Digital image analysis of palaeoenvironmental records and applications

PAPERSof these records is actually originated from structureDigital image analysis ofand/or composition changes of the records. Although thepalaeoenvironmental recordsimages may be different from one record to another due tovarieties of image acquiring procedures for different rec-and applicationsords, these images are actually identical in the digitalinformation format, and their image pattern typies alsoSUN Donghuail.2, LIU Yu2 & TAN Ming3show some similarities. They generally consist of growth1. South China Sea Institute of Oceanology, Chinese Academy of Sci-layers with their special patterns, which allow a compati-ble image processing technique to be carried out for dif-ences, Guangzhou 510301, China;2. State Key Laboratory of Loess and Quaternary Geology, Earth Envi-ferent palaeoenvironmental records to obtain their growthronmental Institute, Chinese Academy of Sciences, Xi' an 710075,parameters.China;The digital image is actually made up of data matrix3. Institute of Geology and Geophysics, Chinese Academy of Sciences,that represents color values. The purpose of image analy-Beijing 100029, ChinaCorrespondence should be addressed to Sun Donghuai (e-mail: sundh@sis is to extract environmental signals from the data matrix.scsio.ac.cn)Data at monochrome image pix are gray index value. ItAbstract Environmental change signals in geological orresolution may be 2, 256 or higher if the image format is 2,biological records are commonly reflected on their reflecting8 or more bits respectively, of which 8 bit gray image isor transmitting images. These environmental signals can begenerally used (fig. 1(a)). Data at color image pix consistextracted through digital image analysis. The analysis prin-of red, green and blue indexes. 24-bit color image, isciple involves section line selection, color value reading andknown as“true color image", gives 256 scales for eachcalculating environmental proxy index along the sectioncolor index (fig. 1(b)), which can express most colors inlines, layer identification, auto-chronology and investigationnatural world. From the image pattern of environmentalof structure evolution of growth bands. On detailed ilustra-records, it can be seen that color value varies consistentlytions of the image technique, this note provides image ana-in any growth direction. Therefore, the image analysis islyzing procedures of coral, tree-ring and stalagmite records.The environmental implications of the proxy index fromcarried out in a typical growth direction. The processingimage analysis are accordingly given through applicationprocedure includes section line selection， color valuedemonstration of the image technique.reading along the section line, growth parameter calcula-tion and their environmental interpretation, layer identifi-Keywords: digital image, environmental records.cation, auto-chronology, etc. These analyzing items arePalaeoenvironmental investigations generally em-compatible for most palaeoenvironmental records.(. i) Section line selection. Section line selection isploy environmental proxy index of geological or biologi-the first step of the analyzing procedure. Section lines arecal records to reconstruct environmental changes. Alt-selected in the growth or accumulation direction accord-hough sophisticated isotope or geochemical proxy indexing to the representative and cleaning of growth layer.have been developed in these records, high expanse andDue to inconsistent developing of records over the devel-long time consuming of measurement procedures limit theefficiency of extracting the environment information.oping surface, the section line is not necessarily continu-Fortunately, environmental and climatic changes areous throughout the sequence and may transfer from Ocommonly preserved on image color variations due toline to the discrete next so that the subsequent imagecomposition and structure change of records. The digitalanalysis is done in the typical developing direction of theimage technique made it possible to extract the color in-records.(_ i) Color value reading. Reading color data alongformation that is associated with environmental changes.For example, valuable results have been achieved fromselected section lines is the second step of image analysis.the tree-ring record using image analysis". Color dataBefore color value reading, the active point for eachread along a scanning-line of image is actually the pri-reading is calculated using the cross point between themary application of image analysis on palaeoenviron-section line and the horizontal scanning line of the image.mental investigation2. In this work, the modern digitalIn simple way, only one point on the section line is read.image technique is applied to palaeoenvironmental re-To remove high-frequency noise from the single pointsearch to build a universal method for image analyzing ofvalue, multiple pix data on the same growth line are readpalaeo- environmental records.