Evaluation of Trend in Vegetation Variations using Time Series Images and Mann-Kendall test over Gavkhuni Basin

Introduction
Investigation of the future trend of changes in the density and health of the vegetation due to climate and land use/land cover changes can be useful in ecological and environmental studies using time series data especially in dry and semiarid brittle ecosystems. Time series analysis is a powerful tool for investigating the trends of natural phenomena in the past and present. The Satellite Time Series provides the most appropriate data for analysing the trend of changes in natural and human phenomena and predicting for the future. A time series is the collection of statistical data collected at regular intervals. Determining the trend (increasing, decreasing and unchanged) of the vegetation can be one of the ways to help manage vegetation monitoring. In this regard, the use of statistical tests to determine the trend of time series data is essential. There are different methods for determining the process, which are divided into two sets of parametric and nonparametric methods.
Since satellite imagery is one of the best time series illustrations, the use of these data in trend studies is fruitful. Vegetation indices are used to monitor and evaluate vegetation dynamics using satellite imagery. NDVI is the most famous useful in plant monitoring. The NDVI, is used as a useful indicator of photosynthesis capacity identification, usually used to examine environmental and ecological changes. Therefore, remote sensing through the NDVI satellite index can measure surface vegetation changes due to the strong correlation between the vegetation and this index. The NDVI is also reliable for identifying vegetation stress because, as a result of degradation of vegetation in the land, vegetation characteristics such as health, density are majorly changed
Materials and methods
Monthly MODIS NDVI times series images from the MOD13A3 series, were collected from the site of NASA from 2003 to 2014 (144 images). Then the images from Gavkhuni basin have been extracted to determine the trend of vegetation changes.
Mann-Kendall test was presented based on the significance of the Tau Kendall correlation coefficient (τ) and then developed. Correlation coefficient measures the relationship between two series of variables to determine if the first variable increases with increasing the second variable, or decreases or the patterns of change are not relate together. This method is widely used to handle time series. In general, Mann-Kendall statistical test can be used to determine the uniform trends of ubnormal distribution data based on the ranking. Given that the Mann-Kendall test, in addition to the trend, also specifies the type of trend occurring, this model does not require the input of normal data. In the test, for time series data trends, positive values indicating incremental trends and negative Z values represent decreasing trends.
Results
Vegetation fluctuations of mean monthly MODIS NDVI time series images of Gavkhoni basin from 2003 to 2014 were presented. The maximum vegetation cover is observed in spring months. This increase reached its maximum during the course of the study in 2007, and then reached its maximum four months later and then sharply declined in 2008. In 2009, vegetation cover in Gavkhuni basin is relatively modest. In fallowing, the fluctuations are almost the same, but declined sharply in 2014. In addition, the highest vegetation density is observed in the western and central regions of Gavkhuni basin. Also, this figure shows that vegetation density is high in 2003-2006 and has declined significantly by 2014. Accordingly, the vegetation cover increased in January 2003 through January, a dramatic drop in February, a slight decrease in March to May, a modest increase in June to July, a slight increase in the months of July to September, in October And November has been a slight decrease, and December has seen an increase in vegetation.
The trend of long-term vegetation changes in Gavkhuni basin based on 144 images of the monthly MODIS-NDVIwere implemented in the Idrisi Tersset software. At first 144 time series images were converted into TSF file format. The resulting trend map indicates that the vegetation trend is incremental with a value of τ equal to 0.68 to a decreasing trend with the value of τ equal to -0.65. Accordingly, the trend of increasing vegetation is more dispersed in the north-eastern and central to west parts of the basin, and slightly to the south of Gavkhuni basin. In addition, the largest decline in vegetation occurs in the central regions of the east and south of Gavkhuni basin. This trend is evident in the northern and western basins as well as in the south of the basin. In addition, decreasing trend of vegetation is more concentrated in areas of basin which referring to the use/land cover map are aquaculture. This trend indicates a sharp decrease in the rain-fed agriculture. Referring to the land use/land cover map, the vegetation cover of the Gavkhuni wetland during the study period has also been subject to decreasing or unchanged trend. The vegetation of the rocky areas of the north-east of the basin has also increased marginally. Accordingly, the most extensive areas indicates unchanged trend of vegetation in Gavkhuni basin.
Discussion & Conclusions
The results showed that the largest decline in vegetation was observed in 2007, when the catchment area was faced with drought at this time. Also, the annual average vegetation map in 2014 with the least amount of vegetation density has been encountered. Due to the fluctuation of vegetation, the need for serious attention of planners is required because it can be said that vegetation as the representative of environmental health also reflects the characteristics effective biophysics in growth, such as adequate water, undiluted and nutritious soil, etc., provide both the protection of wildlife and soil conservation against erosion. Also, since vegetation cover has declined from 2003 to 2014 in the months of April and May (when is the beginning of the growth season in the Gavkhuni basin), it indicates a crisis in agriculture in the Gavkhuni basin.
The regional variations of vegetation in Gavkhuni basin from 2003 to 2014 through the time series images of MODIS NDVI was determined using non-parametric Mann-Kendall Tau and contextual tests, and areas with increasing, decreasing and unchanged vegetation cover were determined. The results showed that the extent of the vegetation variation trend as well as the extent of the decreasing trend in the Contextual Mann-Kendall method was more than Tau Mann-Kendall, and the trend in most of the area remained unchanged, which only indicates whether vegetation is present in this area or not. This case is achievable by comparing the trend map with land use / land cover map in 2014 or following years.
In addition, it was revealed that the decreasing trend with the sever vegetation decline has occurred in the central regions of the Gavkhuni basin and along the Zayandehrud River and also in the urban areas, which can indicate the high urban growth and land use /land cover change from the vegetation to the construction site in this basin. Accordingly, a sharp decreasing trend was observed in the northern and western regions as well as in the south of the basin, which with referring to the land use /land cover maps and type /vegetation maps, these areas are highest proportion of rangeland vegetation and fertile farming along the Zayanderroud River. These lands affected by land cover/land use and have been severely subjected to land degradation due to the dryness of Zayandehrud Rive in recent years. This urged urgent measures to prevent land degradation. In addition, since the highest reduction in temporal monitoring in the Gavkhuni basin has been observed in the months of April to June, and these months are growth season and cultivation in the Gavkhuni basin, this necessitates a serious actions to mitigate the risks. Also, the study of vegetation trends showed that during the study years, saline lands in the Gavkhuni basin increased and the agricultural land was significantly reduced.