【正文】 and dense sand extending to badrock with the varied thickness of 10 to 20m. 3. 3 Soil improvement and embankment construction Dry Jet Mixing and stage construction with sand blanket have been designed to reduce the total settlement and post construction settlement . The embankment filled with residual clay from Dec. 1999. After the first stage construction of ～ m high , about 6 months were left along for soil consolidation. The settlements are observed regularly by settlement plates . During the period of the first stage consolidation , some observational settlements are found to be larger than the corresponding designed total settlement . It is necessary to repredict the behavior of the embankments based on the settlement observational results of the first stage construction , in order to modify the original design and make the reliable decision for next stage construction. 3. 4 Prediction of the final settlement The typical fill heights with time and measured ground surface settlements at the center of embankments with sand blankets are shown. It can be seen that the settlement curve drops significantly between the first stage and second stage construction. The final settlements of the second stage construction are predicted from the first stage observational data by the stage observational method. It indicates that the predicted settlements are basically consistent with the measured settlement . Table 1 also shows the bake analyzed values of undrained elastic modulus Eu and coefficient of consolidation Cv based on the first stage observational data , giving ranges of Eu / cu ratio of 50～ 100 and CV (field) /CV (lab) ratio of 6～ 12. It seems to be in the lower range of the Eu / cu ratio for Jiangsu Marine clay pared with the ratio of 70～ 253 for Bangkok clay and other clays existing in the different literature. The actual coefficient of consolidation appears one order of magnitude larger than the laboratory value , this resulting in the faster rate of measured settlement . 4. Conclusions On the basis of theoretical derivation of Asaoka method and case study of Jiangsu Marine clay , this paper presented a Stage Observational Method for settlement prediction of embankments on soft ground with stage construction. The following conclusions can be given: (1) Considering the available observational methods for ultimate settlement prediction , the Asaoka method may be successfully extended to make the settlement prediction for stage construction embankments . (2) The immediate settlement Se is verified to be equal to the intercept 0? in the Sj and Sj1 space of the Asaoka method , therefore , the undrained modulus of soft ground can be obtained from the first stage construction measurements , this contributing to the more accurate estimation of immediate settlement of the next stage construction. (3) Assuming the actual coefficient of consolidation and the thickness of the soft ground remain constant during stage construction , the shifting distances of the parallel lines with the slope of 1? is equaled to the immediate settlements , which can be calculated with the inverse modulus from the first (last) stage construction. (4) The distinct advantage of the remended method is that the values of undrained modulus and coefficient of consolidation for next stage construction are inversely analyzed from the first (last) stage construction. (5) From the case study , the value of Eu / cu ratio ranges from 50 to 100 for Jiangsu Marine clay , while the actual coefficient of consolidation is almost one order of magnitude larger than the laboratory data. Acknowledgement The research is sponsored by Doctoral Program Funds of Education Ministry of China. The support is grateful acknowledged. 軟土地基上分期施工的路堤沉降預測方法 摘要： 沉降量和沉降速率控制是軟土地基上路堤工程設計的關鍵問題，由于固結理論的局限性和參數的不確定性，理論預測的精度較低，而基于現場實測數據的觀測法則顯示出了較高的精度。 因此，人們非常渴望發(fā)展這樣一種觀測方法 —— 只要記錄足夠的