Vegetation Growth Monitoring Under Coal Exploitation Stress by Remote Sensing in the Bulianta Coal M

Journal of China University of MiningAvailableonlineatwww.sciencedirect.comDIRECT·J China Univ Mining Technol 2007, 17(4): 0479-0483Vegetation Growth Monitoring Under CoalExploitation Stress by remote Sensingin the bulianta Coal mining areaLU Xia, HU Zhen-qi, LIU Wei-jie, HUANG Xiao-yanInsitute of Land Reclamation and Ecological Restoration, China University of Mining Technology, Beijing 100083, ChinaAbstract: Coal exploitation inevitably damages the natural ecological environment through large scale undergroundexploitation which exhausts the surrounding areas and is the cause of surface subsidence and cracks. These types oflimits growth of vegetation, which is a very important indicator of a healthy ecological system, nly an impact on anddamage seriously lower the underground water table. Deterioration of the environment has certavegetation growth under coal exploitation stress by remote sensing technology provides advantages such as large scalecoverage, high accuracy and abundant information. a scatter plot was built by a TM(Thematic Mapper)infrared anded bands. a detailed analysis of the distributional characteristics of vegetation pixels has been carried out. Resultsshow that vegetation pixels are affected by soil background pixels, while the distribution of soil pixels presents a linearattern, Soil line equations were obtained mainly by linear regression. A new band, reflecting vegetation growth, hasbeen obtained based on the elimination of the soil background. a grading of vegetation images was extracted by meansof a density slice method. Our analysis indicates that before the exploitation of the Bulianta coal mining area, vegetationgrowth had gradually reduceially intermediate growth vegetation had been transformed into low vegetation. Itay have been caused by the deterioration of the brittle environment in the western part of the mining area. All thesame, after the start of coal production, vegetationhas gradually improved, probably due to large scale aerialseeding. Remote sensing interpretation results proved to be consistent with the actual situation on the ground. Fromtailed research on vegetation growth needs to be analyzedKey words: coal exploitation; linear regression; density slice; vegetation growthCLC number: TP 751 IntroductionAt the same time, it is an important step during theprocess of ecological restoration near coal miningThe process of coal resource exploitation damages areasand changes the natural environment by such occur-Remote sensing has characteristics that can dealrences as subsidence and abundant cracks. More over, with large scale problems, has short repetition periodsit may damage surface vegetation. These kinds of. and provides abundant information. Therefore, it be-damage can induce serious ecological and environ- comes a very useful tool for investigating such thingsment problems and increase the conflict between theas local environmentssurveying, exploiting mineraldemand for resources and environmental protection resources, monitoring soil erosion and land use. Inand ultimately the preservation and self-modulation the area of vegetation research, national and internafunctions of the original ecological system, which tional experts have recently largely focused on thewill be, to some extent, impacted. vegetation is an identification of vegetation types and vegetation in-important part of the global environment and also its diceso-3. However, research into vegetation growth,best indicator. How to obtain effectively vegetation leaf area index (Lab and vegetation stress by multigrowth under coal exploitation stress is always a de- sp中国煤化 T is relatively raretection problem in monitoring a mining environment.sually use hyperCNMHGeceived 02 April 2007: accepted 17 June 2007roject 2003AA322040 supported by the National High Technology Research and Development Program of ChinaCorresponding author. Tel: +86-10-62331339-8006: E-im2Journal of China University of Mining Technology17No4spectral remote sensing data to analyze vegetation at every level of coal measure gradually increasesbiomass, various stress and chemical parameters such from high to low. It shows clearly that the quality ofas nitrogen and chlorophyll content and their concen- the coal production at the top part is