中国煤化工le short line perpendicularto ththe growth layer. Special1 Image analysis principle of palaeoenvironmentalalgoMHC N M H Ghe active reading point inrecordsprogramming as growth layers are usually different fromPalaeoenvironmental records generally developthe scanning line of image.growth or accumulation layers. The visual image pattern(i i) Regularity estimation of growth layer. GrowthChinese St亦数据/etin Vol. 47 No. 23 December 20021957PAPERSlayer regularity is indicated by color value variabilitylayer finding”(strong micro-layer is identified as annualwithin the growth layer. It can be estimated using standardlayer) may occur during the processing for some layersdeviation of color values of the same growth layer. Thewith color value outside the threshold value. These mis-pix width used for calculating deviation depends on thetakes can be recovered in two methods. In simply way,curvature of the developing surface of the record.repeat identification using modified threshold will find the(“- 1) Extension rate or accumulation rate calculation.lost layers when identified window is locked around theOn determination of time controlling of records, thelost layers. Alternatively, other layer properties, like layerextension rate or accumulation rate can be calculated us-width and position, are taken into consideration in layering the thickness between any two controlling points.identification through programming. Any of the twoAnnual band markers are commonly used as the timemethods can overcome the“layer lost”or“wrong layercontrolling in the high-resolution climate record.finding”in layer identification.(- v) Seasonality calculation of color value. After(.Vi) Computer assistant auto-chronology. On theobtaining the color series, seasonality of color valueannual layer identification, chronology work can be donewithin a year is calculated in two ways. In simple way,automatically through designing a reciprocal dialog box toseasonality is estimated using the difference between thdelete the annual layers that are identified by the abovemaximum and minimum values of an annual band. Alter-sub-routine.natively, the color value deviation of annual band is usedThe above processing involves the modern digitalto estimate the seasonality.(L V) Annual structure of growth layer.The annualimage technique and the complicated algorism method.Most computer languages are qualified for achieving abo-band structure probably reflects the seasonal configurationve tasks using any popular file format in image processing.of the environment variables that influence the color valueIn our processing program, VB was used as the processingof the records, which is represented by annual curvelanguage, and popular BMP file as the image-processingshapes of color value series.format.(iv) Identification of growth layer.After readingcolor value series, growth layer identification is achieved2 Image analysis methods of high-resolution recordsthrough identifying peaks on the color value curve. TwoHigh-resolution environmental records are particu-substantial jobs are critical in the identification. The firstlarly suitable for image analysis as they generally haveis setting a suitable threshold value according to growthcomplete image with consistent color variation. The mostlayer level to identify. There are two types of growth lay-potential high-resolution records include the coral growthers, the annual layer and micro-layers within annual layers.sequence, tree-ring and stalagmite accumulation in caThe threshold for annual layer identification is generallybonate caves.set a higher value, but some weak annual layers may be(- 1) Image analysis of coral growthsequencelost if over high value is used. To identify micro-layersMassive reef- corals continuously excrete carbonatewithin annual bands the threshold must be set to lowto build their living skeleton during their living period.enough, but the noise of image will be identified as micro- Extension and density of coral skeleton are two primarylayers if the threshold is too low. The second substantialgrowth parameters. Calcification, an indicator of the x-job is algorism of determining the growth layer. There arecumulation flux of carbonate material, is calculated usingtwo cases in program processing. In the case that colorthe production of extension and density. The other growthvalue increases with regard to the last point, if the differ-parameters and growth properties can be calculated usingence between color value of active point and the lastthe above three parameters, including regularity of layer,identified peak value is equal to or higher than the thresh-density seasonalty, annual band structure， etc. Theseold value, the active point is identified as the beginning ofgrowth parameters are generally connected to marinea new dark layer; otherwise goes on to the next pointenvironment. The digital image technique provides a pre-processing. In the case that color value decreases withcise and efficient method for extracting