Let $\{Z(\x):\x\in\cD\}$ be a stochastic process in a domain $\cD\subset\IR^d$, $d\ge1$. To apply statistical procedures, it is often necessary to decompose theprocess into several parts. The most commonly used decomposition is based on the separation according to different scales: a large-scale variation, a smooth small-scale variation, a microscale variation and a measurement error. Although this additive partitioning is of considerable utility it also has several drawbacks. In this paper, we present an analysis based on state-space representation. Let $Z(\x)=W(\x)+\epsilon(\x)$, the observation equation, and $W(\x)=\int_{\cD}k(\x,\s)W(s)ds+Y(\x)$, $\x\in\cD$, the state equation, where $k(\x,\s)$ is a sufficiently regular function, $Y(\x)$ is a second-order stationary spatial process and $\epsilon(\x)$ is a zero-mean white-noise. The state at the point $\x$ is then a weighted mean of its neighborhood states plus a spatial process. Other existing decompositions can be reconstructed by the new representation. The new model takes account of diverse shapes of trends and one does not have todecide whether the process is stationary or not. We consider estimates of the parameters of the covariogram of $Y(\x)$ based on nonlinear optimization. Finally we discuss the efficiency of the proposed method and compare the results to other commonly used models.