higher than at thetrations"-4.However,we all know that hyperspec- bottom part. This phenomenon fully conforms totral remote sensing data is very expensive and very Xierte' s Law of coal degeneration.difficult to obtain. In a limited number of situations, 2.1.3 Situation of coal exploitation stressextracting information on vegetation growth byThe Bulianta coal mine has now been exploited formulti-spectral remote sensing technology becomes ten years or so. Many years of over exploitation cer-imperative. In this paper we mainly analyze vegeta. tainly has caused some damage, especially to thetion growth near the Bulianta coal mining area in the vegetation. Field investigations show that coal exarid and semi-arid area under coal exploitation stress, ploitation has resulted in the exhaustion of large areasing the method of three time phase TM (Thematic under the surface. Exhausted areas cause surfaceMapper) of remote sensing data. The distributional cracks. The width of the cracks is usually very narrowcharacteristics of vegetation pixels have been ana- but the number of cracks is large. Almost everylyzed in a scatter plot constructed by TM 4 and TM 3 working face can cause hundreds of cracks where thebands. Soil lines have been extracted from the scatter direction of most cracks is along the direction of explot by linear regression and the TM 4 infrared band ploitation while other cracks are perpendicular to thehas been rotated with the soil line in order to elimi- direction of the working face. Moreover, the length ofnate the impact of the soil background. Then, the new the cracks along the direction of exploitation is theband, which can entirely reflect the information on same as that of the working face. As well, exhaustedthe growth of the vegetation, will be sliced by a areas can cause subsidence on the surface, where thethreshold. The setting of thresholds is largely deter- vegetation has already been moribund for some time.mined from the inflection points of the frequency The soil type in the Bulianta coal mining area is anaccumulation. Hence, the grading of the remote sens- aeolian sandy soil. In this type of soil, whose struc-images of vegetation growth has been achieved. ture is unconsolidated, water penetrationThe method introduced above may provide ecological strong and the capacity to hold moisture very smallrestoration and environmental protection with very The soil fertility is very poor. The major plant typesgood technical supportare typical steppe vegetation: deciduous trees, broadleaf scrubs and sandy vegetation, whose characteris-2 Experimenttics are those of a short growing season, a long dormancy stage, poor canopy closure and a low coverage2.1 Bulianta coal mine and exploitation stressper cent. The main vegetation varieties are thyme2.1.1 Introduction of the Bulianta coal mining areagrass and sand sagebrush, etc. Therefore, in this coalmining area the environment is very vulnerable, theThe Bulianta coal mining area is one of the major ecology fragile with seriously damaged vegetation,mines of the Shendong Company 5. It is located in combined with many years of coal exploitationthe south-eastem part of Yinjinhuoluo Qi, Yikezhaomeng, Mongolia, about 35 km from Yijinhuoluo 2.2 Data source and remote sensing image pre-Qi. The mine was opened in 1997, with a designedproduction capacity of 6 million ton per year. In re-The remote sensing data used in our paper consistscent years, the production has been well over the de- of TM data obtained from the Landsat satellite for thesigned production capacity. In 2004, the production time phases of August 1986, August 1990, July 1995,was over 10 million ton The coal area for the Buli- July 2000 and July 2006. For some reason, the remoteanta coal mine is about 225 km" and the provensensing data for 2000 misses almost half of the entirelogical reserve are about 0.5 billion ton Thestudy area and the data for 2006 just misses a smallmine can be exploited for about seventy yearpart of the study area. In order to monitor vegetation2. 