the environmentalinformation. X-ray photographing was conventionallyregard to the last point, if the difference between colorused to produce density image of coral skeleton. On thevalue of the active point and the last identified vale valueX-ray image, the coral growth sequence is composed ofis equal to or higher than the threshold value, the activedark high-density sub-bands altered with light less-densitypoint is identified as the beginning of a new light layer,sub-中国煤化工annual bands can be db-otherwise goes on to the next point processing. The meas-image analysis of coraluring of growth layer width also involves image identifi-sequMHC N M H Gbe selected along maincation. The mediun value between adjacent peak and valegrowth axis. Image data along the section line are grvalues is the target to identify. Identification mistakes ofvalue of image. According to the physical principle of X-“layer lost" (some weak annual layer is lost) or“wrongray photographing, the gray value of the image is a proxy1958Chinese Science Bulletin Vol. 47 No. 23 December 2002PAPERSindex of coral skeleton density if the sample thickness isgrowth parameters, which are conventionally measured onconstantBI. It has been suggested that skeleton densityhe tree-ring cross-section. The digital image analysisreflects the living environment quality of coral colony,provides an efficiency technique not only for measuringcontrolled greatly by seawater temperature, available lightwood density and ring width, but also for obtaining otherand nutrient plankton'4. Investigations on coral growth in growth parameters.the South China Sea demonstrate that the annual bandColor value of reflecting image depends on woodcycles on density image are produced from seasonalmaterial type, density and its structure, of which woodvariation of SST (see the next part). Regularity of growthdensity plays the dominant role in image colorl6-I. Redlayers provides an indicator of poison substance in sea-color is generally the most important index in tree-ringwater. This is because normal grow layers will be distort-images. It has been indicated that light color, lower den-ed to form unhealthy irregular layers when single coralsity sub-band represents the early wood grown in spring,animal assumes different growth rates probably due toand dark color, high density sub-band represents the late-suspension grains or poison substancel5l. From fig. 2(a), itwood grown in autumn. The boundary between late woodcan be seen that the growth layers are obviously distortedand the early wood of the next year has been convention-when the regularity index is high. Annual layer and mi-ally used as the time markers of annual bandsi7]. The alcro-layer within annual bands are two important layernual growth width, the width of early and late wood suttypes in layer identification, which request higher andbands can also be calculated using the identification tech-lower threshold values respectively. The darkest growthnique on a proper threshold given. On the suggestion thatlines in annual bands are generally used as the time con-the color value is a proxy index of wood density, the sea-trollers of the coral growth sequence, which are developedsonality of color value, therefore, reflects the seasonalin the coolest month of each year, January in the Southvariability of wood density, which probably contains in-China Sea (Plate - I). The thickness between two annualformation of seasonality of local climate.controlling points is the annual extension rate. Widths ofInvestigations on the mechanism of tree-ring growthwinter and summer half-year sub-bands can be calculatedindicate that tree-ring density and wood width are greatlyusing layer identification when the medium value betweeninfluenced by local precipitation and temperaturelthe maximum and minimum values of a year is set as thethe relationship between tree-ring parameters and climatetarget to identify. As shown in Plate . I, the annual exten-variables are quite complicated due to varieties of gsion rate of Sanya coral Porites is about 2- -3 cm. Sub-graphy locations and tree speciesl7. For example, in theband variation demonstrates that the winter type sub- bandarid and semi-arid area, the tree growth is sensitive toshows little variation, whereas summer type sub-bandprecipitation, in humid regions, however, the temperatureexhibits obvious variation, which means that the annualis important in tree growth in humid areas. Therefore, theband width is dominated by summer half-year extension.correlation analysis between growth parameters and cli-Density seasonality of coral sequence provides a proxymate variables is necessary to determine the climate influ-index of climate seasonality, high seasonality value gener-encing factors of tree growth in an area. Plate - I(b) givesally appears on the sharp contrast bands in the densitya demonstration of image analysis of a cross-section ofimage. Micro-layer identification suggests that annualcypress from Huangling of northern Chinal9. It