A new algorithm for the energy–water balance model to quantitatively reconstruct Holocene precipitation and vegetation: a case study from Dali Lake, North China

Abstract

Quantitatively reconstructing precipitation is an important component of paleoclimate research. The mean annual precipitation (MAP) of a closed-basin lake can be estimated by the energy–water balance model (EWBM). The EWBM can be divided into linear and nonlinear models, with determination of evaporation over water and land surface being the main components. However, the influence of vegetation changes on different models and the potential of the models to reconstruct vegetation remain unclear. Dali Lake, a closed-basin lake in the marginal region of the East Asian summer monsoon (EASM), is selected as the study region to test the feasibility of a new algorithm, i.e., the comparative method. This method is performed by comparing the reconstructed MAPs given by linear and nonlinear models for each stepwise reduced or increased forest fraction until both MAPs are approximately equal within the error range. The results indicate that the linear model is more sensitive to vegetation changes than the nonlinear model; however, both linear and nonlinear models can obtain the modern MAP correctly. Additionally, the comparative method can successfully determine the vegetation cover. Using the reported paleolake levels of Dali Lake, we calculate the Holocene MAP and vegetation of the Dali Lake basin. The results show that the early and middle Holocene MAP obtained by the comparative method is ∼10 mm higher than that obtained by the nonlinear model, while the difference in the late Holocene is not insignificant. The reconstructed vegetation is generally consistent with the pollen-indicated vegetation changes in the Dali Lake region and the monsoon marginal zone in North China. Our study suggests that both linear and nonlinear models can provide reliable MAP values for basins with known vegetation. For basins with unknown vegetation, the nonlinear model with variable parameters or the more advantageous comparative method is preferred. In cases where the basin mainly consists of two vegetation types, the comparative method enables to reconstruct the fraction of two vegetation types.