Li CHEN, Jie GAO, Mengdi CHANG,3,4,5, Xuan REN,3,4,5

1.Xinjiang Academy of Environmental Protection Science, Urumqi 830011, China; 2.Xinjiang University Urumqi, Xinjiang 830046; 3.Xinjiang Key Laboratory for Environmental Pollution Monitoring and Risk Warning, Urumqi 830011, China; 4.Xinjiang Engineering Technology Research Center for Cleaner Production, Urumqi 830011, China; 5.Junggar Desert-oasis Ecotone Station for Scientific Observation and Research of National Environmental Protection, Urumqi 830011, China

Abstract [Objectives]To explore the effects of spatial density of farmland shelterbelts on NDVI on the northern slope of Tianshan Mountains.[Methods]Taking the economic belt of the northern slope of the Tianshan Mountains as the research area and using the grid method, the spatial density distribution of farmland shelterbelts can be known.Combining the grid method with NDVI data, the average value of normalized difference vegetation index(NDVI)during the growing period of crops can be obtained.In addition, the protection benefits of the shelterbelts on crops were analyzed through comparing the growth of crops within the protection zone with and without protection.[Results]The grids with shelterbelt distribution were better than the grids without shelterbelt distribution, and the shelterbelt played a great role in promoting crop growth.In the middle stage of crop growth, the protection benefit of shelterbelt was significant.The spatial density of shelterbelts was unevenly distributed in the range of 0.6 to 0.8, and the protection benefits were poor.[Conclusions]It is expected to explore the effects of shelterbelts on crop growth at a larger regional scale, so as to provide a basis for the management and design of shelterbelts in the future, and to provide a theoretical basis for studying the protective effect of farmland shelterbelts on crops.

Key words Farmland, Shelterbelts, Spatial density, Normalized difference vegetation index(NDVI)

1 Introduction

Building a strong ecological security barrier is an important measure for sustainable development of China[1].The ecological environment in Xinjiang is very vulnerable, natural disasters occur frequently, the resistance of farmland ecosystems is weak, and the development of agriculture, forestry and animal husbandry is difficult, and the development of oasis agriculture faces a great challenge[2].In order to better provide ecological barriers and strengthen industrial support, it is necessary to continuously promote the development of forestry, then the forestry can undertake the important mission of ecological construction and industrial development[3].Farmland shelterbelt is an artificial ecosystem.Developing farmland shelterbelt is mainly to improve the farmland ecological environment and ensure high and stable crop yield[2].Farmland shelterbelt is the ecological barrier of the oasis in the desert ecosystem[4].It is of great significance in improving farmland ecosystems, effectively resisting natural disasters, and increasing agricultural income, and is also an important prerequisite and guarantee for development of agricultural economy[5].

The 3S technology refers to remote sensing technology, geographic information system technology and global positioning system.Foreign countries have applied 3S technology to carry out research on shelterbelt planning and ecological evaluation since the 1960s, while China has just started to use 3S technology to conduct research on the ecological benefits of shelterbelt since the 1990s.Liu Kang[6]evaluated the effects of farmland shelterbelt benefits on wheat yield according to the presence or absence of shelterbelts.Using remote sensing image maps and land use data of three time periods, Wei Yaling[7]analyzed the characteristics of dynamic changes in shelterbelt landscape pattern.Liang Yietal.[8]elaborated the application of remote sensing technology in the fields of coastal shelterbelt resource survey, dynamic monitoring, and pest and disease monitoring, as well as the comprehensive application of 3S technology in the management of coastal shelterbelts.The 3S technology has been widely applied in shelterbelts, but most scholars generally use crop yield to analyze the protective effect of farmland shelterbelt[9-13].Most of the research methods are based on sample plot measurement to obtain crop yield[14-15]and require much time and effort.Due to the limitation of calculation methods, it is difficult to monitor the growth process of crops from a large spatial scale and a long time.In addition, there is a shortcoming of fanning out from point to area.As a result, it is impossible to effectively reveal the rules of effects of farmland shelterbelts on crops on a larger regional scale.