can bebands with a clear micro-layer have 12- 13 micro-layers,seen that the late wood shows a little width variation andwhich coincide with the lunar cycle. This reveals the ge-early wood shows an obvious variation. Thus, the width ofnetic connection between micro-layer formation and theannual bands is greatly determined by early wood. Colormoon moving through the tiding process. From Plate. Ivalue seasonality of annual bands suggests that the densityit can be seen that micro-layers also show a positive cor-difference between early and late wood serves as a proxyrelation with annual extension, which implies that obviousindex of seasonal variability of local precipitation. A no-micro-layers generally develop within wide annual bands.table correlation exists between the seasonality and annualBecause skeleton density of the South China Sea coral isband width as well as early wood width, reflecting theirgreatly controlled by SST driven by surface current, thegenetic connection.annual band structure reflects the configuration of winteri i) Image analyzing method of stalagmite. Micro-and summer type surface currents.structure investigation reveals that accumulation layers of(- i) Image analysis of tree-ring. Tree-ring imagesstalagmite can be observed on both transmitting and re-are acquired in two methods. In the first method, the pol-flec中国煤化irinciple suggests that theished tree cross-section is directly scanned to obtain itsaccu木u二gmites is formed from thereflecting image. The second method is similar with thatcom];YHC N M H Gifference layers. Accom-of coral sequence. Transmission image is taken from thepanying organic matter and other dark material plays thetwo-side polished disk of tree-ring section using the X -raydominant role in gray bands of stalagmite image. There-machine. Wood density and band width are the primaryfore, it is believed that the gray value of stalagmite imageChinese S府亦数letin Vol. 47 No. 23 December 20021959PAPERS .reflects the percentage of dark material including organicbands and less dense sub-bands. Systematic work suggestsmatter and other dark material that accumulated simulta-that the annual density cycles on coral sequence resultneously with carbonatelIl. There are two types of accu-from the seasonal variation of marine environment drivenmulation layers in stalagmite, annual layers and non-an-by the climate. Here we try to determine the main climatenual layers. The annual layers and seasonal sub-layers canvariables to control the seasonal variability of coralbe identified as the time markers of chronology seriesgrowth. Abundant references have demonstrated theusing image analysis on their proper threshold given indominant role of sea surface temperature on the coralidentification. Different types of non-annual layer may begrowth!2- 14. Observation has found that massive coralidentified through setting appreciated threshold value inmay live in temperature range of 16 - 36. CI1SI. Calcifica-identification. The accumulation rate is obtained after thetion experimental indicated that the optimum temperaturetime controls are markers using image identification. Itrange is about 26- -29_ (16. Sea surface temperature ofcan be understood that the seasonality of gray value ofnortherm SCs generally varied below this optimum value,annual band indicates the seasonal variability of darkproposing positive correlation between temperature andmaterial accumulation. Structure type of annual bandscoral growth. Fig. 1 gives the comparison of growth vari-probably reflects the seasonal configuration of dark mate-ables with the main environment variables of Sanya. Timerial flux induced by the seasonal climate change.series of coral density are detrended in order to analyzePlate - I demonstrates the image analysis of a sta-the inter -annual and seasonal changes. It can be seen thatlagmite sequence from a karst cave near Beijing. Primarythe density curves show similar annual and seasonalpalaeoenvironmental investigation demonstrates that thevariations to temperature (fig. 1(a)). The correlation ana-predominant accumulation cycles in transmitting imagelysis consistently confirms their correlation (fig. 1(b)). .are annual layers resulting from the seasonal climateMeanwhile, we can see no obvious correlation betweenvariationl. On transmitting images, an annual band con-growth parameters and precipitation. It should be notedsists of a wide sub-band with light color and a narrow sub-that some density characteristics show a poor correlationband with dark color, which is represented by narrowto temperature change, and also, skeleton density andpeaks altered with wide vales in gray series curves. Sedi-mentary work suggests that the annual bands were pro-temperature display the inconsistent variation amplitudeduced in the early season of each year due to fast accu-in some annual bands. In addition, micro-layers within themulation of dark materials within down-passage of groundannual band cannot be ascribed to temperature changes.water