Page 718 - NGTU_paper_withoutVideo

Modern Geomatics Technologies and Applications

more efficient, the best variables must be selected from the available variables. For this purpose, Cramer's V method were used
in this study.

2. Study area and research process
th
2
The study area is Central Massachusetts in the United States, which has an area of 27363 km (Figure 1). It is the 13
most populous state in the United States and its capital, Boston, has played an important economic and cultural role in American
history. This area has coordinates as described below (Table 1) with a data resolution of 60 m. (Reference system spc83ma1)

Table 1 Study area coordinate

X Y
Min 33847.928761 775398.018214
Max 332107.928761 961488.018214

Fig. 1. Study area

In this study, land use/land cover maps related to 1996, 2006, and 2016 have been used. The first two maps were used to
create a model and predict changes, and the third map was used to control computations and validation. Land use maps for 1996
and 2006 have been compiled from the Clark Institute's website (www.clarklabs.org) and the 2016 map from the Massachusetts
Office of Environmental Affairs (MassGIS). Afterwards, factor maps were obtained for 1996 and 2006 and the development of
each user was predicted separately using the binary logic regression method. Accordingly, using the Markov chain method, a
forecast was made for changes in land use and land cover for 2016, and finally, a comparison was made between the generated
map and the existing map, and validation would be performed.

3. Proposed method

3.1 Logistic Regression
In most land use research, the relationship between the set of independent parameters and a dependent binary output must
be specified. In this regard, the logistic regression is commonly used for this purpose. This method can obtain the effect of the
independent parameters (here is the factor map) according to the dependent variable in the form of 0 and 1, which value 1

2