were used. The kappa coefficient and overall accuracy of the classification images for 2017  and 1991 are 0.81, 90%, 0.83 96%, respectively. The Fragstats 4.1 software was used to calculate the landscape metrics. In the present study, according to the aim of the study, Class Area (CA), Number of patches (NP), Percentage of Landscape (PLAND), AREA, Radius of Gyration (GYRATE), Euclidean Nearest Neighbor Distance (ENN), and Contagion landscape metrics (CONTAGE) were selected and evaluated at the class and landscape level.

In this research, by studying different metrics in the two scales of class and landscape, it was inferred that the landscape in open land and green spaces are being crushed and discrete over a period. While built-up has become more integrated and more expansive over the period, it shows the destructive effects of human activities on the environment. During the studied period, the highest increase in area to other classes belongs to the built-up class. In this study, the maximum number of patches is related to other classes of a green space class. The number of vegetation patches increased and the number of patches opens land and urban class decreased. The results of this metric along with area metrics show the phenomenon of fragmented in Yazd city. Changes in agricultural and gardening land use to residential areas cause disintegration of vegetation patches. The mean patch size of the built-up class has increased and in two vegetation and open land decreased. It shows that the impervious area in the studied city has increased. The average distance metric of each cell in the patches with the center of gravity in two classes of open land and vegetation decreased and the largest reduction is in the green space class. The Euclidean nearest-neighbour distance metric of patches in all uses has been increased which is related to open land.

The results of the study of metrics changes in the class area show that the built-up has increased by 4346.82 ha in the studied period. The reason for this is the increase in the population of Yazd city, which in 1991 and 2017 were 275298 and 529673  respectively. Therefore, more space is needed for the growth and expansion of the city, which causes the physical development of the construction. Due to the spatial expansion of the city, about 1667.61 ha of the agricultural lands and gardens in Yazd city has been destroyed and integrated into urban infrastructures. Increasing human infrastructure and activities without considering the capacity and ecological capability of this area can cause many environmental problems. Therefore, in order to prevent further degradation of the environment and reduce its quality. Monitoring and evaluation of land use patterns should be measured continuously so that they can be used as a guide to assess the current status of the urban ecosystem.

Nowadays, in order to logical use of water for agricultural products, an accurate understanding of the evapotranspiration process is needed. Evapotranspiration is one of the most significant components of water balance hence it is a key variable for the optimal management of water resources. In this paper, we aim to the analysis of the spatial and temporal and distribution of actual evapotranspiration (AET) at monthly time scale using the METRIC approach, driven by MODIS satellite observations over the Vanak Basin and check the accuracy of the METRIC results with (SEBAL, Surface Energy Balance Algorithm for Land).

There are many methods for correct estimation of point evapotranspiration, such as weighing lysimeters, the Bowen ratio, and the eddy correlation methods. The weakness of the mentioned methods is that these techniques only provide evapotranspiration for a specific site and they can't estimate regional evaporation. The METRIC model was developed by Allen et al., (2007) based on the well-known SEBAL model (Bastiaanssen, 1998). METRIC model is a remote sensing-based method that estimates actual evapotranspiration as a residual of the surface energy balance. Herein, the spatial and temporal distribution of actual evapotranspiration of the Vanak Basin from April to November 2013–2014 was estimated using the METRIC model and using MODIS satellite data, the feasibility of using METRIC was investigated. Vanak Basin is located in the southeastern part of the Northern Karoon Basin. It is geographically placed between Chaharmahal va Bakhtiari and Isfahan provinces. 60 MODIS products of Leaf Area Index (MOD15A2), land surface temperature LST (MOD11A2) and surface reflectance (MOD09A1) in 8-day time step were extracted. The mentioned images were downloaded from the USGS website and the images were re-projected from the Sinusoidal projection to UTM projection. The scale factor for LAI, LST and Surface Reflectance were 0.1,0,02 and 0.0001, respectively. Estimation of ET with the METRIC model begins with energy balance. Data sets such as MODIS observations and weather data from the stations in and near the Vanak Basin are used to calculate instantaneous surface energy fluxes including net radiation flux (Rn), soil heat flux (G) and sensible heat flux (H) in the processing technique. ET at the instant of the satellite image is computed for each pixel by dividing LE values by latent heat of vaporization and density of water.

