Environ Monit Assess DOI 10.1007/s10661-014-3995-z

Analysis of multi-temporal landsat satellite images for monitoring land surface temperature of municipal solid waste disposal sites Wai Yeung Yan · Prathees Mahendrarajah · Ahmed Shaker · Kamil Faisal · Robin Luong · Mohamed Al-Ahmad

Received: 13 November 2013 / Accepted: 12 August 2014 © Springer International Publishing Switzerland 2014

Abstract This study presents a remote sensing application of using time series Landsat satellite images for monitoring the Trail Road and Nepean municipal solid waste (MSW) disposal sites in Ottawa, Ontario, Canada. Currently, the Trail Road landfill is in operation; however, during the 1960s and 1980s, the city relied heavily on the Nepean landfill. More than 400 Landsat satellite images were acquired from the US Geological Survey (USGS) data archive between 1984 and 2011. Atmospheric correction was conducted on the Landsat images in order to derive the landfill sites’ land surface temperature (LST). The findings unveil that the average LST of the landfill was always higher than the immediate surrounding vegetation and air

temperature by 4 to 10 ◦ C and 5 to 11.5 ◦ C, respectively. During the summer, higher differences of LST between the landfill and its immediate surrounding vegetation were apparent, while minima were mostly found in fall. Furthermore, there was no significant temperature difference between the Nepean landfill (closed) and the Trail Road landfill (active) from 1984 to 2007. Nevertheless, the LST of the Trail Road landfill was much higher than the Nepean by 15 to 20 ◦ C after 2007. This is mainly due to the construction and dumping activities (which were found to be active within the past few years) associated with the expansion of the Trail Road landfill. The study demonstrates that the use of the Landsat data archive can provide additional and viable information for the aid of MSW disposal site monitoring.

W. Y. Yan () · P. Mahendrarajah · A. Shaker · K. Fasial · R. Luong Department of Civil Engineering, Ryerson University, 350 Victoria Street, Toronto, Ontario M5B 2K3, Canada e-mail: [email protected]

Keywords Municipal solid waste · Landfill · Landfill gas · Land surface temperature · Multi-temporal · Landsat · Remote sensing

P. Mahendrarajah Department of Civil Engineering, McMaster University, 1280 Main Street West, Hamilton, ON L8S 4L7, Canada

Introduction

K. Fasial Department of Geomatics, College of Environmental Design, King AbdulAziz University, Makkah, Saudi Arabia M. Al-Ahmad Environment Public Authority, Safat, Kuwait

The municipal authority is responsible for monitoring waste disposal (or landfill) sites to prevent any potential threats toward natural habitat and public health. Recycling and incineration are two common solutions for waste disposal. Nevertheless, the majority of wastes are still buried in landfills. Generally, municipal solid waste (MSW) disposal sites are well

Environ Monit Assess

designed and established with comprehensive monitoring systems including leachate collection facilities, deep and shallow water monitoring stations, soil sample collection, etc. With the collected data samples (e.g., leachate, aquifer, water, etc.), the organic content and biochemical properties can be examined for assessing the landfill biodegradation stability (Cobo et al. 2008). Nonetheless, implementation of these monitoring schemes requires proper equipment and laboratory testing which consume both time and effort. In view of these costs, researchers have attempted to investigate other viable methods to support landfill site monitoring. Jones and Elgy (1994) suggested that remote sensing technology can be used as a supplement to, rather than a substitute for, ground monitoring systems. Remote sensors are able to record the spectral reflectance of topographic features across visible to near-infrared wavelengths. By studying the recorded spectral measurements, geographical and biophysical information of the observed features can be determined. Early researches utilized historic aerial photography to identify hazardous waste sites based on the size, shape, and spectral reflectance of the landfill sites, as well as the texture of surrounding features (Erb et al. 1981; Lyon 1987; Bagheri and Hordon 1988). Pope et al. (1996) proposed deriving information for characterization of waste disposal sites using multi-temporal aerial images and geographic information system (GIS). Stereoscopic measurements were also carried out to delineate and extract the boundary of landfill sites (Stohr et al. 1987; Philipson et al. 1988; Stohr et al. 1994). However, these early studies mainly relied on visual interpretation and manual digitization, which are labor intensive when dealing with distributed landfill sites located in large areas. Digital image analysis techniques were thus being introduced to retrieve additional geographical and biophysical characteristics of the landfill sites. Brivio et al. (1993) introduced a spatial autocorrelation (semivariogram) method to identify and classify landfill sites by studying the contrast of their spatial features with respect to the surrounding rural landscape on a Landsat Thematic Mapper (TM) image. Jones and Elgy (1994) performed unsupervised classification on an Airborne Thematic Mapper (ATM) image to produce a crop condition map, which was used to establish a relationship with the crop (and soil) condition data collected by ground survey. Any

