31 Aug 2023
 | 31 Aug 2023
Status: this preprint is open for discussion and under review for Climate of the Past (CP).

New probabilistic methods for quantitative climate reconstructions applied to palynological data from Lake Kinneret

Timon Netzel, Andrea Miebach, Thomas Litt, and Andreas Hense

Abstract. Quantitative local paleoclimate reconstructions are an important tool for gaining insights into the climate history of the Earth. The complex age–sediment–depth and proxy–climate relationships must be described in an appropriate way. Bayesian hierarchical models are a promising method for describing such structures.

In this study, we present a new age–depth transformation in a Bayesian formulation by determining the uncertainty information of depths in lake sediments at a given age. This enables data-driven smoothing of past periods, which allows for better interpretation.

Furthermore, we introduce a systematic way to establish transfer functions that map climate variables to biome distributions. This includes consideration of various machine learning algorithms for solving the classification problem of biome presence and absence, taking into account uncertainties in the proxy–climate relationship. For the models and biome distributions used, a simple feedforward neural network wins.

Based on this, we formulate a new Bayesian hierarchical model that generates local paleoclimate reconstructions. This is applied to plant-based proxy data from the lake sediment of Lake Kinneret. Here, a priori information on the recent climate in this region and data on arboreal pollen from this lake are used as boundary conditions. To solve this model, we use Markov chain Monte Carlo sampling methods. During the inference process, our new method generates taxa weights and biome climate ranges. The former shows that less weight needs to be given to Olea europaea to ensure the influence of the other taxa. In contrast, the highest weights are found in Quercus calliprinos and Amaranthaceae, resulting in appropriate flexibility under the given boundary conditions. In terms of climate ranges, the posterior probability of the Mediterranean biome reveals the greatest change, with an average boreal winter (December–February) temperature of 10 °C and an annual precipitation of 700 mm for Lake Kinneret during the Holocene. The paleoclimate reconstruction for this period shows comparatively low precipitation of about 400 mm during 9–7 and 4–2 cal ka BP. The respective temperature fluctuate much less and stays around 10 °C.

Timon Netzel et al.

Status: open (until 09 Nov 2023)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse

Timon Netzel et al.

Model code and software

Reconstruction code in R (includes the data sets) Timon Netzel

Reconstruction code in python (includes the data sets) Timon Netzel

Timon Netzel et al.


Total article views: 182 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
130 44 8 182 6 5
  • HTML: 130
  • PDF: 44
  • XML: 8
  • Total: 182
  • BibTeX: 6
  • EndNote: 5
Views and downloads (calculated since 31 Aug 2023)
Cumulative views and downloads (calculated since 31 Aug 2023)

Viewed (geographical distribution)

Total article views: 162 (including HTML, PDF, and XML) Thereof 162 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
Latest update: 03 Oct 2023
Short summary
New probabilistic methods for local quantitative palaeoclimate reconstructions are presented in a Bayesian framework and applied to plant proxy data from Lake Kinneret. We use recent climate data and arboreal pollen from the sediment of this lake as predefined boundary conditions. The result shows a climate reconstruction of the mean December–February temperature and annual precipitation with the corresponding uncertainty ranges during the Holocene in this region.