In view of the shortcomings of the actual measurement method, we used GIS technology combined with the spatial density method to analyze the spatial distribution and normalized vegetation index(NDVI)of the shelterbelts on the northern slope of Tianshan Mountains.It is expected to explore the effects of shelterbelts on crop growth at a larger regional scale, so as to provide a basis for the management and design of shelterbelts in the future, and also provide a theoretical basis for studying the protective effect of farmland shelterbelts on crops.

2 Data and methods

2.1 Overview of the research areaThe economic belt on the northern slope of the Tianshan Mountains is an area with high concentration of productive forces in Xinjiang Uygur Autonomous Region and relatively developed modern industry, agriculture, transportation information, education and technology, and is also the core area for the construction of the National Silk Road Economic Belt[16].It is located in the hinterland of the Eurasian continent, in the middle of the northern foot of the Tianshan Mountains(80°93′-90°79′ E, 43°42′-45°41′N).In the economic belt, the core oasis distribution area has a temperate continental arid and semi-arid climate, with hot summers and cold winters[17], an average annual temperature of 7.4 ℃[18], an average annual precipitation of 181.99 mm, and an average annual evaporation of 1 948 mm.Oasis farmland is concentrated in belts and areas.With the promotion of the Western Development Strategy and the Belt and Road Initiative, the economy of the northern slope of the Tianshan Mountains has developed rapidly[19], and the farmland area has also increased significantly.

Fig.1 Map for distribution of economic belt on the northern slope of the Tianshan Mountains

2.2 Data acquisitionAccording to the forestry survey of Xinjiang Uygur Autonomous Region in 2017, we extracted the data of water conservation forest, soil and water conservation forest, shelterbelts and sand fixation forest, farmland and pasture protection forest, bank protection forest, road protection forest, and other protection forests on the northern slope of Tianshan Mountains.The 16 d of MODIS NDVI data with a spatial resolution of 250 m×250 m was synthesized, and the NDVI data in the study area were finally generated after stitching, projection conversion, cropping of the research area, and removal of outliers.Finally, we calculated the average value.

2.3 Data processing

2.3.1Normalized vegetation index(NDVI).NDVI can reflect the growth information of crops to a large extent, and it has a high correlation with important growth parameters such as leaf area index and net primary productivity of crops.Therefore, we directly used the NDVI value in the economic belt of the northern slope of the Tianshan Mountains to represent the growth of crops[20].

The calculation formula is as follows:

NDVI=(NIR-R)/(NIR+R)

(1)

whereNIRis the near-infrared band with the wavelength range of 0.845-0.885 μm, andRis the infrared band with the range of 0.630-0.680 μm.

2.3.2Grid method.The grid method, as an analysis method in geographic statistics, refers to the use of grids to summarize and analyze ground elements to reveal the correlation between different geographic elements[21].In this study, according to the size of the economic belt on the northern slope of the Tianshan Mountains, a total of 65 388 grids of 1 000 m×1 000 m were generated.Combined with the extracted shelterbelt data, we calculated the area ratio of shelterbelt in each grid.Using the area ratio field to generate the shelterbelt spatial density raster data, we obtained the spatial density distribution of shelterbelts.

2.3.3Division of presence or absence of protection.According to the relevant findings, we set a 200 m buffer zone for the decoded shelterbelt data.Assuming that the protection scope is 200 m, the scope covered by the buffer zone is the protection area, and the uncovered area is the unprotected area.

3 Results and analysis

3.1 Current status of shelterbelt system constructionAccording to the classification calculation results, it can be seen that the northern slope of Tianshan Mountains is mainly shelterbelt and sand-fixing forest with the largest area of 60 399 ha, followed by farmland and pasture shelter forest area of 54 277 ha, bank protection forest area of 9 840 ha, soil and water conservation forest area of 6 204 ha, road protection forest area of 1 785 ha, water conservation forest area of 1 776 ha, other shelter forest area of 232 ha.

3.2 Spatial density of farmland shelterbeltsAccording to the spatial density distribution of shelterbelts, we can see that the shelterbelts on the northern slope of the Tianshan Mountains are mostly concentrated in the central and western regions, the distribution of shelterbelts in northeastern region is relatively sparse and scattered, and the density is low, which is mainly related to the topography and development time between regions.