unhealthy condition of surface vegetation or soil would result in an indication of landfill gas migration. Dewidar (2002) used two Landsat TM images and investigated various change detection techniques including image differencing, image ratioing, and post-classification change to determine the landfill areas from 1984 to 1987 in Hurghada, north Red Sea, Egypt. Silvestri and Omri (2008) introduced IKONOS satellite images to study the radiometric properties of stressed vegetation which gives an indication on the possible location of illegal landfill sites. In addition to multispectral satellite images, remote sensors such as ground-penetrating radars (Well et al. 1994; Daniels et al. 1995) and hyperspectral sensors (Slonecker et al. 2010) were used to investigate hazardous waste sites and detect contaminated areas. Ottavianelli (2007) investigated the possibility of distinguishing landfill sites from other land features using synthetic aperture radar (SAR) and correlated the SAR data with onsite conditions and operational procedures. The experiments demonstrated the usefulness of spatial characteristics of SAR backscatter and complex degree of coherence for landfill site identification. Im et al. (2012) used airborne hyperspectral sensors (HyMap) to characterize the vegetation cover on two capped landfill sites in Utah and Arizona, USA. In order to aid in the characterization of vegetation covers, the leaf area index and vegetation species mapping were computed. Noomen et al. (2008) and Noomen et al. (2012) used different hyperspectral indices to measure the canopy reflectance and investigated its relationship with the soil oxygen concentration. This can serve as an indicator of underground gas/methane leakage. Studying the thermal behavior of waste could be another possible alternative in characterizing landfill sites since the decomposition of waste increases the temperature differently during the aerobic and anaerobic digestion processes (Yes¸iller et al. 2005). Zilioli et al. (1992) utilized a thermal camera to detect potential anomalies in a landfill site by measuring the waste temperature. Lewis et al. (2003) performed a similar study and addressed the limitations of using infrared thermography for detecting landfill gas leakage. The research unveiled that the accuracy of measurements is influenced by weather conditions, nature of the ground surface, and distance between the sensor and target. Nevertheless, these findings verify the feasibility of using thermal measurements to

Environ Monit Assess

differentiate the waste under different biodegraded conditions and pave the way for future experiments with airborne and space-borne sensors. Recently, Kwarteng and Al-Enezi (2004) and Yang et al. (2008) explored the use of Landsat thermal images to monitor and detect landfill sites by deriving the land surface temperature (LST). In comparison to the immediate surroundings, higher LST values (in a few degrees Celsius) were reported within the study landfill site. In addition, Shaker et al. (2010) introduced temporal Landsat data to compute the LST for identifying suspicious dumping locations within an uncontrolled landfill in Kuwait. In view of these attempts in using different remote sensing data and techniques for landfill site monitoring, this study reaps the benefits of using Landsat TM images obtained from the US Geological Survey (USGS) data archive to monitor two adjacent MSW disposal sites in the city of Ottawa, Ontario, Canada. More than 400 images were acquired from the USGS EarthExplorer1 for analyzing the change of LST between 1984 and 2011 in the study landfill sites.

Study sites The study sites are the Nepean landfill and the Trail Road landfill (45◦ 14 N, 75◦ 45 W) which are located at the west of highway 416 in Ottawa, Ontario, Canada. Both landfill sites, which are the major MSW disposal sites for the city of Ottawa, are approximately 12 km away from the Ottawa MacDonald-Cartier Airport. Within the immediate vicinity of the 500acre landfill space, farmlands, grasslands, trees, and minor industrial facilities exist. Other distinguished roads that enclose the landfills include Barnsdale Road which bounds the southern border of the landfill and Cambrian Road which runs northeast, enclosing the northern region of the landfill (see Fig. 1). The Trail Road landfill is situated adjacent to the Nepean landfill and began to accept solid waste from the Ottawa residences in late 1980. Nonhazardous wastes in the landfill are mainly garbage from residents, industries (lightweight waste), construction sites, and commercial businesses residing in Ottawa

1 http://earthexplorer.usgs.gov/

(Dillon Consulting Limited 2008). The establishment of the Trail Road landfill was mainly due to the full operation of the adjacent Nepean landfill (operated since early 1960s and accepted waste until early 1980s). With the increasing amount of waste disposal, the Trail Road landfill was established based on the new property area that was acquired to the Nepean landfill site in March 1975 on the north side of Trail Road. In December 1978, the Trail Road landfill was constructed and waste disposal operations began in May 1980. Trail Road landfill comprises of four stages which were developed sequentially beginning at stage 1 (situated at the far east) and ending with stage 4 in the west. According to the annual landfill monitoring report (Dillon Consulting Limited 2008), the first two stages of the landfill were closed and capped with a polyethylene liner in 1988 and 1991, respectively. However, the design of the initial two stages did not include a geomembrane cover beneath the landfill for solid waste containment. Therefore, waste was buried directly over the sand or open pits, which may cause potential contamination underground. On the other hand, stages 3 and 4 incorporated a geomembrane cover and a leachate collection system in order to prevent leachates from contaminating groundwater and soil. A small portion of the collected leachate from stage 3 was rereleased into stage 4 and the rest was treated at the Robert O. Pickard Environmental Centre (Dillon Consulting Limited 2005) for purification. In addition, the Power Trail Inc. is currently operating and converting the leachate obtained from the landfill into green energy. Due to the increased disposal rate of waste, approval was granted in mid-2007 for expanding the previous stages and initiating construction of stage 5. In fact, in 2007, 216,700 tons of waste was required for disposal which was 28 % higher than the preceding year (Dillon Consulting Limited 2008). The Nepean and Trail Road landfill sites are currently operated and managed under the environmental monitoring program which is approved by the Ontario Ministry of the Environment. Samples of groundwater from shallow and deep aquifers are acquired from monitoring wells, whereas samples from surface water are acquired from ditches, rivers, surrounding streams, and ponds. The tests performed on nearby water bodies are analyzed for its chemical composition in order to determine the degree of immediate environmental damage caused by operation of the landfill. Landfill

