INFLUENCE OF ZR ON ESTIMATING PALEOPRECIPITATION AND PALEOTEMPERATURE USING A NEW PALEOSOL GEOCHEMISTRY MODEL

Zirconium oxide (ZrO₂) concentrations in the soil reflect parent rock composition and they share similar chemical and physical characteristics with titanium. Yet, the enrichment of ZrO₂ and the extent of mobility has been shown to vary by climate. The majority of ZrO₂ resides from the mineral Zircon and can be a component of dust, so using ZrO₂ as an immobile tracer has a downside. A previously developed paleosol-paleoclimate model (PPM_1.0) was created using ZrO₂ and 10 other elements from B horizons to predict mean annual temperature (MAT) and mean annual precipitation (MAP). The PPM_1.0 model differs from similar models due to its larger range of MAP and MAT values. A key importance to the research is that by removing and testing the model without ZrO₂, it can have access to model data sets that currently have no analyses of ZrO₂. The influence of ZrO₂ is evaluated in that model by observing actual and estimated MAP and MAT values for two different scenarios: (1) with ZrO₂ and (2) without ZrO₂, to further understand its influence on the PPM_1.0.

Observed vs. estimated plots, and residual plots were created to display MAT with ZrO₂, MAT without ZrO₂, MAP with ZrO₂, and MAP without ZrO₂. The scatter plot regression lines fit as follows: without ZrO₂ (MAT R²=0.58, MAP R²=0.87) and with ZrO₂ (MAT R²=0.35, MAP R²=0.52). Towards lower MAT and MAP values, the error between observed and estimated decreases. At higher values, the model fit overestimates MAT and MAP with ZrO₂ and underestimates without ZrO₂. Residual plots indicate that without ZrO_2,they display lower residual error. Both graphs display symmetrically distributed data with outliers at higher estimated values.

It is unclear that ZrO₂ has significant influence over the MAP and MAT output; it’s also not clear whether the model works better or worse, with or without ZrO₂ based on current results. The code to compute MAP and MAT involves using different x-values once ZrO₂ was removed, so graphs with and without ZrO₂ can’t be directly compared since it’s unsure whether the oxide had a direct relation to significant change. ZrO₂ influence on the PPM_1.0. model can provide more insight as to how to improve the current edition. In addition, the ability to remove an oxide from the model is essential to testing more data that originally doesn’t include ZrO_2.