1.2 Coal measuregrowth effectively and make certain comparisonsThe Bulianta coal mining area belongs to the coal between each research result, we mainly used themeasure of the Jurassic period 69. There are six 1986, 1995 and 2006 3-time phases of the remoteworkable and partial workable coal measure from sensing data based on the consideration of the time ofhigh to low, i. e,1,243, 44,5 and 54. Among exploitation over a fixed number of years. At thethese, the 2 coal measure is the most important part same time, topographic scale maps of 1: 50000 whichof all the workable measure. The coal measure found include the Bulianta coal mining area were obtainedin this mine has the following characteristics in its The中国煤化工 ages used infavor: the mine has abundant water, (2 the coal paperhas a low ash content, it has a low sulfur content imagCNMHGnO clouds in theand it has high heating value. The type of coal is a In order to accomplish a geometric cnon-caking coal. The average reflectance of vitrinite initially calibrated the topographic maps, then madeLU Xia et alvegetation Growththe topographic maps as reference images and made the tm 3 band and the y axis the TM 4 bandthe remote sensing images as base images. Thirdly,Fig. la shows that the line at the bottom is the soilwe set up the system as a GK projection and chose 42 line. The left end of this line indicates hare and wetgeographic control points. Finally, the geometric cor- soil, whose reflectance is very low in both TM 3 andrection was completed by a one degree polynomial 4 bands. The water content in the soil should be at amodel. The error is not greater than 2 pixels and en- maximum here, while the right end represents baretirely meets the requirement of our studyand dry soil, whose reflectance is relatively high inboth bands. The line which is perpendicular to the3 Results and Discussionsoil line reflects the different levels of growth of the3. 1 Theoretical basis of extracting information of vegetation. At the juncture of the two lines, the vege-tation may begin to germinate. The reflectance is verylow because of the impact of the soil backgroundThe reflectance spectrum of typical vegetation has When following the vegetation line, the reflectance invery obvious characteristics. Reflectance in the red thethe red band decreases while the reflectance in theband is very low, while reflectance in the infrared infrared band increases. It shows that the growth ofband is very high. The ratio of the infrared band to the vegetation improves. When the reflectance in ththe red band is an important consideration factor for infrared band reaches its maximum and the reflecanalyzing vegetation by remote sensing. According to tance in the red band its minimum, vegetation growthJensenthe scatter plot(shown in Fig. 1)was estab- is optimum and its canopy is at its most extensivelished by the TM 3 and 4 bands where the X axis isLargest canopy closurest vegetation growthar maturearger canopy closurerowing plantBefore germination乏| Bare and wet seBare and wet soil0 Reflectance in red wave hand0 Reflectance m red wave hand(a)Scatter pl of difTerent canopy cluse(b) Scatter plot of growing period of vegetationFig 1 Typical graph of plant growthSome people are of the opinion that the line per- ate. In some study areas, the scatler plots were estab-pendicular to the soil line in Fig. 1b, can reflect the lished by remote sensing data. If the impact of thefull growth process of vegetationsoil background were not considered, the curved lineof bare and wet soil, where the reflectance in the red specific time. The level of growth of the vegetation g wand infrared bands is very low. When vegetation start- different because of the effect of various factors suched to grow, it moved away from the soil line, the re- as plant type, the type of canopy, physical andflectance in the red band became gradually even chemical parameters of the vegetation. However, alllower while the reflectance in the infrared band in these impact factors should be considered relative tocreased slowly. Gradually the reflectance in thethe type of soil, the moisture and nutrient contents offrared band reached its maximum and that of red the soil and the location of plants On the base of thisband decreased to its minimum value. Later, when the interpretation, we rotated the tm 4 band around thevegetation started to fade, the reflectance in the in- soil line in the scatter plot in order to eliminate therared band gradually reduced, while that of the red soil background. The soil pixels in the rotated scatterband increased somewhat. Finally, when the vegeta- plot are located at the bottom. The reflectance of thetion died, its reflectance characteristics