Environ Monit Assess

Ottawa Trail Road Landfill Nepean Landfill

0

250

500 m

Esri, HERE, DeLorme, MapmyIndia, ' OpenStreetMap contributors, Source: Esri, DigitalGlobe, GeoEye, i-cubed, USDA, USGS, AEX, Getmapping, Aerogrid, IGN, IGP, swisstopo, and the GIS User Community

Fig. 1 Location of the Nepean and Trail Road landfills (left) and an aerial photo showing the study sites (right)

gas monitoring wells are also utilized within the landfill sites to measure the concentration of methane (CH4 ) in the samples. All these ground measurements form a comprehensive environmental monitoring program that is required to report annually to the city of Ottawa Municipality.

carried out in PCI Geomatica to manipulate the acquired Landsat images. Five individual EASI programs were developed to accomplish the aforementioned tasks (data import, image clipping, atmospheric

Methodology Overall workflow Figure 2 shows the overall workflow for the Landsat data processing. Since the downloaded Landsat images were geo-referenced (L1T format), the images were directly imported and appended into PCI Geomatica V10.3.2. The imported data in PCI proprietary format were cropped to a common spatial extent, and atmospheric correction was conducted in each of the clipped images together with sensor and weather information. With the atmospherically corrected Landsat images, LST and normalized difference vegetation index (NDVI) were derived from the multispectral bands. Finally, the LST and NDVI images were exported in GeoTIFF format for data analysis under a GIS platform (ESRI ArcGIS Version 10). Due to the massive downloaded data (∼75 GB) from the USGS EarthExplorer, batch processing was

Fig. 2 Overall workflow for Landsat data processing

Environ Monit Assess

correction, computation of LST and NDVI, and data export). Since the study area was found to be covered by either cloud or snow in some of the Landsat scenes, manual checks were needed to ensure the usefulness and correctness of the results. Finally, 81 Landsatderived LST images were selected for the season of spring (March 26 to May 1), summer (June 15 to July 15), and fall (September 15 to October 30) from 1984 to 2011. The LST and NDVI of the Trail Road landfill (stages 1 to 4), Nepean landfill and surrounding vegetation were computed for further analysis. Landsat data Since the launch of the first Earth observation satellite, Landsat Multispectral Scanner System (MSS), scientists have extensively utilized these satellite images to serve a variety of Earth science and environmental applications. In 2008, the USGS has started to release the Landsat data archive for free download through the USGS EarthExplorer. Such an action has apparently rendered a significant impact toward academician and industries since remote sensing studies are no longer limited by the previous prohibitive cost of Landsat single-scene data. Recently, a large number of temporal studies were reported in biomass estimation (Gasparri et al. 2010), urban heat island (Li et al. 2012), monitoring vegetation phenology (Bhandari et al. 2012), and forestry management (Griffiths et al. 2012) using time series Landsat data. The Landsat satellite images acquired for the Nepean and Trail Road landfills (paths 15 and 16 and rows 28 and 29) from 1984 to 2011 were obtained from the USGS EarthExplorer. Due to the scan line corrector problem (Markham et al. 2004), Landsat ETM+ images were not considered. The downloaded Landsat TM images have seven bands: band 1 (blue), 2 (green), 3 (red), 4 (near-infrared), 5 (mid-infrared), 6 (thermal), and 7 (mid-infrared). All the bands were ortho-rectified by USGS in UTM 17N coordinate system with reference datum of WGS84. The multispectral bands, including bands 1 to 5 and 7, have a resolution of 30 m, whereas the thermal band (band 6) has a resolution of 60 m. Among the 465 images that were downloaded, a total of 83 images were found to be incorrectly geo-referenced and another 139 images are worthless due to the presence of cloud and snow cover. The experiment intended to incorporate

