The Forest as Sensor, a Case Study in Data Model Intercomparison Forimproved Paleoclimatic Data Assimilation Exercises

The paleoclimatic data assimilation problem requires data (or proxy system) models that map climate models or simulations to the observations. But what level of complexity, generality and formulation is appropriate in the data models, given constraints imposed by limited observations, parameter estimation, and structural uncertainties? To explore these questions, we develop experimental design and narrowly focus on the prediction of forest measures responding to atmospheric moisture, temperature and carbon dioxide levels over the past century, using as case study simulation of tree ring width and analog measures at Sodankyla, Finland (67.4^oN, 26.7^oE, 179m). By application of models with and without carbon allocation processes and the potential for CO₂ fertilization effects, and varying levels of complexity and structural independence, we quantify uncertainties and assess relative merits with respect to their use in constraining paleoclimate data assimilation products. We call for participation in the development of so-called "data model intercomparison projects" as part of PAGES/DAPS (Data Assimilation and Proxy System) modeling activities.