Fig.2 Spatial density of farmland shelterbelts

3.3 Effects of shelterbelts on crop growthBy counting the average NDVI data in each grid in the research area(Table 1), we found that the number of grids without shelterbelt spatial distribution was 44 045, and the number of grids with shelterbelt spatial distribution was 21 343.The results indicate that the average NDVI value of grids with shelterbelt distribution was 0.02 higher than that of grids without shelterbelt distribution, and the crop growth in the grid with shelterbelt distribution was better than that in the grid without shelterbelt distribution.

Table 1 Average NDVI with and without shelterbelts

We compared the crop growth in grids with and without shelterbelts.For the data of the protected area and the non-protected area, we calculated the average NDVI of the corresponding days.Fig.3 reflects the changes in average crop growth with and without shelterbelts during the growing period(1-353 d in 2017)in the research area.

Fig.3 Average growth curve for crops with and without shelterbelts

In order to further explore the protection of shelterbelts for each growth stage of crops, we divided the crop growth into three stages.(i)Early stage(1-161 d): the early stage is the sowing and emergence period.(ii)Middle stage(177-273 d): the middle stage is the main growth and jointing period of crops.(iii)Late stage(289-353 d): the late stage is the mature harvest period of crops.Using SPSS software, we performed the paired samplesttest on NDVI data at each growth stage.The results(Table 2)show that: at the early stage, the significance levelP=0.08, which is greater than 0.05, and there is no significant difference; at the middle stage, the significance levelP=0.03, which is smaller than 0.05, and there is significant difference; at the late stage, the significance levelP=0.07, which is greater than 0.05, and there is no significant difference.These indicate that at the middle stage of crop growth, there is a significant difference in crop growth between the areas with and without shelterbelts, and there is no significant difference between the early stage and the late stage.

Table 2 Significance levels of different growth stages

3.4 Effects of different shelterbelt densities on the average value of NDVIIn order to further analyze the effects of different shelterbelt densities on the average value of crop NDVI, we classified the grids with shelterbelts in the research area at an interval of 0.1.From Table 3, it can be known that the density of shelterbelts is divided into 7 grades.According to statistics, because the number of grids in the density range of 0 to 0.2 accounted for 90% of the total number of grids, there is a certain analysis error in the NDVI value.Therefore, to remove errors, data in the density range of 0 to 0.2 and 1 were not analyzed.As indicated in Table 3, the NDVI first decreases and then increases with the variation trend of shelterbelt spatial density.At spatial densities in the range of 0.2 to 0.6, the NDVI values tended to be stable and the protective effect is good.When the density of shelterbelt exceeds 0.6, the density is too large, which leads to the decrease in the average NDVI value and the low protection benefit.After the range is greater than 0.8, the NDVI value rises again.These indicate that the spatial density of shelterbelts in the research area is in the range of 0.6 to 0.8, the distribution of shelterbelt is uneven, and the protection benefit is poor.Grids without shelterbelts accounted for 67% of the total grids.

Fig.4 Distribution of NDVI on the northern slope of Tianshan Mountains

Table 3 Classification of shelterbelt spatial density

4 Conclusions and discussion

From the relationship between the spatial density of shelterbelts and NDVI, we analyzed the protection benefits of shelterbelt using 16 d synthetic MODIS NDVI data.We found that the grids with shelterbelt distribution was better than the grids without shelterbelt distribution, and the shelterbelt played a great role in promoting crop growth.At the same time, we monitored the growth of shelterbelt crops at different times in the research area.We found that in the middle stage of crop growth, the protection benefit of shelterbelt was significant.Finally, we analyzed the changes of NDVI in different density ranges.The results showed that the spatial density of shelterbelts was unevenly distributed in the range of 0.6 to 0.8, and the protection benefits were poor.The difference in the protection benefits of shelterbelts in other ranges is small, and the number of grids without shelterbelt accounts for 67% of the total number of grids.

In this study, we used remote sensing data to replace the traditional measurement method, and analyzed the protection benefits of shelterbelt from the regional scale.Compared with traditional monitoring methods, our method is faster and more intuitive.However, in the spatial classification of shelterbelts, the results may be subject to certain errors due to the influence of classification accuracy.Besides, using the NDVI data to replace plant growth, we realized time series monitoring of crop growth.We analyzed the effects of the presence or absence of shelterbelt on the growth status of crops at different growth stages, and realized the dynamic monitoring of crop growth.