three images representing different seasons in each year. Due to the lack of cloud-free or snow-free Landsat images, data in Fall 1991, Fall 1994, and Summer 1998 were not included. Finally, 81 images were selected for the seasons of fall (September and October), summer (June and July), and spring (March and April). Table 1 lists the Landsat TM images used in this study. Landsat image processing Atmospheric correction is a significant step in many remote sensing applications for removing atmospheric attenuation due to molecular/aerosol absorption and scattering. Atmospheric correction is particularly important if biophysical information (i.e., vegetation indices, LST, etc.) or temporal studies are conducted on the remote sensing images (Ou et al. 2002; Hadjimitsis et al. 2010). There are two different ways of performing atmospheric correction: absolute correction and relative correction. Although the relative correction approach is easier to implement, absolute correction preserves better results if sensor parameters and weather condition can be acquired (Paolini et al. 2006). In this study, the absolute atmospheric correction model, ATCOR2 (Atmospheric Correction and Haze Reduction), built-in PCI Geomatica was used (Richter 1990, 1996). Several pieces of information were imported in the ATCOR2 model including the sensor information (i.e., sensor type, acquisition date, and pixel size), atmospheric condition (with reference to the weather information from the Ottawa weather office), the atmospheric model (US standard model and six supplementary models), correction parameters (i.e., solar zenith and azimuth, visibility and adjacency), and calibration parameters. The revised calibration parameters of Landsat TM were used as reported in Chander et al. (2007). After running the ATCOR2 model, the surface reflectance was retrieved in each of the Landsat bands. Then, LST and NDVI values for each Landsat dataset were derived using the corrected Landsat band 6 (thermal band) and bands 3 and 4 (red and near-infrared bands), respectively. Finally, all the results of LST and NDVI were exported as a TIFF file for further analysis. Figures 3 and 4 show an example of multi-temporal Landsat images and the derived LST during the summer, respectively.

Environ Monit Assess Table 1 List of Landsat TM images and the corresponding air temperature Spring

Summer

Date (M/D/Y)

Path/ row

Air temperature

04/12/1984 05/01/1985 04/18/1986 04/14/1987 03/31/1988 04/26/1989 04/13/1990 04/25/1991 04/27/1992 04/14/1993 04/08/1994 03/26/1995 04/06/1996 04/16/1997 04/12/1998 04/15/1999 04/24/2000 04/20/2001 04/06/2002 04/17/2003 04/12/2004 04/06/2005 04/02/2006 04/21/2007 04/07/2008 04/17/2009 04/13/2010 04/07/2011

16/29 16/28 16/29 15/29 15/29 16/29 16/29 15/29 15/29 15/29 16/28 16/29 15/29 16/29 15/29 15/29 16/28 15/29 16/28 16/28 15/29 16/28 15/29 15/29 15/29 16/29 15/29 16/28

10.9 ◦ C 10.9 ◦ C 15.3 ◦ C 11.2 ◦ C 6.4 ◦ C 11.8 ◦ C 4.9 ◦ C 14.7 ◦ C 11.3 ◦ C 9.0 ◦ C 5.1 ◦ C 3.1 ◦ C −4.7 ◦ C 13.9 ◦ C 12.2 ◦ C 8.7 ◦ C 12.1 ◦ C 13.5 ◦ C −3.2 ◦ C −1.6 ◦ C 6.0 ◦ C 12.8 ◦ C 5.6 ◦ C 20.0 ◦ C 6.2 ◦ C 18.1 ◦ C 10.7 ◦ C 6.7 ◦ C

Fall

Date (M/D/Y) 07/01/1984 07/13/1985 07/07/1986 06/24/1987 07/28/1988 06/29/1989 07/02/1990 06/19/1991 07/16/1992 06/24/1993 06/20/1994 06/23/1995 07/11/1996 06/28/1997 07/03/1999 06/20/2000 06/14/2001 07/03/2002 07/06/2003 06/15/2004 07/04/2005 07/07/2006 06/15/2007 07/12/2008 07/15/2009 07/02/2010 07/05/2011

Path/ row 16/28 15/29 16/29 16/29 16/29 16/29 16/29 16/29 15/29 16/29 15/29 15/29 15/29 15/29 No Data 16/29 15/29 16/28 16/29 16/29 15/29 15/29 15/29 16/29 15/29 15/29 15/29 15/29

Results and analysis Results of LST Figure 5 shows the LST of the landfill and the surrounding vegetation acquired from the Landsat images as well as the air temperature obtained from the nearby weather station between 1984 and 2011. Temperature variation demonstrated a periodic cycle due to the seasonal change. Peak values of LST were prominently found in the landfill during the summer while minima mainly appeared in either fall or spring. In general, it is apparent that the LST of the landfill (red line) was

Air temperature

Date (M/D/Y)

26.0 ◦ C 25.3 ◦ C 26.3 ◦ C 30.9 ◦ C 26.3 ◦ C 18.7 ◦ C 23.2 ◦ C 27.8 ◦ C 22.0 ◦ C 24.0 ◦ C 26.6 ◦ C 26.1 ◦ C 23.8 ◦ C 28.2 ◦ C

