Presentation

Impacts of Urban and Industrial Development on Arctic Land Surface
Temperature in Lower Yenisei River Region, Russia
Ziqi Li and Nikolay Shiklomanov
The George Washington University, Department of Geography, Washington D.C.
Methods
Urbanization and industrial development have significant impacts on arctic climate
that in turn controls settlement patterns and socio-economic processes. In this study
we have analyzed the anthropogenic influences on regional land surface temperature
of Lower Yenisei River Region of the Russia Arctic. The study area covers two
consecutive Landsat scenes and includes three major cities: Norilsk, Igarka and
Dudingka. Norilsk industrial region is the largest producer of nickel and palladium in the
world, and Igarka and Dudingka are important ports for shipping. We constructed a
spatio-temporal consistent surface temperature model by including 1km MODIS LST,
Modern Era Retrospective-analysis for Research and Applications (MERRA) air
temperature , DEM, and Land cover classification. Those aforementioned spatial data
have various resolution and coverage in both space and time. We analyzed their
coupled relationships and reconstruct a monthly spatio-temporal consistent surface
temperature (Ts) model at 1km resolution from 1980 to 2011. The temperature model
then was used to examine the characteristic summer mean surface temperature
signatures, related to several representative assemblages of Arctic urban and industrial
infrastructure in order to quantify anthropogenic influence on regional surface
temperature.
All spatial data were resampled to 1km resolution and reprojected to EASE-Grid. From 2002-2011, 80%
of summer mean MODIS LST (JJA) were used to derive the multiple linear regression using four variables:
MERRA air temperature, XY coordinates and elevation.
MODIS LST(land cover) = a MERRATair + bX + cY + dDEM
The relationship was used to reconstruct surface temperature for the period 1980-2001. Three
different time interval land cover maps were used. Temperature maps were smoothed out using xyz
coordinates of MODIS pixels.
Land Cover
Ts for different land cover types
Reconstructed Ts
MERRA Tair
1980s
10-yr Mean
Introduction
Study Area
Results
To validate the model, 20% of summer mean temperature data
from 2002 -2011 are randomly chose to test the accuracy. Figure 6.
shows comparison between reconstructed Ts with MODIS LST for
forest, tundra, barren, built-up respectively. All land cover types
yield mean difference (SD) of 0 with standard deviation around 1
degree and Pearson’s correlation around 0.8. Built-up is the best
reconstructed land cover type.
MD = 0.00
SD = 1.26
R = 0.79
MD = 0.00
SD = 1.32
R = 0.79
MD = 0.00
SD = 1.32
R = 0.81
MD = 0.00
SD = 0.93
R = 0.88
1990s
10-yr Mean
a)
b)
Conclusions
• Reconstructing historical surface temperature from MODIS LST
and NASA-MERRA air temperature yields promising accuracy.
• Based on reconstructed surface temperature (Ts), From 1980 –
2011, built-up areas have the highest temperature and warm
faster than the regional warming.
• The urban warming indicates urban and industrial development
have significant influence on land surface temperature in Lower
Yenisei Region.
Figure 2. Photos of Norilsk (a) and Talnakh (b),
industrial area (c), and damaged building (d).
2000s
10-yr Mean
d)
c)
Figure 1. Map of Lower Yenisei River Region. Study area
is bounded by black line. Map courtesy to Kelsey Nyland.
Figure 6. Comparison between reconstructed Ts and MODIS LST.
Acknowledgements
This research was supported by U.S. National Science
Foundation (NSF) grants 1231294, 1558389,1204110, 1304555 and
by the National Research Council of Norway grant
34306/1/ECNS21015N to The George Washington University. We
gratefully thank the University of Waterloo to deliver the MODIS
Monthly LST product. Also, I appreciate the land cover classification
maps provided by Kelsey Nyland.
Data Used
Data
Spatial resolution
Temporal coverage
MODIS LST product1
MERRA 2m Air Temperature
Landsat NDVI - driven Land
Cover Classification2
DEM
1km x 1km
0.5°x 0.66°
30m x 30m
2002-2011
1980-2011
Three time intervals: 1985-1987,
2000-2002, 2012-2014
N/A
1.
2.
1km x 1km
MODIS 1km LST product was preprocessed by University of Waterloo, Canada. The preprocess approach
involves MODIS Aqua/Terra spatio-temporal aggregation and interpolation (Claude et al., 2012).
Land cover classification is based on Landsat NDVI dense time stacking method (Nyland, 2015).
Figure 4.
Water
Barren
Forest
Built-up
Tundra
Figure 3. Land cover maps, reconstructed Ts maps,
MERRA Tair maps and boxplots for three decades.
Figure 5.
References
Nyland, K. E. (2015). Climate-and Human-Induced Land Cover
Change and its Effects on the Permafrost System in the Lower
Yenisei River of the Russian Arctic (Master dissertation, The
George Washington University).
Duguay, Claude R; Soliman, Aiman; Hachem, Sonia; Saunders,
William (2012): Circumpolar and regional Land Surface
Temperature (LST), version 1, with links to geotiff images and
NetCDF files (2007-2010). University of Waterloo, Canada.