Spatiotemporal analysis of soil moisture observations within a Tibetan mesoscale area and its implication to regional soil moisture measurements Long Zhao a,b,c,⇑ , Kun Yang a , Jun Qin a , Yingying Chen a , Wenjun Tang a , Carsten Montzka c , Hui Wu a,b , Changgui Lin a,b , Menglei Han a,b , Harry Vereecken c a Key Laboratory of Tibetan Environment Changes and Land Surface Processes, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing 100101, China b University of Chinese Academy of Sciences, Beijing 100049, China c Research Centre Jülich, Agrosphere Institute (IBG 3), Leo-Brandt-Strasse, 52425 Jülich, Germany article info Article history: Received 14 August 2012 Received in revised form 17 December 2012 Accepted 21 December 2012 Available online 7 January 2013 This manuscript was handled by Konstantine P. Georgakakos, Editor-in-Chief Keywords: Tibetan plateau Soil moisture network Number of required sites Most representative site Optimal combination of sites summary A mesoscale Tibetan Plateau Soil Moisture/Temperature Monitoring Network (SMTMN) has been estab- lished to study large-scale soil–vegetation–atmosphere interactions and to validate satellite soil moisture products. Soil moisture at four layers (0–5 cm, 10 cm, 20 cm, and 40 cm) of 30 sites was monitored since July, 2010. This paper firstly introduces the network and then presents preliminary spatiotemporal anal- yses based on the in situ soil moisture measurements in SMTMN. Three temporal scales (half-hourly, daily, and 10-days) and three time periods corresponding to typical soil wetness conditions, including frozen soil in winter times, are discussed. Primary findings are: (a) generally 13 randomly distributed sites in the study domain are required (i.e. number of required sites) to estimate areal mean soil moisture with correlation coefficient P0.99 and root mean square difference 60.02 m 3 m 3 . This provides guid- ance for future soil moisture network establishment in similar regions; (b) both number of required sites and the most representative site are insensitive to temporal scales while conversely sensitive to soil wet- ness conditions; (c) the combination of a few optimally-selected sites can give more robust estimate of areal mean soil moisture than a single site does because the former contains more information on spatial heterogeneity. These findings can provide not only a practical compromise between maintenance cost and risk on reliability for an existing soil moisture network, but also insights for soil moisture upscaling studies and satellite soil moisture products validations. Ó 2013 Elsevier B.V. All rights reserved. 1. Introduction Near-surface soil moisture plays an important role in the terres- trial water cycle and land–atmosphere interactions (Betts et al., 1996; Entekhabi et al., 1992; Montzka et al., 2011), especially in semiarid regions where strong coupling between surface layer soil moisture and precipitation occurs (Koster et al., 2004). When ap- plied at global scale, soil moisture can be obtained through land surface model (Entin et al., 1999; Henderson-Sellers, 1996), remote sensing (Bartalis et al., 2007; Kerr et al., 2001; Njoku et al., 2003), or land data assimilation (Reichle et al., 2004; Rodell et al., 2004) with different levels of accuracy and reliability. All these estimates need validation against in situ measurements. For decades, a num- ber of soil moisture monitoring networks have been established worldwide (Dorigo et al., 2011) to obtain the ground truth. In spite of this, gaps of spatial representativeness between a single site and remote sensing pixel or a land surface model grid still exist and thus require the upscaling of ground-based soil moisture observa- tions (Crow et al., 2012). Prior to the upscaling, a basic understand- ing on how the spatiotemporal variability of surface layer soil water content influences the estimation of areal mean soil mois- ture (AMSM) is necessary. There are already a lot of related studies, and most of them focused on number of required sites (NRS) (Fam- iglietti et al., 2008; Hupet and Vanclooster, 2002) and the most representative site (MRS) (Brocca et al., 2010; De Lannoy et al., 2007) that may determine the AMSM. The NRS stands for the minimum number of sites required to estimate AMSM within a predetermined accuracy, given that the sites are randomly but relatively uniformly distributed in space. Generally, a consensus is that for a certain experiment area with uniformly distributed sites, the more sampled sites are included, the more accurate AMSM estimate can be achieved (Famiglietti et al., 1999), but the cost to install and maintain more sites will in- crease as well. On the other hand, as the coefficient of variation (CV) and standard deviation (std) of soil moisture measurements increase with the spatial scale (Crow et al., 2012), a larger number of sites are required for a wider area (Reynolds, 1974). For a 0022-1694/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jhydrol.2012.12.033 ⇑ Corresponding author. Address: Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Building 3, Courtyard 16, Lin Cui Road, Chaoyang District, Beijing 100101, China. Tel.: +86 10 84097094; fax: +86 10 84097079. E-mail address: zhaolong@itpcas.ac.cn (L. Zhao). Journal of Hydrology 482 (2013) 92–104 Contents lists available at SciVerse ScienceDirect Journal of Hydrology journal homepage: www.elsevier.com/locate/jhydrol