10/05/1984 10/17/1985 10/11/1986 10/14/1987 10/16/1988 10/28/1989 10/31/1990

26.0 ◦ C 22.8 ◦ C 29.3 ◦ C 31.4 ◦ C 27.4 ◦ C 23.9 ◦ C 27.5 ◦ C 25.3 ◦ C 26.6 ◦ C 23.7 ◦ C 20.5 ◦ C 24.5 ◦ C 27.5 ◦ C

10/04/1992 10/14/1993 09/18/1995 09/20/1996 10/25/1997 09/19/1998 10/07/1999 10/01/2000 10/20/2001 10/23/2002 10/10/2003 10/05/2004 09/13/2005 09/16/2006 10/05/2007 10/23/2008 10/10/2009 10/13/2010 10/09/2011

Path/ row

Air temperature

16/29 15/29 16/29 16/29 16/29 15/29 15/29 No Data 15/29 16/29 No Data 16/29 16/29 16/29 15/29 16/29 16/28 16/28 16/28 16/28 15/29 16/28 16/28 16/29 16/28 16/28 16/29 15/29

3.5 ◦ C 8.4 ◦ C 10.8 ◦ C 11.1 ◦ C 11.0 ◦ C 16.5 ◦ C 1.8 ◦ C 7.9 ◦ C 7.2 ◦ C 11.1 ◦ C 18.5 ◦ C 3.6 ◦ C 24.4◦ C 5.3 ◦ C 20.8 ◦ C 14.5 ◦ C 4.0 ◦ C 15.8 ◦ C 7.2 ◦ C 28.8 ◦ C 21.0 ◦ C 23.8 ◦ C 5.5 ◦ C 10.0 ◦ C 13.0 ◦ C 21.1 ◦ C

found to be higher than the surrounding vegetation (green line) and the air temperature (blue line). Nevertheless, the difference in the temperature varies with respect to the season. During 1984, the LST of the landfill in spring was 19.8 ◦ C, whereas the LST of vegetation and air temperature was 13.8 and 10.9 ◦ C, respectively. Gradual changes were noticeable in the LST when compared to the summer of 1984. In fact, the LST of the landfill spiked to 40.4 ◦ C, whereas the LST of vegetation and air temperature were both lower than 30 ◦ C. During fall, changes in weather condition caused a drop in the air temperature and LST of the vegetation

Environ Monit Assess

Fig. 3 Multi-temporal Landsat images for the MSW sites

Fig. 4 The derived land surface temperature for the MSW sites

Environ Monit Assess 50 LST of Landfill

LST of Vegetation

Air Temperature

Degree Celcius (°C)

40 30 20 10 0 1984 -10

1986

1988

1990

1992

1994

1996

1999

2001

2003

2005

2007

2009

2011

Fig. 5 A temperature plot showing the LST of the landfill and the surrounding vegetation derived from the Landsat images and the corresponding air temperature between 1984 and 2011

by approximately 4 ◦ C. Thus, the LST of the landfill was approximately 6 ◦ C higher than the air temperature and the LST of the vegetation. During the study period, this trend was predominant; however, it should also be noted that after mid-2007, the difference between the LST of the landfill and the air temperature became significant. Primarily, this was due to the approval of vertical expansion of previous stages and construction of stage 5 in the Trail Road landfill. For a further breakdown on the LST for different stages of the Trail Road landfill, refer to Section 4. Figure 6 further summarizes the average temperature difference between the LST of the landfill and the surrounding vegetation. In addition, it also summarizes the correlation between the landfill LST and air temperature in accordance to the three seasons.

Obviously, the temperature difference was notable in spring where 6.5 and 8.8 ◦ C was found between the LST of the landfill and the vegetation and between the LST of the landfill and the air temperature, respectively. Referring to Figs. 3 and 4, it is obvious that the LST of landfill was very high during the summer. Since the Landsat images were acquired at around 11:00 a.m., some of the bare ground and urban areas located in the north and east of the landfills also indicated high LST due to the heat absorption. However, the vegetation cover located in the south of the landfill consistently showed a relatively low LST during summer. The landfill LST was higher than the vegetation by 10.2 ◦ C and the air temperature by 11.5 ◦ C, whereas the temperature difference dropped to more than a half in fall. Approximately 5 ◦ C was found in the two categories of temperature difference.

Temperature Difference (°C)

LST between the closed landfill (Nepean) and active landfill (Trail Road) 14 12

Landfill v.s. Vegetation Landfill v.s. Air

10 8 6 4 2 0 Spring

Summer

Fall

Fig. 6 Average temperature difference between the LST of the landfill and the surrounding vegetation and the difference between the LST of the landfill and the air temperature in spring, summer, and fall

Figure 7 shows the LST of the Nepean landfill (closed) and the Trail Road (active) landfill between 1984 and 2011. In general, the active Trail Road landfill had a slightly higher LST than the closed Nepean landfill. However, in certain years, there are slightly larger differences in the LST between the two landfills. For instance, after the Spring 2008, the Trail Road landfill had a greater difference (approximately 15 to 20 ◦ C) in its LST as opposed to prior years. This further reinforces the fact that increased construction due to the approval of the vertical expansion of the landfill may be the cause for the higher LST. In 2006, the Trail Road landfill was approximately 3 ◦ C greater than