returned to soil pixels in the new coordinate system is very lowthose of the dry soil domain. However, we think that Therefore, the new band after rotation will reflectthis interpretation ignored the fact that the scatter plot fully the growth of the vegetation in the coal miningwas established from remote sensing data obtained area at a specified timeduring a specific time phase. That is to say, in the 3.2scatter plot, all vegetation existed at this specific中国煤化工growth period which is the time when the remoteIf tX-Y. the rotatedsensing images were made. At this specific growth coordCNMH Coordinates in theperiod, the interpretation from the reflection of the original coordinate system are (a, b), the rotated xfull growth process of the vegetation is not appropri- coordinates are(a, b)(shown in Fig. 2). In the figJournal of China University of Mining TechnologyVol 17 No 4ure, a is the angle of rotation by which X axis rotatesThe subtended angle between the soil line and thewhile B is the subtended angle between line segment X axis is a in equation( 3). Before the rotation, theox and the X axis. Therefore, in the X'Y coordinate soil line would have been a linear regression equasystem, the x coordinates(a b')satisfy the following tion.3.3 Dynamically monitoring vegetation growthin the bulianta coal mining areaFrom the analysis above, initially three soil lines,2-3(a, b)for each of the three scatter plots established from thethree pre-processed time phase. remote sensing im-ages, have been constructed. They are as follows:Fig. 2 Diagram of rotation of coordinate systemy=0.99x-5443a'=va?+b Lcos(B-a)y=1068x-73749y=1.1026x-9.2271After a was obtained from the soil line equation,After the above rotation, the computational rela- the TM 4 band was rotated, the soil backgroundtionship between the coordinates of every point in the eliminated and the rotated band reflected the vegetarotated coordinate system and the original coordinate tion growth. The new band was sliced by the thresh-system is as follows:olds determined by the inflexion points in the fre-quency accumulation graph and the grading of thecos a sIn aimages of vegetation growth was achieved(shown in-sina cosa] lyFg.3)(a)ln1986b)In1995(c)ln2006Fig. 3 Slice of remote sensing density image of vegetation growth in Bulianta coal mining areaModerate vegetation growtHigh vegetation grFrom these three pictures, we see that vegetation seeds by airplane in the entire coal mining area andgrowth from 1986 to 1995 was very low. Only along provisions for water facility hadthe river was vegetation growth relative high. As well, cessfully and the area came under closer scrutiny andthe growth of the intermediate vegetation during this protection. These activities increased vegetation covperiod was obviously reduced and degenerated into erage very much. The ecosystem functions in thislow growth vegetation. The reason for this reduction coal mining area have considerably improved. Thismay be that growth is correlated with a further dete- inference agrees well with the interpretation resultsrioration of the vulnerable ecology in the western area. from remote sensing technologyFrom 1995 to 2006, the vegetation growth clearlyimproved; low growth vegetation was transformed 4 Conclusionsinto medium growth vegetation, and some mediumgrowth vegetation became high growth vegetationOver ten years of exploitation of the Bulianta coale know that the Bulianta coal mining area began mining area, local subsidence has occurred and manyproduction in 1997. Our interpretation of the remote cracks appeared on the soil surface. These environ-sensing data shows that the vegetation growth was mental geological hazards have impacted the growthnot subject to subsidence caused by coal exploitation. of v中国煤化工 ng information.onActually the growth of vegetation improved. We have vegethods is very timecontemplated the problem seriously and made someCNMHGRemote sensingfield investigations from which we deduced that in technology has developed to the point where it canugust, 2005, the Shendong Coprovide large scale, abundant information. The resultsLU Xia et algestation Growth Monitoring Under Coal Exploitation Stress by Remote Sensof our investigation into vegetation growth by remote much microscopic information. As well, it is not en-sensing demonstrate that before the development of tirely