Soil moisture content retrieval based on apparent thermal inertia for Xinjiang province in China

Abstract

In the arid to semi-arid regions of north-western China, soil moisture is the main hydrological driver for vegetation growth. With the launch of the Moderate Resolution Imaging Spectroradiometer (MODIS), the local MODIS reception capacity and the strong pressure on water resources in the province, the detection and mapping of soil moisture content (SMC) has become a major issue for the regional water management authorities of the province.

In this article, we apply the apparent thermal inertia approach to quantify SMC in the soils of the province of Xinjiang using locally received MODIS data. We report on SMC mapping for the entire province for the year 2005.

For the estimation, diurnal land surface temperature (LST (T_σ)), LST difference (ΔT_σ) and broadband albedo (α₀) were applied to determine the space–time variability of SMC. The retrieval of SMC was based on the rationale that high apparent thermal inertia (I_τ) values correspond to high SMCs and low I_τ values correspond to the minimal ones. To enable the application of the technique, a new classification of soil texture was established based on an existing Chinese soil type classification.

Typically only topsoil surface moisture content is retrieved with thermal remote sensing (RS). However, SMC retrieval for a 1 m soil profile was performed by applying a semi-empirical modelling approach. The model uses a two-layer water balance equation, and its SMC (θ_g) input is based on its linear relationship with the soil moisture saturation index (θ_si) at time t.

For validation purposes, the automatic weather station and time domain reflectometry (TDR) monitoring network included eight sites in the province, including the Mosuowan and Tianshan snow sites and the Turpan, Bayangburk, Kuerle, Yinsu, Alagan and Yiganbujima TDR sites for which data for the year of 2005 were acquired by the Xinjiang Meteorological Bureau (XMB), the Tarim Management Bureau (TMB) and the Xinjiang Institute for Ecology and Geography (XIEG).

When time series of SMC determined by using I_τ are compared with the measurements at the different validation sites, regression curve slopes of the validation relationships vary between 0.499 and 0.922. The R² values vary between 0.25 and 0.83. The minimum and maximum root mean square errors (RMSEs) are 0.001 and 0.028, respectively. Results suggest that apparent thermal inertia application is quite suitable for θ_g retrieval of a 1 m soil moisture profile in an arid to semi-arid region. The Aqua MODIS 10-day mean soil moisture product is proven to deliver quantitatively correct SMC imagery representing seasonal changes quite realistically.

Acknowledgement

The authors thank the Ministry of Science and Technology (MOST) of China as well as the Belgian Science Policy Office (BELSPO) for the financial support received to execute the work described in this article.