Though the urban land use spatial dynamic simulation and forecasting based on cellular automata (CA) model have
achieved remarkable progress, the CA model still has some problems and drawbacks in forecasting urban land use
changes. In view of the deficiencies of traditional urban CA, an improved CA model based on spatial dynamic data
mining and random forecast is proposed in this paper, which establishes an operable CA method to forecast and simulate
the discrete status attribute. This improved CA model is examined in analyzing the urban land use structure changes in
Jinan 2002-2006 and testified both feasible and effective. Based on the remote sensing images in Jinan 2002 and 2006,
the urban land use spatial structures are classified into five types, commercial land, residential land, education facility,
industrial land and the other. With the improved CA model, the urban land use framework in Jinan in 2010 was
calculated, the result of which can be used as a reliable reference information for the following urban land use planning.
Urban function partitioning is one of the important tasks of urban land utilization and management. The goal of urban
function partitioning is to make the urban land form a set of specific function units and regular spatial structures. Spatial
clustering is the key process of urban function partitioning. Based on geographical information system (GIS) and the
principle of self-organizing feature map network, this paper presents a combined -type of spatial clustering. That is to
say, the attribute features and their spatial positions are processed by a unitive spatial clustering model, and then the
isolated "islands" or the "holes" of spatial clustering outputs should be smoothed into their neighborhood. This method
fully mines the connotative spatial clustering information in spatial attribute data and spatial positions. The experiment
shows that the unitive spatial clustering method can provide a sufficient and reliable basis for urban function partitioning.
Using a conventional optical receiver model, we analyzed the penalties on the optical signal performance due to low-frequency subcarriers with a new approach, regarding the subcarrier as a sinusoidal distributed random variable of which the standard deviation is applied in the formula derivation. Then, the Q factor is derived as a simple analytic function of modulation index, OSNR, and the number of subcarriers. By using numerical analysis, our method agrees well with the proved theory and is more efficient and convenient for calculation.