Modelling future land use/land cover and seasonal land surface temperature changes based on CA-ANN algorithm to assess its impacts on Chennai Metropolitan Area (CMA), India
Keywords:
Remote sensing, Land use/Land cover, Land surface temperature, Artificial neural network, Cellular automataAbstract
Urbanization may cause huge amount of Land use and land cover (LULC) changes, which
creates a significant influence on land surface temperature (LST) in fast developing
megacities. This study first analyzed the pattern of LULC changes and then evaluated their
implications on LST in the Chennai Metropolitan Area (CMA) for the years 2000, 2010, and
2020 using Landsat TM/EMT+/OLI satellite images. The study ultimately predicted the future
LULC and LST scenarios for the years 2030 and 2040 using Artificial neural network and
cellular automata algorithms based on previous predicted change maps of LULC and LST. The
study then used correlation analysis to examine the relationship between LULC, LST, and other
vital spectral indices such as NDVI, NDWI, and NDBI for both the summer and winter seasons.
Overall accuracy assessment of 91% in 2000, 89% in 2010 and 92% in 2020, with Kappa
coefficients more than 85% for LULC. The results indicated a considerable decrease in
agricultural land (40.91 %), Forest land (51.60 %) and an increase in built-up area (64.39 %)
from 2000 to 2020, respectively. Maximum LST increases from 34.29°C in 2000 to 41.51°C in
2020 and 35.06°C in 2000 to 41.26 in 2020 during summer and winter seasons respectively
and with substantial LST differences seen among different LULC classes. The predicted
outcomes for 2030 and 2040 show considerable losses of agricultural land, forest land by 4.69
% and 38.95 %, respectively, as well as increases in built-up areas by 16.96 %. The predicted
seasonal LST revealed that in 2030 and 2040, more than 70% and 80% of the summer and
22% and 13% of the winter seasons will likely have LSTs in the 32-34 °C range. The study shows
that LST with NDVI and NDWI are negative correlation and on the other hand, LST with NDBI
are positive correlation. This study can help urban planners, environmental engineers and
agricultural officers design successful policy efforts to protect agricultural and forest areas for
sustainable development.