Fuzzy Patches on the Earth's Core-Mantle Boundary

Fuzzy Patches on the Earth’s Core-Mantle Boundary? Introduction • Seismological investigations revealed the presence of ULVZ in many locations at the base of the mantle • Siginifcant reductions in VP and Vs were invoked to match the observed waveforms: -δVp≤5-20%, δVs≤10-50% and thickness ~5-50 km • Such low-velocities were interpreted in terms of partial molten just above the CMB • Significant heterogeneity in properties within ULVZ were suggested Modeling trade-offs 1. Significant tradeoffs are recognized in ULVZ modeling, but only a relatively narrow range of models satisfying the seismological observations have been explored 2. Modeling of the SPdKS waveform allows other models to be evaluated 3. Considering the effects of perturbations in ρ and wave velocities (Vs, Vp) in modeling data Modeling trade-offs Fig. 1 summarizes results for several models that match data obtained from the 3/31/94 Fiji event Clearly, a range of ULVZ thicknesses (2-10km), density perturbations (0-60%), and velocity perturbations can be found that match data. A more comprehensive analysis of tradeoffs among ULVZ characteristics for the Fiji data set ULVZ Density Considerations The modeling results indicate the possibility of surprisingly large density increases of up to 60% • require a major variation in bulk composition but no phase transition having density change >10-15% • Can be ascribed to chemical reaction/contamination at the lowermost mantle by the core ULVZ (reductions in Vs and VP) M decrease Partial melting ? V=[M/ρ]1/2 Density increase Partial melting (δVS:δVP=3:1) ? Chemical contamination (and reaction) by the core Mutually reinforcing Partial melting is in some cases incompatible with a ULVZ containing large ρ anomalies Partial melt vs no melt condition V=[M/ρ]1/2 δlnV=(1/2)δlnM-(1/2)δlnρ and -δlnV=(1/2)δlnρ − (1/2)δlnM If δlnM<0 (caused by partial melting) then –δlnV> (1/2) δlnρ for melting else if –δlnV≤ (1/2) δlnρ (δlnM>=0) then partial melting is not possible For a sufficiently large density increase, the entire drop in velocity is due to density; the modulus does not decrease, and an increase in modulus may even be required in order to match the velocity change No melt condition –δlnV≤ (1/2)δlnρ Model solutions w/o melt correspond to conditions involving the most extreme perturbations in velocities and density, along with the small values of thickness (< 3-7km) Not all anomalies require lowermost mantle partial melting, the authors suggested that (1) A thin (~1km) “core rigidity zone” at the top of the outer core (CRZ) (2) A core-mantle transition zone (CMTZ) at the top of the outer core Can also explain the observations Comparing synthetic seismograms for CRZ with those of ULVZ that fits the Fiji data shows the strong similarity in SPdKs behavior But CRZ must have a non-zero VS, the perferred thickness is ≤ 1km. ULVZ 5km δVP -10% δVS -30% δρ=0 CRZ VP Vs ρ 1.5km 8km/s 3km/s 9.6Μg/m3 ULVZ 5km δVP -10% δVS -30% CRZ 1km VP 8km/s Vs ρ 3km/s 9.6Μg/m3 δρ=0 CMTZ Properties change from 2km pure mantle to pure core PREM Preliminary Reference Earth Model ULVZ, CRZ and CMTZ waveforms match data well, and are clearly distinguishable from the PREM waveforms CMTZ or CRZ models of thickness ~1km can produce observable waveform disortions Changing the thickness yields different separations between the arrivals of SKS and SPdKS waves, or PcP (or ScP) and precusors (or postcursors) Discussion & Conclusion • The physical interpretation of seismologically anomalous zones at the CMB are necessarily speculative and waveform modeling sucks • Both partial melting and chemical contamination may be necessary to explain the anomalous data • Fuzzy patches at the CMB may be zones of intense chemical and physical interactions between the mantle and core