用数值模拟方法研究第四纪古气候──以LLN二维模式为例

SIMULATING QUATERNARY PALEOCLIMATE WITH THE LLN 2─D MODEL

本文着重介绍ＬＬＮ模式对中晚第四纪大陆冰量变迁的模拟研究。由于天文学研究已经推算出大气层顶部太阳辐射过去和将来的长期变化，因而对本文的ＬＬＮ二维模式而言，能否有效地进行古气候模拟在很大程度上取决于对古ＣＯ２浓度的重建。随着对第四纪ＣＯ２历史研究的不断深入，ＬＬＮ模式对古气候长期变迁的模拟已从最初的末次冰期－间冰期延伸到０．６ＭａＢ．Ｐ．。模拟的大陆冰量变化与地质记录对比良好，二者在与地球轨道要素有关的天文频率上的波动具有很高的相关性。

To test the Milankovitch climate theory, a two dimensional climate model was designed at Louvain-la-Neuve[11]. It is a latitude-altitude model.In each latitudinal belt (5 degrees), the surface is divided into 5 oceanic or continental surfaCe types (oceanic surfaces free of ice or covered by ice, continental surfaces with or without snow cover, and ice sheet). The atmospheric dynandcs is represented by a zonally averaged quasi-geostrophic model; The atmosphere interacts with the other components of the climate system through vertical fluxes of momentums, heat and water vapor.The model explicitly incorporates detailed radiative transfer, surface energy balances, and snow and sea-ice budgets. The surface albedo changes resulted from the 'snow aging' process and the taiga /tundra boundary shifts are also included. The vertical profile of the upper ocean temperature is computed by an integral mixed-layer model. The LLN 2-D model uses van4tions of insolation and atmospheric CO2concentration as climatic forcings, and the integral bine step is 3 days, except that a 1-day time step is used for the ocean mixed layer and sea-ice. The simulation of present climate shows that the model is able to reproduce the main characteristics of the general circulation. The seasonal cycles of the oceanic mixed-layer depth,the sea-ice extent and the snow cover are also well reproduced.The first experiments simulating the paleoclimate are from 0.122 Ma B. P.to the present[12]. The insolation computation follows Berger[14], and the atmospheric CO2.variations are from the Vostok ice core record[16]. The simulated northern hemisphere ice volume is added to the Antarctic ice volume changes so that a global ice volume variation is obtained over the last glacial-interglacial cycle. Comparison between this result and the deep-sea oxygen isotope record reveals general similarity be tween them except for the high frequency variations at time scales shorter than 5000 years.To test whether the LLN 2-D model could simulate climate change over multiple glacial-interglacial cycles, an experiment is performed using the astronomically driven insolation and CO2 concentrations adapted from Shackleton et al.[17] over the last 0.2Ma[22]. The simulation results correlate well with the SPECMAP oxygen isotope record[23], and a spectral analysis of both time series gives similar spectral characteristics,i. e., markable periods around 100, 41, 23 and 19 thousand years, signal of the 100000-year period being the strongest. The difference between the CO2of Shackleton et al. and the Vostok does not cause significantly different simulation results, implying the dominance of insolation over the simulated ice volume.Further attempts are made to simulate climate in the farther past. Here the ma- jor difficulty is that so far the reliable geological records of atmospheric CO2cover only the last 0.22 Ma.Recent study of Li et al. (1995) documented the coherent relationship between the Vostok CO2 and SPECMAP δO:y =214.53-25.66x+ 5.95x2+ 8.78x3-1.26x4- 2. 17x5where x is for SPECMAP δ18O[23] at time t, in standard deviation; and y is for Vostok CO2[18] at a time 4 500 years earlier than t, in ppmv。With this equation we calculate the atmospheric CO2 concentrations over the past 0.6 Ma. Incorporating the results to the LLN 2-D model to simulate the northern hemisphere ice volume variation over this period. The simulation reconstructs the main characteristics of the middle and late Pleistocene climate recorded in the marine sediments, and close to the simulation result of Gallee et al[22] within the recent 0.2 Ma,suggesting that our statistically-built atmospheric CO2 series could approximate the real CO2 values. The major periods of the simulated ice volume changes are 100, 41,23 and 19 thousand years, corresponding to the changes in parameters of theEarth's orbit and rotation (eccentricity, obliqinty and precession, respectively). Cross spectral analysis reveals that the simulated ice volume highly correlates with the SPECMAP δ18 O record at the frequency bands associated wi