Geostatistical estimation of tropical peat-soil volume at Bacho, Thailand: impact of spatial support size and censored information

Abstract

The peat-soil volume at the Bacho site in Thailand was estimated using indicator geostatistics. The original peat layer thickness data include two types of censored observations: the physical and measurement limits. To avoid assigning values arbitrarily to them, the indicator approach was used, in which all observations are transformed into either 0 or 1 depending upon the exceedence of any given threshold. In this study, peat-soil thickness values are estimated using blocks with different sizes (i.e., change of support). Indicator kriging is applied to construct conditional cumulative distribution functions (ccdf) defines peat layer thickness at centers of estimation blocks. From the point ccdf obtained, an optimal estimate of peat layer thickness at each block is obtained by taking the mean of the point ccdf. The ccdf value beyond the maximum threshold is modeled by the linear model as it fits well to the original cumulative distribution of data. The choice of the extrapolation model needs to be guided by available information. The total peat-soil volume is estimated simply summing up the volume at each block (optimal thickness×area of the block). The estimated volumes are almost constant when the block size less than 200 m is used, indicating that the mean of the point ccdf at the center of the block can be used as the representative value of the block. When much larger block sizes are used, the ccdf at the center of the block cannot be used to estimate the block value. In such a case, the block ccdf need to be derived using different approaches such as stochastic simulation. The estimated total peat-soil volume within the investigation boundary at the Bacho site is 2.14×10 7 m 3, when the block size smaller than 200 m are used. Using the average organic carbon content value measured at the site, the total organic carbon storage is estimated 1.22×10 9 kg C.