ИСТИНА

Mapping the terrain in small scales requires the appropriately generalized elevation data. Existing coarse-resolution digital elevation models such as GMTED2010, GEBCO, GTOPO30, ETOPO1, ETOPO2v2 and ETOPO5 have a wide range of spatial resolutions and can be used for mapping purposes in scales of 1:1 000 000 and smaller. However, the appropriate spatial resolution is not enough to obtain the cartographic relief representation of high quality. These models were obtained using the various sources and the simple resampling as a main generalization method. As a consequence, in resolutions of 30 arc seconds and smaller these models have excessive terrain details. Every pixel represents just the elevation sampled at its position, often from different subpixel terrain features, which results in lower spatial autocorrelation of neighbouring elevations. As a consequence the resulting hypsometric and shading images are not morphologically sound. For example, the shape of the neighbouring contours may not be coherent, while terrain shading includes too many details in relation to spatial resolution. The images must be simplified and amplified to produce the desired readability and visual effect. To solve this problem effectively we must obtain the appropriately generalized digital elevation models for various small scales. We used the methodology presented by Samsonov (2011) to obtain the morphologically generalized raster DEMs with worldwide land coverage (the sea is not included). This approach combines previously developed methods and builds the triangulation of main structural lines (streams and watersheds) to reconstruct the terrain surface in a simplified form. Valleys and ridges are widened in the resulting surface to amplify the readability of terrain features. As this approach is computationally intensive, a new technological scheme was developed that dices the source DEM into overlapping fragments and processes them in parallel. GMTED2010 30 arc seconds model was used as a source dataset. Resulting models were called HYPSO inspired by hypsometric approach often used in small-scale terrain mapping. HYPSO1, HYPSO2, HYPSO5, HYPSO10 models with spatial resolutions of 1, 2, 5 and 10 arc seconds were derived. For derivation of every model the optimal values of source parameters such as stream length and flow accumulation were selected experimentally. Resulting models can be used for land terrain representation in scales smaller than 1:1 000 000 using the hypsometric, hillshading and other approaches. It is planned to make them available for free. We also plan to include the sea surface in the next version of the model, which will be possible with addition of GEBCO model as a source dataset covering the oceanic areas. References: Samsonov T. Multiscale hypsometric mapping // Advances in Cartography and GIScience, Vol. 1. — Lecture Notes in Geoinformation and Cartography. — Heidelberg, Germany: Heidelberg, Germany, 2011. — P. 497–520.