Environ Monit Assess

Degree Celcius (°C)

50

LST of Nepean

(a)

LST of Trail Road

40 30 20 10 0 1984

1985

1986

1987

Degree Celcius (°C)

50

1988

1989

1990

1991

LST of Nepean

1992

1993

1994

1995

1996

1997

(b)

LST of Trail Road

40 30 20 10 0 1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

Fig. 7 The LST of Nepean and Trail Road landfills between (a) 1984 and 1997 and (b) 1998 and 2011

the closed landfill, which may be as a result of the construction of the energy plant which was undertaken during 2006. Probable causes for these proximities in LST between the two landfills throughout the study period may be due to the seasonal change. During summer, the LST of landfill reaches a maximum spike since warmer climates induce decomposition, whereas during spring, dry and cooler climate inhibits bacteria from decomposing, thereby causing lower temperatures (minimum on graph). In some instances, such as in Spring 2007, the LST of the landfill is not near the minimum or maximum. This may be due to the presence of rainy weather, which produces moisture in the landfill, thus influencing bacteria growth and producing more landfill gasses. Therefore, changes in weather condition and season may be the cause for fluctuations in the LST. Some other potential reasons for fluctuation may be due to use of machinery for expansion of the landfill, capping of the landfill, installation of new wells, or the heterogeneity of dumped wastes. LST between the capped (stages 1 and 2) and open (stages 3 and 4) stages in the trail road landfill Figure 8 indicates the mean LST of stages 1 and 2 (capped landfill) and stages 3 and 4 (open landfill) of the Trail Road landfill in spring, summer and fall, respectively. Generally, it is noticeable that waste buried in stages 3 and 4 of the Trail Road landfill had higher LST than that of stages 1 and 2 (regardless of the seasons). Figure 8, which depicts

the LST in spring, indicates no significant difference in LST between the capped and open stages from 1984 to 2007 (with a difference of ±2 ◦ C). Similar patterns can be observed in the fall season, where the LST difference fluctuated within a minimum range (±3 ◦ C). Nevertheless, starting from 2008, there existed a notable difference in the LST between the closed and the open stages (for both spring and fall seasons) due to the recent approval of expansion in stages 3 and 4. The LST of open stages was always higher than the closed stages by 7.3 to 14.6 ◦ C in spring and 7.3 to 13.3 ◦ C in fall (with an exception of the minimum LST found in stages 3 and 4 on April 7, 2008). This can be explained by partially covered snow and haze on the landfill sites which may affected the validity of the computed LST. Similar to the pattern of spring and fall, the LST of open stages and closed stages were almost the same in summer (Fig. 8). However, this pattern is only valid between 1985 and 1990. One should also note that in 1984, there was a 4 ◦ C higher LST in stages 1 and 2 than stages 3 and 4. This is mainly due to the fact that stages 3 and 4 were not readily used for waste disposal as stages 1 and 2. Between 1991 and 2008, significant LST differences can be observed. Although the difference of the LST between the open and capped stages fluctuated, the LST of the open stages always demonstrated a higher value than that of the closed stages by 0.4 to 10.4 ◦ C. The most obvious LST differences were noted after 2008 where the LST of open stages was always greater than the LST of closed stages by 10 to 15 ◦ C.

Environ Monit Assess

Degree Celcius (°C)

50

(a)

Stages 1 and 2

40

Stages 3 and 4

30 20 10 0 1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

Degree Celcius (°C)

70 60

Stages 1 and 2

50

Stages 3 and 4

(b)

40 30 20 10 0 1984

Degree Celcius (°C)

2010

40 35 30 25 20 15 10 5 0

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

2010

(c)

Stages 1 and 2 Stages 3 and 4

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

2010

Fig. 8 The LST of capped stages (stages 1 and 2) and open stages (stages 3 and 4) of Trail Road landfill in (a) spring, (b) summer, and (c) fall

Discussion There are a number of potential reasons to explain the fluctuation of the measured LST in the landfill sites. The biodegradation process is ultimately a main reason for the elevated LST of the landfill sites. Once wastes accumulate in the landfill, the biochemical reaction begins in the organic components of wastes, such as food and yard wastes. Such process can be categorized as aerobic digestion and anaerobic digestion (Agency for Toxic Substances and Disease Registry 2001). The aerobic digestion is undergone with the presence of oxygen, and as a result, the digestion process generates high amounts of heat flux during the

decomposition of the organic compounds. The process is most apparent near the surface of the landfill due to active dumping activities resulting in a high surface temperature. For the anaerobic digestion process, the principal bio-reaction in landfills, decomposition does not require oxygen but employs water as the main reagent and releases a small amount of heat. As a result of this process (which may last for decades), landfill gas is produced with mainly methane and carbon dioxide, plus small amounts of ammonia, hydrogen sulfide, and other minor constituents (Themelis and Ulloa 2007). Since Landsat image acquires the instantaneous heat flux of the land surface in the thermal infrared

