Spatial Recovering of Agricultural Values from Aggregate Information: Sequential Downscaling Methods

Fischer, G., Ermolieva, T., Ermoliev, Y., & van Velthuizen, H.T. (2006). Spatial Recovering of Agricultural Values from Aggregate Information: Sequential Downscaling Methods. IIASA Research Report (Reprint). IIASA, Laxenburg, Austria: RP-06-009. Reprinted from International Journal of Knowledge and Systems Sciences (IKSS), 3(1):1-6 [2006].

Preview

Text RP-06-009.pdf Download (127kB) | Preview

Abstract

In this paper, we propose a downscaling procedure that provides a basis for recovery and estimation of incomplete, aggregate, unknown or indirectly measurable variables. It makes maximum use of information and dependencies on various levels relying on the cross-entropy maximization principle. We show that the maximum entropy principle can be viewed as the extension of the maximum likelihood principle. In this sense, the convergence of the proposed downscaling methods to solutions maximizing an entropy function can be considered as an analog of the asymptotic consistency analysis in traditional statistical estimation theory.

The main motivation for the development of the procedure has been a practical example of spatial estimation of agricultural production values. We briefly discuss the main challenges related to the choice of priors (location specific information) and their inherent uncertainties that to large extent determine the success of the downscaled results.

Actions (login required)

View Item