Quartz contents derived from particle density measurements improve the accuracy of soil thermal conductivity estimates

Abstract

Quartz content (fquartz) is a key value used to estimate soil thermal conductivity (λ) in land surface models. Due to the difficulties in measuring fquartz directly, many studies use sand content (fsand) to approximate fquartz in λ models, causing large uncertainties in λ estimates. The existing methods for determining fquartz are quite limited and complicated, and it is still desired to derive fquartz values with simple parameters. Here, we present an empirical equation to estimate fquartz values from soil particle density (ρs). The empirical equation was developed from two published datasets including laboratory measurements on 56 soils, and its performance was evaluated by comparing measured λ values to estimated λ values. Four models were used to estimate λ using inputs of fsand, actual measures of fquartz, and the new empirical equation estimates of fquartz. The root mean square errors, RMSEs, of modeled λ values (based on the empirical equation estimated fquartz) were less than 0.27 W m−1 K−1, which were similar to the RMSEs (< 0.26 W m−1 K−1) when actual fquartz values were used as model inputs. The new empirical equation estimates of fquartz led to much lower RMSEs in modeled λ values than the RMSEs (< 0.39 W m−1 K−1) obtained when fsand values were used to represent quartz content. The new empirical equation provides a simple and easy approach to estimate fquartz as an input to λ models.