satisfactory to extract information on vegetationthe Bulianta coal mine, vegetation growth was very growth only from a TM 4 band. Therefore, we con-low. From 1986 to 1995, vegetation growth gradually clude that a study of the impact on vegetation suchdeteriorated. After the start of exploitation of the Bu- water content, vegetation structure and some chemilanta coal mine at 1997, the growth of vegetation cal parameters involved in coal exploitation needimproved. The reason for this improvement maygher spectral and spatial resolution in remote sens-related to aerial seed in 2005. Our interpretation re ing data"9.sults agree with the actual situation on the groundThis fact indicates that obtaining information on the Acknowledgementsgrowth of vegetation by remote sensing technology isfeasible. However, our research results do not indThe authors wish to express their most sincere ap-cate whether coal exploitation has any impact on preciation to He Fenqing who took part in our fieldvegetation. The reason for this is that the information investigations and for obtaining the remote sensingon vegetation growth, extracted from the infrareddata used in ourch. Tremendous thanks areband and its transformation, where the spectral reso- owed to director Zhou from the mining area for hislution is very low, is insufficient owing to a lack of considerable helpReferences[11Chen S P earth information Science and Regional Sustainable Development. Beijing: Survey Press, 1995.(In Chinese)Wang Z H Landslide research along Qingzang road and railway. Modern Geology, 2003, 17 (4): 355-362.(In Chinese)[4] Tian S F, Wang XN. Remote sensing technology application to the goldmine prediction in Yanggao, Shanxi province. Mod-ern Geology, 2001. 15(1): 64-68. (In Chinese[5] ZhaoG X, Li J, Wang J Y Land use change and its environmental effect based on TM data and GIS. The Journal of Agricularal Engineering, 2006, 22(10): 78-81.(In Chinese)[6] Richardson A J, wiegand C L, Distinguishing vegetation from soil background information. Photogrammetric Engineeringnd Remote Sensing,1977,43(2):1541-1552.[7 Benediktsson J A, Sveinsson JR, Amason K. Classification and feature extraction of aviris data. IEEE Transactions onGeosciences and Remote Sensing, 1995, 33(5): 1194-1205[8]Chen Y H, Li X B, Li J, et al. Study on aNdVi and its application for airbone remote sensing. Journal of China Universityof mining Technology, 2004, 33(4): 438-442. (In Chinese)[9] Clevers JGP W. Application of a weighted infrared-red vegetation index for estimating leaf area index by correcting for soilmoisture. Remote Sensing of Environment, 1989, 29(1): 25-37[10] Onisimo M, Andrew K S. Hyperspectral band depth analysis for a better estimation of grass biomass( Cenchrus ciliaris)neasured under controlled laboratory conditions. Internationa! Joumal of Applied Earth Observation and Geoinformation2004,5:87-96[11] Pablo J Z T, John R M, Harmon J, et al. Needle chlorophyll content estimation through model inversion using hyperspectraldata from boreal conifer forest canopies. Remote Sensing of Environment, 2004, 89: 189-199[12]Zhi H, Brian J T, Dury J S, et al. Estimating foliage nitrogen concentration from HYMAP data using continuum removalanalysis. Remote Sensing of Environment, 2004, 93: 18-29,[13] Lydia S, Ustin L S, Dar AR, et al. Deriving water content of chaparral vegetation from aviRis data. Remote Sensing of EnoInment,2000,74:570-581[14] Smith K L, Steven M D, Colls J J. Use of hyperspectral derivative ratios in the red-edge region to identify plant stress re-ponse to gas leaks. Remote Sensing of Environment, 2004, 92: 207-217[15] Jiang Y, Wu L Q. The coaly character and utilization approaches of Bulianta coal mine in Jurassic, Coal Science and Technology,2004,32(12):6971.( Chines[16] Yang XM. The basic characteristics and utilization of coal in Bulianta coal mine, Shendong company. Coaly Technique, 2004,(5): 12-13. (In Chinese)17] Chen B, Liu S F, Mei X F. The new method to extract vegetation pollution information by multi-spectral remote sensing dataModern Geology, 2005, 19(3): 458464(In Chinese)18] Lei G J, Liu S F, Cheng S Y. Remote sensing application to extract vegetation pollution information, Jiangxi Non-ferrousMetal, 2006, 20(2): 1-5. (In Chinese)[19] PuR L Gong P. Hyperspectral Remote Sensing and Application. Beijing: High Education Press, 2000. (In Chinese中国煤化工CNMHG