APPLICATION OF MARKOV CHAIN METHOD TO ESTABLISHMENT SEDIMENTARY FACIES MODEL

Observed the cores from 11 wells and the loggings from approximate 300 wells in 3rd Member of the Lower Tertiary Shahejie Formation in the west Huimin seg, and analyzed sedimentary facies, Linpan Delta and the Shanghe deepwater turbidite fan and other sedimentary environments were determined. Linpan Delta and the Shanghe turbidite fan have a lot of different characteristics in lithology, sedimentary structure, grain size, logging, and paleontological assemblage etc. The lobate Linpan Delta Paralleled the axis direction of Huimin seg, 35 kilometers in length, over 700 meters in accumulating thickness. The delta displays a reverse cycle on the whole vertical section. The reverse rhythm of the predelta and delta front consisted of grey mudstone and light grey siltstone, fine grained sandstone, about 8 meters in thickness, and appeared frequently; the normal rhythm of the delta plain consisted of light grey, and greyish green siltstone, fine grained sandstone, medium grained sandstone and brown mudstone, carbonaceous shale, 1-4 meters in thickness. The normal rhythm often overlaies the reverse one, constituting a complete delta composite rhythm. The maturity of sandstone composition is lower, the content of quartz 50-60%, feldspar 20-30%, rock fragment 10-20%, clay 5-10%. The maturity of its texture is higher, moderate-well sorted and grain supported texture. The grain size data indicate the feature of typical drawning-current. The probability graphs of the delta are two-lines which consits of suspension and saltation populations, CM diagram consists of segment QRS. There are many types of sedimentary structures, such as cross bedding, parallel stratification, and vegetable foot etc. The vertical sequence of the sedimentary stuctures reflects the delta porgradation towards lake centre, paleontological assemblages, SP curves and seismic sections all show the existence of the Linpan delta. The Shanghe deep-water turbidite fan is located in the front of Linpan Delta, irregular oval in shape, smaller in size, the area is about 10-20 km2. The frequent interbeded light grey, greyish brown muddy siltstone, siltstone, fine grained sandstone, and deep grey mudstone constituted a lot of normal rhythms, 0.5-5.5 meters in thickness. The maturity of the sandstone composition is relatively low, the content of quartz 40-60%, feldspar 15-30%, rock fragment 5-25%, clay 10-20%, having mud supported, pore-matrix cementation. The information of grain size shows the characteristic turbidite sedimentation, probability graphs are single line, CM diagram appears segment QR paralleled to C = M base line. The complete incomplete Bouma seqences and many kinds of base prints were developed. The deep and shallow fossils were associated, their types were monotonous and a few in quantity. There are many typical electric loggings and seismic responses of the Shanghe turbidite fan. In order to establish a local facies model quickly and accurately, and play a role in facies studying, to eliminate some artificial influnce, and to favor the dialogue between geologists, in this paper authors used the Markov chain to determine vertical relationship of the sedimentary sequence of the Linpan delta and the Shanghe deep-water turbidite fan and to establish the local facies models. Practice has indicated that the Markov chain is an effective method to establish local facies model, and it is suitable for rhythmical complex stratigraphic sequence. The steps of Markov chain established facies modle are following: 1. Observe stratigraphic section carefully, based on a varity of facies marks, determine facies with different characteristics and their relationship plotted relation diagram of facies change; 2. Count the whole number of facies changes and the number of facies changes from one into another, establish the matrix of facies change number, determine facies change probability of sequence; observed 3. Fix the facies change probability of random sequence; 4. Determine the diffe- rence of facies change probability between the observed and random sequences,