Process of Reconstruction Within a Dataset

The process of reconstruction within a dataset aims to replace missing or invalid values with synthetic values. This process, which you can perform in the Reconstruct Single Dataset window of Windographer, consists of two phases:

Phase 1: Pattern-based reconstruction

Windographer can perform pattern-based reconstruction on four types of data columns: speed, speed SD, direction, and temperature. For each dataset, for each of those four data types:

Choose as the ‘anchor’ the data column of this type and subtype with the highest data coverage rate.
For the anchor column, perform the pattern-based reconstruction mechanism.
If using Markov-based synthesis, perform the Markov-based reconstruction mechanism on the anchor column.
For all other columns of this type and subtype, perform the pattern-based reconstruction mechanism.

Phase 2: (Optional) Markov-based reconstruction

For each dataset, for each data column not already reconstructed, perform the Markov-based reconstruction mechanism.

This approach maximizes the use of pattern-based reconstruction and minimizes the use of Markov-based reconstruction. For data columns that can be reconstructed using the pattern-based mechanism, it reconstructs them as much as possible with that pattern-based mechanism. It uses the Markov-based mechanism only on the anchor column of each type, and only in those time steps that contain no valid data of that type at any measurement height. As a result, in time steps in which it reconstructs values for multiple data columns, those synthetic values always obey the observed correlation patterns.