Environ Monit Assess

wavelength, the derived LST refers to the heat flux generated at the near surface wastes of the landfill. In addition, the type of vegetation/soil cover may have certain influence on the thermal absorption and emission. The significantly higher landfill LST than air temperature (as shown in Fig. 6) echoes the field measurements of landfill temperature as reported in Yes¸iller et al. (2005) regardless of the seasons. In addition, the rate of temperature increase is higher for newly dumped wastes compared to the old wastes (Yes¸iller et al. 2008), which can also be observed in our findings as shown in Figs. 7 and 8. Other factors that affect landfill gas production include moisture content and temperature. The presence of moisture from rain or snow can induce landfill gas production because such an environment provides an optimal condition for bacteria to decompose during the anaerobic digestion (Themelis and Ulloa 2007). As a result of precipitation, the biochemical decomposition process accelerates and induces heat generation. Such a phenomenon can be observed in Fig. 6, where the temperature differences of landfill with the air temperature or surrounding vegetation are always higher in spring compared to those in fall. With even higher air temperature and precipitation in summer, bacteria thrive better than the other two aforementioned seasons, which thus increase the rate of the biodegradation process. More heat turns out in the buried wastes, and thus increases the overall landfill site surface temperature. Such argument echoes our findings in Figs. 6 and 8 since the elevated LST was always found during the summer. The findings of this research is able to provide fruitful information for landfill site monitoring in various aspects. Due to the fact that landfill site produces higher LST than its immediate surroundings, such a phenomenon can be used as an indicator to detect suspicious or unauthorized dumping activities (Shaker et al. 2010). On the other hand, emission of methane can be somehow estimated since there is indeed a moderate correlation between the methane emission and the LST (Ishigaki et al. 2005). Most importantly, once a landfill site is stabilized after the decomposition process, the difference of LST between the landfill and the surroundings should reach a minimum. Thus, remote sensing-derived LST can be used to examine the stability of landfill sites, which is necessary to confirm before any landfill mining tasks (Krook et al. 2012). Ultimately, it is desirable to

have a space-based landfill information system for regional/national environmental monitoring. The free Landsat data archives together with the Google Earth Engine2 can be integrated to serve such a purpose.

Conclusions and future work A number of studies were conducted to use remote sensing images to detect and identify landfill sites, while thermal analysis of MSW sites using remote sensing techniques is limited. This work presents the first attempt, to the best of our knowledge, of using time series Landsat-derived LST for landfill site monitoring and analysis. The study sites include the Nepean landfill and the Trail Road landfill which are the major MSW disposal sites for the city of Ottawa, Ontario, Canada. More than 400 Landsat TM images were downloaded from the USGS EarthExplorer. In order to proceed with the experimental analysis, 81 cloud-free Landsat TM images were selected for the spring, summer, and fall seasons. The LST of the Nepean landfill, Trail Road landfill (including the four stages), and the surrounding vegetation were computed and extracted from the atmospherically corrected Landsat images. Depending on the seasons, the landfill sites had higher average LST than the surrounding vegetation by 4 to 10 ◦ C and air temperature by 5 to 11.5 ◦ C. The temperature difference was obvious in summer and less obvious in fall. It should also be noted that there was no significant difference between the LST of the Nepean landfill (closed) and Trail Road landfill (active) from 1984 to 2007. Since mid-2007, further expansion was approved and initiated. Ultimately, this resulted in the LST of the Trail Road landfill to be 15 to 20 ◦ C higher than the Nepean landfill during the Summer. Within the Trail Road landfill, the capped stages (stages 1 and 2) and open stages (stages 3 and 4) did not show significant LST difference in spring and fall seasons (± 2◦ C to 3◦ C) before 2008. Again, due to the construction activities in the open stages (stages 3 and 4), the LST difference rose up 7 to 15 ◦ C in these two seasons. Throughout the summer, the difference of LST between the capped and open stages was obvious due to the biochemical decomposition process. The open stages had a higher LST than the 2 http://earthengine.google.org

Environ Monit Assess

capped stages by 0.4 to 10.4 ◦ C between 1991 and 2008 and 10 to 15 ◦ C between 2009 and 2011. Both landfill sites appeared to have a consistent higher LST than the immediate surroundings; however, the Landsat-derived LST may be exaggerated due to daytime solar radiation. Since Landsat satellites orbits are sun synchronous, where the revisiting time is fixed at approximately 11:00 a.m. local standard time, thermal measurements taken in predawn condition, as recommended by Zilioli et al. (1992), would be impossible. Therefore, calibration of Landsat-derived LST with ground measured LST is desired for a better analysis of thermal radiation from the landfill sites. Future work will also focus on the correlation between the LST and the ground measured data such as the amount of methane measured from the ground monitoring wells. In addition, the NDVI derived within and nearby the landfill sites will be investigated in order to see if there are any potential contamination which affects the healthiness of the vegetation cover. Acknowledgements This paper is an extended version of the paper “Trail Road Landfill Site Monitoring Using MultiTemporal Landsat Satellite Data” presented at the Geomatics Conference 2010 and ISPRS COM I Symposium, Calgary, Canada. The research was supported by a discovery grant from the Natural Sciences and Engineering Research Council of Canada (NSERC). We acknowledge Mr. Peter Filipowich from the city of Ottawa Government, who provided the Nepean and Trail Road landfill monitoring reports. We also thank the anonymous reviewers for their comments.