resulting from heavy rain after long dry seasons[10.These evidence the nonlinear influence of temperature onTherefore, the gray value of the image indicates the accu-skeleton density and the growing processes of massivemulation rate of dark material that is connected withoral.ground vegetations. Calculation of annual accumulationIt has been found that available light is essential forindicates that annual layers are dominated by light layers photosynthesis of coral alga which is responsible for re-because dark layers show a litte variation and the lightmoving CO2 from coral calcificationl!7-20l. Daily lightlayer varies obviously. Gray value seasonality exhibits acycle, tide process and cloud coverage are three possiblepositive correlation with annual bands, suggesting theirprocesses that could induce available light change:intrinsic connection. Therefore, it is possible to extractcoral growth, of which the former two processes areprecipitation and vegetation information through imagecorded in daily21] and lunar layers17] in the coral sequence.analysis.Observation counting and computer identification of th3 Application demonstration: Image analysis of coralmicro-layers in annual band of Sanya coral Porites indi-Porites of the South China Seacate that the annual numbers of micro-layers are about 12Environmental factors of coral growth are the pri-一13 (Plate - I(b)), showing a lunar tide cycle. Therefore,mary topic of coral research. From time scale, seasonalsome micro-layers of Coral Porites in the northern SCSvariability and the long-term variation trend of coralare explained in the tide process from lunar moving. P1growth are two important aspects of coral growth investi-vious work suggested that cloud coverage could influencegation. Porites lutea samples for this research were col-coral growth through available sunlight. Here we presentlected from northern Xidao Island, 10 km south of Sanyathe precipitation data (an indicator of cloud coverage) ofCity. The investigation mainly focuses on the influencingSanya in fig. 1. We could not see an obvious correlationenvironmental variables of coral growth in seasonal vari-between cland nvpraceand coral growth parameters. Forability and long-term variation trends, using digital imageexan中国煤化I:tuation of cloud coverageanalysis to obtain the growth parameter.|YHC N M H Ghe calcification variation,(_ ) Environmental controlling of seasonal variabil-showing " une maepenuent variations. The salinity ofity of coral growth. Coral Porites is characterized byseawater also displays the rare influence on the coralannual band sequence consisting of altered dense sub-growth that was found in earlier research2211960Chinese Science Bulletin Vol. 47 No. 23 December 2002PAPERSExtension/cm-al Density/g . cm~'Precipitation/m1.4 1.3 1.2 1.10200 4002001200019991998y= -0.0169x+ 1.7371997..R= -0.72乏199519941993.2 t19911990.1L19892025303540SST/C19881989861985198405101522 26 30Calcification/mg .d"'Fig. 1. Correlation of coral growth parameters and climatic index. (a) Comparison between coral growth parameters and climateindex; (b) correlation analysis.Therefore, the sea surface temperature is recognized the optimum value of coral growth. The global warmingas the most important environment variable that influ-of the last several ten years should result in the decreasingences the coral skeleton density. Some micro layersof skeleton density. Therefore, the coral skeleton densitywithin annual bands are genetically associated with thevariation in the last twenty years cannot be explained bysea level changes resulting from the tide process.temperature changes.(- il) Influence of human activities on recent coralA notable change that might influence offshore envi-growth.The long-term variation trend of coral growthronment is human activities. Primary statistic data indicateparameters reflects the variation of living environmentthat economic activities and tourism industry have pro-quality. The image analysis technique is used to investi -gressively grown in the last twenty years. The moderngate recent coral sequences in Sanya tourism offshore.industry system was built initially in the 1980s and be-The results indicate that the coral density and the layercame further flourishing in the 1990s. Along with theregularity of all samples show consistent variation trends.economic development of the region, tourism populationThis trend is characterized by the abnormal density in-has increased by 20 times in the last twenty years, mean-creasing and the layer distorting in the last twenty years,while the industry and economic activities have alsowhich began in the 1980s and progressively developed insimultaneously grown. Human activities could influencethe 1990s (fig. 2). The skeleton density of coral Poritescoral growth in two ways. In one way, suspended grains inseawater from human activities may be trapped into coralsamples remains at the average of about 1 - -1.1 unittissue or skeleton to prevent coral animals from