Throughout this research, the upper limit of the variation of AET showed a gradual increase from April to July in both 2013 and 2014. According to the results, the maximum amount of actual evapotranspiration in 2013 and 2014 for the July month was obtained 244 and 263 mm per month respectively. In general, the results of this paper will help us better understand the variations of regional AET. Comparison of the spatial distributions of AET, LAI and LST in the study area showed that the spatial distribution of AET was affected by two factors, LAI and LST, that Pearson correlation test was used to assess the relationship between two variables LAI and LST with actual evapotranspiration. Based on the results, the regions which had dense vegetation and low land surface temperatures had high AET rates, while in the regions with sparse vegetation and high land surface temperatures, the AET rate was low. The results showed that the trend of changes in the mean monthly temperature is in line with the monthly actual evapotranspiration; the same trend was observed in the case of albedo and net radiation flux. It should be noted that the absence of ground measurements for comparing them to the modelled AET amounts was a potential limitation of the current study.  However, our approach of evaluating AET estimates derived from the METRIC model with the AET estimates derived from SEBAL model is a widely used (as standard approach) approach to tackle such limitations. In the second step of the analysis, this paper compares the estimated monthly AET using the equations of the METRIC versus the SEBAL, for the Vanak Basin in 2014. The outcome of the SEBAL model was used as a reference to compare the results obtained from the METRIC model. The statistical analysis was performed to determine the differences between monthly AET derived from METRIC vs. monthly AET derived from SEBAL. The Nash–Sutcliffe model efficiency coefficient (NSE), Coefficient of Determination (R2) and Mean absolute error (MAE) are used, that the results showed high R2 values and NS coefficients and low MAE values indicate that METRIC is closely related to SEBAL Model in the most of the months. The monthly AET values estimated by the METRIC model versus the monthly AET values estimated from the SEBAL model were evaluated and compared for the Vanak Basin from April to November 2014. Based on the overall results the scatter of estimations is in an acceptable range. In 2014, there was good agreement between METRIC and SEBAL models (R2=0.96–0.99, NSE = 0.93–0.99 and MAE = 1.3–7.53 mm month−1). In 2014, other results indicated that in both models, the upper limit of the variation of AET showed a gradual increase from April to July.

According to the results, the regions with high leaf area index (LAI) and low land surface temperature have more evapotranspiration than other regions with low leaf area index and high land surface temperature. The trend of the time series of LAI index and evapotranspiration in this study was consistent with the trend of changes in the parameters mentioned in the study, which was described by Reyes-González et al (2019) that use of the METRIC model in Dacota.

Assessing the process of land use and land cover (LULC) changes leads to a clear understanding of how humans and the environment interact. Landscape metrics can be used as a basis for comparing different scenarios of landscape or recognizing changes and developments in landscape over time. The use of landscape metrics, while saving time, makes it possible to assess the environmental impact of human activities in the shortest time. Population growth in recent decades, the region's high potential for farming, rising expectations and higher demand for income, low employment rates, tourism in the region, the development of communication routes, as well as inappropriate decisions in recent years have led to many changes in Ardabil province. Given the ecological importance of the study area, it seems that the analysis of the relationship between landscape metrics and LULC changes of Ardabil province in the period 1987 to 2015 is a way to quantify the impact of these changes on the landscape.

In this study, the data of Landsat 5 Thematic Mapper (TM) sensor (1987) and Landsat 8 Operational Land Imager (OLI) sensor (2015) were used in a 28-year period to evaluate the trend of LULC changes in Ardabil province. In order to accuracy assessment and geometric correction of LULC maps, 1:25000 topographic maps of the National Cartographic Center of Iran were used. After preparing LULC change maps, LULC metrics were extracted using the FRAGSTATS software. In order to avoid the production of additional information, based on the literature review and expert knowledge, and according to the appropriateness of the criteria with the aim of studying and paying attention to the correlation between there, a set of landscape metrics related to LULC change was extracted. Then, the index of class area (CA), percentage of landscape (PL), number of patches (NP), mean patch size (MPS), edge density (ED), largest patch index (LPI), total edge(TE) and patch density (PD) were calculated and their relationship to LU/LC changes was analyzed by comparative analysis method.

