JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 100, NO. D1, PAGES 1093-1101, JANUARY 20, 1995 Reconciling Late Ordovician (440 Ma) glaciation with very high (14X) CO2 levels ThomasJ. Crowley and StevenK. Baum Department of Oceanography, Texas A&M University, College Station Abstract. Geochemical data andmodels suggest a positive correlation between carbon dioxide changes andclimate during thelast540 m.y. Themost dramatic exception to thiscorrelation invc•lve• _the I .•toC•rdovician (440 Ma) glaciation, whichoccurred at a time whenCO2levels may have been much greater than present (14-16X?). Since decreased solar luminosity at that timeonly partially offset increased radiative forcing fromCO2, some other factor needs to beconsidered to explain theglaciation. Priorworkwith energy balance models (EBMs) suggested thattheunique geographic configuration of Gondwanaland at that timemayhave resulted in a small area of parameter space permitting permanent snow cover and higher CO2 levels. However, thecrude snow and sea iceparameterizations in theEBM left these conclusions open to further scrutiny. Herein we present results fromfourexperiments with theGENESISgeneral circulation model with CO2levels14X greater than present, solar luminosity reduced 4.5%, andan orbital configuration setfor minimumsummer insolation receipt. We examined the effects of different combinations of ocean heat transport andtopography on high-latitude snow cover on Gondwanaland. For the no- elevation simulations we failedto simulate permanent summer snow cover. However, for the slightly elevated topography cases (300-500m), permanent summer snow cover occurs where geological data indicate theOrdovician ice sheet was present. These results support thehypothesis based on EBM studies. Further results indicate thatalthough average runoffper grid point increases substantially for the Ordovician runs, thedecreased landarea results in global runoff 10- 30% less than present, with largest runoff reductions for flat topography. Thisresponse has implications for CO2-runoff/weathering parameterizations in geochemical models.Finally, simulated tropical sea surface temperatures (SSTs) arethesame or onlymarginally warmer than present. Thisresult is consistent with evidence fromother warmtimeintervals indicating small changes in tropicalSSTs during time of high CO2. 1. Introduction One of the most significant developments in geology during the past 10 years hasbeenemergence of the paradigm suggesting that variations in CO2 have played an important role in the evolution of past climates. This relationship applies to both Pleistocene glacial cycles [Barnola et al., 1987] and pre-Pleistocene time periods of altered land-sea distribution. The pre-Pleistocene conclusions represent a convergence of results from climate[Barron and Washington, 1985] and geochemical [Berner, 1991] models and proxy data for higher CO2 levels [Berger and Spitzy, 1988; Cerling, 1991' Freeman and Hayes, 1992]. There is generally good agreement between •' ..... '•"'"':""' '-""'•"'• ""'• •'•" •' indices [Berner, 1992], at least in terms of the sign of the response. The geochemical modelcalculations alsoagree well [Crowley and Baum, 1991] with the distribution of glacial deposits (Figure 1) through the Phanerozoic (last540 m.y.). Despite the gratifying convergence of evidence from a number of different approaches, a troubling exceptionto the CO2-climate correlation involves evidence for Late Ordovician (440 Ma, million years ago) glaciationin North Africa and This paperis not subject to U.S. copyright. Published in 1995 by the AmericanGeophysical Union Paper number 94JD02521. adjacentregions (Figure 1). At that time these areas were part of the Gondwanan supercontinent and in high southern latitudes(Figure 2). The most direct evidence for glaciationis based on tillites and striated terrain typical of glacial deposits [Hambrey and Harland, 1985; see also Crowley and North, 1991, Figure 11.6], although the initial suggestion of glaciation was based on equatorwarddisplacements of marine biota [Spjeldnaes, 1961]. The estimatedarea of the deposits, if they represented a contiguous ice sheet, would be of the order of 6-10 X 106 km 2 [Hambrey, 1985;Crowleyand Baum, 1991], about the same size as the present East Antarctic Ice Sheet [cf. Long, 1993]. Analysis of nearshore marine sediments indicate a sea level fall of 50-100 m for this event [Brenchley and Newell, 1980]. The areal estimates of ice cover would be consistent with the lower estimate for magnitude of sea level fall [Crowley and Baum, 1991]. The glaciation also coincides with the second largest extinction event in the Phanerozoic, when an estimated 85% of all species became extinct [e.g., Brenchley, 1984; Jablonski, 1991]. In the context of the CO2 paradigm the problem with Late Ordovician glaciation involves geochemical model calculationsthat CO2 levels may have been very high for the Late Ordovician [Berner, 1991], on the order of 14X present values. This calculation agrees with CO2 estimates from proxy data of a •-16X CO2 increase [Yapp and Poths, 1992]. These values seem incompatible with glaciation of any type. Although 813C data suggest there may have been a "brief" 1093