Numerical simulations and observations are complementary and combining them efficiently has been the topic of many recent studies. This session will focus on two specific aspects: data assimilation and proxy system modeling, and on the applications on these techniques to constrain past climate and environmental changes. Proxy system models have been developed to explicitly simulate the variable measured in an archive (for example, tree ring width, pollen assemblage or d18O in corals) that serves as a proxy for environmental conditions. When driven by outputs from a climate model, this offers the possibility to perform model-data comparison directly, thus avoiding the often ill-conditioned transformation of the observed quantity into a variable simulated by the model (such as the temperature or precipitation). Data assimilation uses observations and simulations to produce estimates based on both sources of information, given uncertainties in each. If certain key assumptions of the method are met, this approach allows estimation of environmental parameters for regions and times which lack observations, potentially leading to a better mechanistic understanding. Ideally, proxy system models should be an essential element of any data assimilation procedure. Nevertheless, biases in climate models and the limitations of the proxy system models themselves may restrict the advantages of including proxy system models in data assimilation compared to simpler choices based on statistical methods. In this session, we welcome contributions describing and applying proxy system models as well as experiments using data assimilation to reconstruct past changes or understand the origin of those changes, on all spatial and temporal scales. Contributions integrating different data sources, comparing various models and methods or integrating proxy system models in a data assimilation procedure are particularly welcome.