The biggest change in the mean patch size of LULC in the period of 1987 to 2015 (75186 and 1164.354 hectares, respectively) is related to rangelands with a decreasing trend. Population growth in recent years and the resulting to decrease distances between urban and rural areas, low-density construction, transportation network development, and strip or linear growth have led to an increase in the number of man-made patches over time. Moreover, in order to access the living needs, Due to the climatic potential there is an increase in farmlands by changing rangelands throughout the region, especially in the north (Parsabad and Bilesvar counties) and south (Khalkhal county) of Ardabil province, which reduces the percentage of rangelands (From 58.17 to 53.89%). In fact, the process of rangelands change is fragmented, which reduces the mean patch size and, in other words, reduces the stability of the rangelands. On the other hand, man-made land use types have spread in a very heterogeneous and inappropriate way in Ardabil province, which has increased the common border with natural patch and as a result has increased the influence on natural patch (forests and rangelands) and further destroyed them. The increase in class area, largest patch index, total edge, patch density and number of patches of water bodies in the period 1987 to 2015 are due to the increased construction of dams in this period. The increase in requirement and the desire to produce more, and on the other hand the availability of water resources in this period, has changed the rangelands to farmlands (irrigated farming) and as a result, the development of agricultural levels in the region. On the other hand, with the increase of water resources in the province, in the largest patch index of the patches occurred during the study period in dry farming lands. Evaluation of the number of patches showed that the biggest change during the period 1987 to 2015 was related to dry farming class with an increasing trend (2651 patches), which due to human population growth in recent decades, high potential of the region for agriculture and increasing expectations and demand to earn money, this increase indicates the fragmentation and disruption of the landscape, which has resulted in a decrease in forest and rangelands and an increase in farmland land uses. The highest changes in edge density in the period 1987 to 2015 are related to rangelands with an increasing trend (11.78 m/ha) and the lowest change in edge density is related to the forest with an increasing trend (0.66 m/ha). Comparison of edge density showed that the expansion of man-made lands and the increase in agricultural land use (dry and irrigated farming) have increased the edge density of rangeland patches and have been fragmented more than other. The increase in the agriculture level due to the joining of the patches to each other has caused the decrease in the number of patches and the increase in the mean patch size due to the reduction of fragmentation. From the increase in the number of patches in the study period (from 20065 to 23802 patches), it can be concluded that due to human intervention and occupation, the tendency of land appearance in Ardabil province is towards fine-grained structure and the number of man-made artificial and semi-natural patches which has reduced forest and rangelands. The results showed that residential and farmlands (dry and irrigated farming) during the study period had a perfectly aligned change trend, while the reverse trend with forest and rangelands, which can be due to the direct effect and intensity of human presence in determining the distribution pattern for all kinds of LULC.

Due to the uncontrolled growth of residential and farmlands (dry and irrigated farming) in recent years, to prevent further destruction and also to preserve forests and rangelands, it is recommended to perform land use planning based on structural elements and concepts of landscape. The change in the landscape structure has taken in land use types with different degrees, and quantifying these changes using landscape metrics is one of the issues which can help analyze the pattern of spatial changes.

The Aerosol Optical Depth index is one of the most commonly used indicators for assessing air pollution in various regions, especially arid and semi-arid areas. The arid and semi-arid regions are the main sources of dust particles. Due to locating in the arid and semi-arid region, Iran faces dust storms several times over the year, which have caused irreparable environmental and socio-economic damages to different parts of the country. The southeastern of Iran is one of these regions that is affected by dust storms in the first half of the year (early spring to late summer) due to 120-day winds, and large amounts of sand and dust particles enter the atmosphere each year. Therefore, it is important to study the temporal and spatial changes of suspended particles in the atmosphere, of which dust is a major part of aerosols in these regions. In fact, knowing the temporal and spatial changes of suspended particles can be helpful in providing appropriate solutions to reduce the damages caused by these particles. In this study, due to lack of ground-based aerosol gauge station, aerosol optical depth feature was considered based on the Aerosol Optical Depth (AOD) product of MODIS sensor to monitor and analyze spatial and temporal changes of aerosol concentrations in Iran during a period of 18 years. Annual, monthly and seasonal temporal changes of AOD were investigated using pre-whitening Mann-Kendall trend test.