healthy(g/cm) before 1985. It increases to 1.3 unit from 1985growing, and subsequently to form denser skeletonl22 -25.and further to over 1.5 unit after 1996. The regularityMe:中国煤化工n material emitted in hu-exhibits a similar variation over the periods (fig. 2). Fig. 1.manral tissue or skeleton inshows that such variations are not observed on tempera-somd;Hc N M H Ge comparison analysis, itture, precipitation, cloud coverage and salinity curves. Wecan be seen that systematic changes in Coral Poriteshave known that the global temperature has gently in-parameters consistently reveal the stepwise deteriorationcreased in the last twenty years, but it still changed underof the living condition of the offshore environment. ThisChinese St亦数据/etin Vol. 47 No. 23 December 20021961.PAPERSDensity/g. cm"'Densityig. cmiDensity/g cm 3.81998199619941994.1991992199221990i9019881988::19881986i986-19841984 t1982 t1982 11982(c19B01980+198051525351978(b)Regularity/0- 25519761976 .15219741972a)19705 25Fig. 2. Skeleton density variation of present coral Porites from Sanya offshore. (a) Xidao XD3; (b) Xidao XD1; (C) XidaoXD4;the shaded areas indicate the two stages of coral growth trend.process coincides with the onset of local tourism andproxy index of coral skeleton density that reflects theeconomy systems in the 1980s and further developing in general quality of living marine environment. The skele-the 1990s, reflecting the influence of human activities onton density of SCS coral Porites is mostly influenced bythe offshore environment of southern Hainan Island.the seasonal temperature change. The regularity of growthlayer provides an indictor of poison substance in sea water.4 Primary conclusions and further applicationLayer identification of coral sequence includes annual andDigital image analysis of palaeoenvironmental inmicro-layers within annual bands. Annual extension,vestigation involves three procedures. (1) Image acquiring:winter and summer half-year extensions can be calculatedimages of micro-sequences are acquired using transmit-through layer identification. Density seasonality of densityting light under a microscope, whereas images of large-serves as a proxy index of climate seasonality. Some mi-scale records are acquired using the special scanner or thecro-layers in annual bands are connected with the lunardigital camera. Density images are generally obtainedcycle through the tiding process and the moon light cycle.using X-ray photographing. Special equipments are need-The annual band structure of SCS coral sequence reflectsed for fluorescence and infrared image acquiring. (2) Im-the configuration between winter and summer type sur-age analysis: including section line selection, color dataface currents.reading along the section lines, growth parameter calcu-The color value of tree-ring reflecting image islating and layer identification. (3) Environmental implica-mainly associated with wood density, which is influencedtion of the image parameters: the composition and densityby precipitation and temperature changes in the monsoonof the record are generally the main determining factors of中国煤化工:lor seasonality of colortheir images. The environmental implication of the calcu-valueerence between early andlated parameters is interpreted through the image analysislateYHCNMHGasonality of precipitationprinciple after color value is designated as specific envi-or temperature in northern China. The width of annualronmental significance.band, early and late wood are calculated using the layerThe gray value of X-ray image of coral sequence is aidentification technique through inputting a proper thres-1962Chinese Science Bulletin Vol. 47 No. 23 December 2002PAPERShold value in processing.1985, 4: 95.The gray value of stalagmite sequence serves as an4. Lough, J. M., Barmes, D. J. Environmental controls on growth ofthe massive coral porites, Journal of Experimental Marine Biologyindicator of dark materials in the sequence, particularlyand Ecology, 2000, 245: 225.organic substance, which is connected with ground vege-5. Brown, B. E, Tudhope, A. W., Tisser, M. D. et al, A noveltation. Gray seasonality reflects the seasonal change 0mechanism for iron incorporation into coral skeletons, Coraldark material accumulation. The annual layer structureReefs, 1991, 10:211.. Cook, E. R., Kairiukstis, L, Methods of Dendrochronology, Dor-indicates the monthly distribution of dark material accu-drecht: Kluwer Academic Publisher, 1990, 1 - 200.mulation. The width of annual layer, sub-annual layers.Wu Xiangding, Te-ring and Climate (in Chinese), Bejjing: Me-and the accumulation rate can be calculated using theteorology Press, 1990, 1- -369.layer identification technique if proper threshold valuesB. Shao Xuemei, Wu Xiangding, Chronology of Hasha tree ring,are input respectively.Acta Geographica Sinica (in Chinese with English abstract), 1994,49(2): 174.This note ilustrates the image analysis principle for，Liu, Y., Leavit, S. W., Wu, X. D. et al., Stable carbon isotope inenvironment