Jump to Main Content
A comparison of spatial interpolation methods for soil temperature over a complex topographical region
- Wu, Wei, Tang, Xiao-Ping, Ma, Xue-Qing, Liu, Hong-Bin
- Theoretical and applied climatology 2016 v.125 no.3-4 pp. 657-667
- autumn, climate change, data collection, databases, ecosystems, latitude, longitude, models, mountains, seasonal variation, soil temperature, spatial data, China
- Soil temperature variability data provide valuable information on understanding land-surface ecosystem processes and climate change. This study developed and analyzed a spatial dataset of monthly mean soil temperature at a depth of 10 cm over a complex topographical region in southwestern China. The records were measured at 83 stations during the period of 1961–2000. Nine approaches were compared for interpolating soil temperature. The accuracy indicators were root mean square error (RMSE), modelling efficiency (ME), and coefficient of residual mass (CRM). The results indicated that thin plate spline with latitude, longitude, and elevation gave the best performance with RMSE varying between 0.425 and 0.592 °C, ME between 0.895 and 0.947, and CRM between −0.007 and 0.001. A spatial database was developed based on the best model. The dataset showed that larger seasonal changes of soil temperature were from autumn to winter over the region. The northern and eastern areas with hilly and low-middle mountains experienced larger seasonal changes.