The daily MODIS level-4 AOD data have been used in this study. The AOD data were obtained from the earth explorer website USGS in 6570 frames for 2001- 2018  and these data were extracted in NetCDF format with programming in Matlab software as annual, seasonal and monthly time series for 13 synoptic stations in the study area. PM10 concentration data were used to validate the AOD product of MODIS. In order to investigate the temporal and spatial changes of aerosol concentrations. The AOD zoning maps were prepared using inverse distance weighted (IDW) interpolation method based on the mean values of AOD. Based on the IDW method, each point/ station has a local effect that decreases with increasing distance, places close to the measuring point will have more weight. Finally, the temporal trend changes of AOD data were determined using the Mann- Kendall trend test in the different time scales.

The results obtained from the validation of AOD data indicated that there was a strong and acceptable relationship between aerosol optical depth and PM10 concentration data at 95% significant level. Therefore, it can be concluded that the AOD data have a good accuracy to investigate and analyze the changes in suspended particles in the atmosphere in the study area. The spatial distribution of AOD in the study area showed that the western and central regions (Kerman, Baft, Sirjan, Rafsanjan) had the lowest concentration of aerosol during the period of 2001- 2018. While, the northeastern, eastern and southern regions (Sistan and Baluchestan province) have been affected by higher concentrations of aerosol. The amounts of AOD have varied from 0.14 to 0.53 in the study period that occurred in the years of 2016 and 2003, respectively. The findings of temporal changes of annual AOD series indicated that the atmosphere suspended particles had the highest amount in the years of 2001, 2003, 2008, 2009 and 2012 at the Zabol, Zahak, Zahedan, Khash and Iranshahr regions, respectively. The seasonal mean AOD had the most concentration in the summer and spring seasons while the autumn and winter had the lowest aerosol concentration. The monthly AOD series showed that the increasing variations of atmosphere suspended particles starts in April and then increases in June and has decreasing changes in July. The results of Mann- Kendall trend test indicated that changes trend of annual aerosol optical depth was decreasing in the most regions except in the Rafsanjan station so that the AOD changes had significant downward trend a 5% level in the Zahedan, Zahak, Zabol, Saravan and Bam stations. The Mann- Kendall statistic of monthly AOD series showed the most frequency of significant trend occurred in May, June, July and October, respectively in the study region. The monthly trend of AOD indicated decreasing changes of aerosol during the study period. As in May, June, July, August, November and February, the concentration of aerosols had a downward trend in the most study stations. The seasonal changes of AOD showed a decreasing trend in aerosols in the summer and spring, while there was an increasing trend in the autumn season.

One of the available sources to access the aerosol optical depth data is the MODIS sensor. Based on our findings, there was an acceptable relationship between AOD product of MODIS and PM10 data obtained from ground-based aerosol gauge stations in the study region. The results of this study showed that the annual mean AOD varied from 0.14 to 0.53 which highest and lowest values occurred in the years of 2003 and 2016, respectively. Annual trend of AOD showed the concentration of aerosols was decreasing in most regions. The decreasing trend can be due to the decline of dust events resulted from an increase in vegetation cover in the study area. The monthly and seasonal mean AOD showed the concentration of aerosols had the lowest value during the autumn and winter in November, December, January and February, while in the summer and spring, the aerosols had the highest concentration in May, June and July. However, the temporal changes of monthly and seasonal AOD were decreasing in the most study regions. Generally, our findings showed the western part of the study area, located in Kerman province, had been faced the lower concentration of aerosols than the eastern part, located in Sistan province, during the study period. In general, the declining of aerosol concentrations in some of the study areas could indicate an improvement in air quality in these regions. It seems that the implementation of appropriate executive and management methods in this region, which has been considered by many managers and decision-makers in recent years, have had a significant effect on the reduction of air pollution.