records. Although the calculation procedurestree rings from Huangling, China and climatic variation, Scienceof most popular parameters are given, and it may be ex-in China, Ser. B, 1996, 39(2): 152.panded to do more parameter calculating or to achieve0. Tan, M.. Qin, X. G.. Shen, L. M. et al, Bioptical microcycles ofparticular tasks according to specific needs of palaeoenvi-laminated speleothems from China and their chronological sig-nificance, Chinese Sc ience Bulletin, 1999, 44(17): 1604.ronmental investigation, such as the parameter correlation11. Gascoyne, M., Palacoclimate determination from cave calcite de-between different radii of tree-ring, investigation on dif-posits, Quaternary Science Reviews, 1992, 11: 609.ferent axes of coral sequence. Although only typical ap-2.Shenn, E. A, Coral growth rate, an environmental indicator, J.plications of the technique are given in the later part of thePaleontol, 1966, 40: 233.l3. Highsmith, R. C, Coral growth rates and environmental control ofnote, we believe that it works successfully on many otherdensity banding, J. Exp. Mar. Biol. Ecol, 1979, 37: 105.climate records. Most records can be investigated using14. Schneider, R. C, Smith, S. W.,. Skeletal Sr content and density inthe method if their color contains valuable environmentalPorites spp. In relation to environmental factors, Mar. Biol, 1982,informants, but analysis procedure and the environmental66: 121.interpretation may be different from one record type to the15. Qi Wentun, Scleractinian Coral (in Chinese), Bejing: SciencePress, 1989, 430 - -442.other. For example, when images of drill core section of16. Gladfelter, E. H., Skeletal development in acropora cervicormis: .I1marine and lucstrine sequences are taken using a digitalA comparison of monthly rates of linear extension and calciumcamera, it can be analyzed using the image techniquecarbonate accretion measured over a year, Coral Reefs, 1984, 3:introduced here instead of manual measuring using specialequipments. It is also possible to apply the image tech-17. Knutson, D. W., Buddemeier, R. W., Smith, S. V, Coral chro-nometers: seasonal growth bands in reef coral, Science, 1972, 177:nique to the loess paleosol sequence using complied fieldphotos as the image. For large-scale images, the color8. Bak, R. P. M., Available light and other factors influencingcalibration work must be done between different parts ofgrowth in stony corals through the year in Curacao (ed. Cameron,the image because the entire images are composed 0A), Proc. 2nd Int. Coral Reef Symp, Brisbane: Great Barrier Reefmany sub-images. On this point, the image technique is19.Comitee, 1974,229一233.Wellington, G. M, Glyn, P. W., Environmental influences onparticularly suitable for high-resolution records becauseskeletal banding in (Panama) corals,, Coral Reefs, 1983, 1: 215.their images are generally taken on one process with the20. Marshal, A. T, Calcification in hermatypic and ahermatypicconsistent color background over the entire image.corals, Science, 1996, 271:637.21. Well, J. W.. Coral growth and geochronometry, Nature, 1963, 197:Acknowledgements We would like to present our special thanks to Dr94Shi Qi for his help in the sample collection. We also thank Porfs. Pen22. Buddemeier, R. W, Kinzie, .I.. R. S., Coral growth, Oceano-Zichen, Nei Baofu, Yu Kefu, Shao Xuemei and Dr. Cai Yanjun for theirgraphical Marine Biology Annual Review, 1976, 14: 183.valuable discussion on coral, tree-ring or staglamite accumulation23. Macintyre, T G., Smith, S. V., X- radiographic studies of skeletalmechanism respectively. This work was supported by the Knowledgedevelopment in coral colonies (ed. Cameron, A.), Proc. 2nd Int.Innovation Project of South China Insitute of Oceanology (Grant No.Coral Reef Symp., Brisbane: Great Barrier Re ef Committee, 1974,5210607), Knowledge Innovation Project of the Chinese Academy of2: 277.Sciences (Grant No. KZCX2-108) and the National Natural Science24. Shen, G. T, Boyle, E. A, Lead in corals: reconstruction of his-Foundation of China (Grant No.499756015).torical fluxes to the surface ocean, Earth Planet Sci. Lett., 1987,References82: 289.Sheppard, P. R., Graumlich, L. J.. Conkey, L. E, Reflected-light25. Brown, B. E., Tudhope, A W., Tissier, M. D. et al, A novelimage analysis of conifer rings for reconstructing climate, TheHolocene, 1996, 6(1): 62.26.中国煤化工。concentration in the tissue and2. Qin Xiaoguang, Liu Tungsheng, Tan Ming et al, Grey character-istics of microbanding of stalagmite in Shihua Cave, Beijing andMYHCNMH Gularis ata Veneyullan reaf (ed.JIral Reef symposium, Panama:its climatic signification(. )一- the study of the microstructureof microbanding, Science in China, Series D, 1998, 41: 151.Smithsonian Tropical Research Institute, 1997, 1847 -1850.3. Chalker, B., Barnes, D., Isdale, P., Calibration of X-ray densi-(Received July 1, 2002)tometry for the measurement of coral skeletal density, Coral Reefs,Chinese S府亦数letin Vol. 47 No. 23 December 20021963中国煤化工MHCNMHG中国煤化工MHCNMHG