|the horizontal contour of a SU was limited by the vertical sections going along the borders of the excavation ground, or by a contour line/multiple lines with a fixed elevation mark (fig. 7). The “lower” surface of a SU coincides with the “upper” surface of the consequent stratification unit/units. The rather different strategy was used when stratification units were sliced by horizontal spits with an elevation interval set to 10 cm. In this case, the SU limits were traced on a planar surface of a spit and marked by one (SU completely inside the excavation ground) or two (SU extends the excavation ground) contour lines (fig. 8). In both cases, if a SU border exceeded the borders of the excavation ground, vertical sections, which stratification was thoroughly recorded once the excavation was finished, provided the wireframe 3-D “upper” and “lower” contours of each stratification unit (fig. 9).
Due to budget limitations, the data collection was limited to manual and optical theodolite measurements. When it was possible, the stratification record, elevation grid points, contour lines, and all other graphical data were directly input or copied in AutoCAD-based virtual 3-D excavation ground environment. Though, weather conditions, battery life limitations, and other unpredictable circumstances made direct on-the-fly computer record impossible. The voids in digital record were compensated with traditionally taken paper-form plans, drawings and sketches, which were scanned and then manually vectorized in AutoCAD module – CAD Overlay (fig. 10). Thus, raw data of stratification units’ measurements were collected. The next step was to create the most precise 3-D virtual solids of SU from the obtained elevation points and contour lines data array. The extrapolation principle of digital relief reconstruction was assumed to be the most appropriate. Stratification unit’s surface is very similar, not to say - exact, to a relief surface. We strongly believe, that elevation points extrapolation in this case is not less precise and accurate then those used in digital cartography. The most common technique, which was chosen, is geometrical extrapolation. The initial points and contour lines for each SU were integrated in a triangular irregulated network (TIN) (fig. 11), which was followed by creation of a regular extrapolation grid with 20x20 cm cell density (fig. 12). Merged together, the extrapolated surfaces formed the final shape of stratification units’ virtual 3-D solids (fig. 13-14).
3.2.2. Artificial and natural remains record
Natural stone scattered chaotically or organized geometrically in the cultural layers; chopped, cut or broken wood of both natural and artificial origin; bones were recovered in hundreds in the process of excavation. Some of them appeared in the cultural layers accidentally, being a natural deposition, the rest were intended by inhabitants of Gnezdovo as parts of different structures and hearths. Large amount of these remains along with very slow speed of manual and optical theodolite measurements made it impossible to transmit their shape to the 3-D modeled AutoCAD environment identically. A series of approximations had to be accepted.
All unearthed stone was of irregular natural shape; its dimensions did not exceed 20x20x20 cm. Therefore, after its’ precise position and dimensions were marked in the virtual excavation ground, the sample stone library, consisting of about two dozens of common stone shapes, was used to find the most similar shape to fit the original (fig. 15).
This kind of approximation strategy, appropriate for stone, could not satisfy the wooden remains and bones. No sample would fit their individual unique shape and represent the significance of small details of appearance. 3-D solids of wooden and bone remains were substituted by their 2D contour outlines taken from the top view and several thickness measurements as it is done in a traditional paper-form excavation record. In some cases, multiple contour outlines could be recorded to give a rather rough idea of its 3-D shape (fig. 16).
3.2.3. Finds record
All artificial finds were registered in a common format Access database with a traditional “logical square” reference. A range of standard fields contain find’s id numbers; year, id and author of an excavation ground; position within cultural layer, “logical layer”, “logical square”, context and a description of each find. A description record is split between