Gully erosion is a type of water erosion that occurs in many climate areas, from arid to humid areas. This type of soil loss causes the displacement and destruction of soil surface horizons by the accumulation of runoff. In many parts of Iran, in the north, south and central faced with this type of soil loss.  In fact, gully erosion occurs in this area due to the complex topography, erodible soils, mismanagement of soil and land use/land cover. Therefore, in order to protect the soil in these areas, it is necessary that a susceptible map should be available to the managers and policymakers. Many parameters affect the occurrence of gullu erosion, including soil, geology, tectonics, hydrology, land use, vegetation and topography, that have been mentioned in various studies around the world. The topographic indices are the most important parameters in the event of gully erosion, which operates differently in each region according to the physical characteristics of the areas. This parameter also indirectly affects the other indicators or criteria (for example, its impact on the vegetation, climate and soil of the area). Even there are many researches on the gully erosion, but there are only a few studies on the modelling with applying the stochastic approaches. This study is the first attempt to the modelling of gully erosion in the central of Iran with applying the maximum Entropy model and topographic indices that have been applied with using the free of charge of digital elevation model. This study uses a new approach to preparing the susceptibility map of gully erosion in the Semirom catchment in the South of Isfahan province. This area is affected by different types of water erosion, same as; gully, rill and landslide. Also, the purpose of this research is to compare the accuracy of two digital elevation model, ASTER and SRTM with 30 m resolution, (DEM) from USGS website, for the modelling of gully erosion in the study area. The emphasis of this research was on the topography indices because it has most important on the event of gully erosion.

In this research for the prediction of the susceptible areas in the result of the main type of gully erosion, the following steps have been applied; In the first step the locations of some sampled gullies, have been digitized randomly with using the Google Earth (GE) images, aerial photos and fieldwork in polygon shapes for each gully. Subsequently, we converted the polygons and into equally spaced points. In the second step, we determine the most important criteria as the environment layers for the modeling. These topography indices including, wetness Index (TWI), curvature, profile curvature, slope, aspect, catchment area, flow length, elevation, slope, LS factor, Stream Power Index (SPI). The topographic indices have been extracted in SAGA GIS from the SRTM DEM with 30m spatial resolution and were then converted to the ASCII format to run in the model.  Before applying the indices, the DEM was preprocessed with low pass filtering to extract artefacts and errors, like local noise and with using ArcGIS. Subsequently, the DEM was hydrologically corrected eliminating sinks using the algorithm proposed by Planchon & Darboux. The Maximum Entropy Model is a general-purpose method for making predictions or inferences from incomplete information. MEM explores applications in diverse areas such as astronomy, portfolio optimization, image reconstruction, statistical physics and signal processing. The idea of Maxent is to estimate a target probability. In fact, this model needs only the gullies feature (present data). The advantages of this model include the following; It requires only presence data together with environmental information for the whole study area. It can utilize both continuous and categorical data and can incorporate interactions between different variables.

With applying the Maxent model in the Semirom catchment, it was trained using 70% of the mapped points of gully features as the target or dependent variable and 30% of the mapped gully as testing samples. The raster type of environmental layers (topographic indices) as the independent variable. The validity of the model used in this study was assessed using the level below ROC or Area Under Curve (AUC). The ROC curve was automatically generated by running this model for both training and testing data. The AUC for training data for SRTM and ASTER is 0.64 and 0.72 respectively and also for testing is 0.68 and 0.72 respectively. These results indicate that the SRTM elevation model has higher accuracy than the ASTER DEM. One of the reasons for the low accuracy of the ASTER DEM can be due to the impact of vegetation, which has caused terrace-like errors, while in the SRTM DEM, due to the radar nature of the waves, this error is reduced and a more accurate map of this the model has been prepared.  Our results show that SPI index with 40.3% contribution, elevation with 22.7% and convergence with 18% are the most important factors for the zoning of the susceptible areas. Regarding the predicted map of the potential of gully erosion, the area in the central and south of the study area are in the high probability.

This study applied the Maxent model to map the susceptibility of gully erosion in the Semirom catchment in the Isfahan Province, using various topographic effective factors and the Maxent model. Stochastic approaches like statistical mechanics provide a powerful tool to study the relations between locations of gully erosion features and corresponding environmental characteristics. The result of this study can be used for land-use planning and management of the areas with gully erosion for sustainable development in the prone areas. Although the results of this study show the prominent role of the topographic indicators for the prediction of the potential gully map, to increase the accuracy of the modeling results, in the furture researches the other criteria such as land use, vegetation and used soil, etc according to the availability of information to can be applied.