References Agency for Toxic Substances and Disease Registry (2001). Landfill gas primer—an overview for environmental health professionals. www.atsdr.cdc.gov/HAC/landfill/html/intro. html. Bagheri, S., & Hordon, R.M. (1988). Hazardous waste site identification using aerial photography: a pilot study in Burlington County, New Jersey, USA. Environmental Management, 12(1), 119–125. Bhandari, S., Phinn, S., Gill, T. (2012). Preparing Landsat Image Time Series (LITS) for monitoring changes in vegetation phenology in Queensland, Australia. Remote Sensing, 4(6), 1856–1886. Brivio, P.A., Doria, I., Zilioli, E. (1993). Aspects of spatial autocorrelation of Landsat TM data for the inventory of waste-disposal sites in rural environments. Photogrammetric Engineering & Remote Sensing, 59(9), 1377–1382. Chander, G., Markham, B.L., Barsi, J.A. (2007). Revised Landsat-5 thematic mapper radiometric calibration. IEEE Geoscience and Remote Sensing Letters, 4(3), 490–494.

Cobo, N., L´opez, A., Lobo, A., Zamorano, M., Brebbia, C.A., Kungolos, A.G., Popov, V., Itoh, H., et al. (2008). Biodegradation stability of organic solid waste characterized by physico-chemical parameters. In Waste management and the environment IV. International conference on waste management and the environment (pp. 153–162). Granada: WIT Press. Daniels, J.J., Roberts, R., Vendl, M. (1995). Ground penetrating radar for the detection of liquid contaminants. Journal of Applied Geophysics, 33(1-3), 195–207. Dewidar, K.M. (2002). Landfill detection in Hurghada, north Red Sea, Egypt, using Thematic Mapper images. International Journal of Remote Sensing, 23(5), 939–948. Dillon Consulting Limited (2005). Trail road landfill site 2004 monitoring and operating report. Tech. rep., City of Ottawa, Canada. Dillon Consulting Limited (2008). Trail road landfill site 2007 monitoring and operating report. Tech. rep., City of Ottawa, Canada. Erb, T.L., Philipson, W.R., Teng, W.L., Liang, T. (1981). Analysis of landfills with historic airphotos. Photogrammetric Engineering & Remote Sensing, 47(9), 1363– 1369. Gasparri, N.I., Parmuchi, M.G., Bono, J., Karszenbaum, H., Montenegro, C.L. (2010). Assessing multi-temporal Landsat 7 ETM+ images for estimating above-ground biomass in subtropical dry forests of Argentina. Journal of Arid Environments, 74(10), 1262–1270. Griffiths, P., Kuemmerle, T., Kennedy, R.E., Abrudan, I.V., Knorn, J., Hostert, P. (2012). Using annual time-series of Landsat images to assess the effects of forest restitution in post-socialist Romania. Remote Sensing of Environment, 118, 199–214. Hadjimitsis, D.G., Papadavid, C., Agapiou, A., Themistocleous, K., Hadjimitsis, M.G., Retalis, A., Michaelides, S., Chrysoulakis, N., Toulios, L., Clayton, C.R.I. (2010). Atmospheric correction for satellite remotely sensed data intended for agricultural applications: impact on vegetation indices. Natural Hazards & Earth System Sciences, 10(1), 89–95. Im, J., Jensen, J.R., Jensen, R.R., Gladden, J., Waugh, J., Serrato, M. (2012). Vegetation cover analysis of hazardous waste sites in Utah and Arizona using hyperspectral remote sensing. Remote Sensing, 4(2), 327–353. Ishigaki, T., Yamada, M., Nagamori, M., Ono, Y., Inoue, Y. (2005). Estimation of methane emission from whole waste landfill site using correlation between flux and ground temperature. Environmental Geology, 48(7), 845–853. Jones, H.K., & Elgy, J. (1994). Remote sensing to assess landfill gas migration. Waste Management & Research, 12(4), 327– 337. Krook, J., Svensson, N., Eklund, M. (2012). Landfill mining: a critical review of two decades of research. Waste Management, 32(3), 513–520. Kwarteng, A.Y., & Al-Enezi, A. (2004). Assessment of Kuwait’s Al-Qurain landfill using remotely sensed data. Journal of Environmental Science and Health, Part A, 39(2), 351–364. Lewis, A.W., Yuen, S.T.S., Smith, A.J.R. (2003). Detection of gas leakage from landfills using infrared thermography– applicability and limitations. Waste Management & Research, 21(5), 436–447.

Environ Monit Assess Li, Y.Y., Zhang, H., Kainz, W. (2012). Monitoring patterns of urban heat islands of the fast-growing Shanghai metropolis, China: using time-series of Landsat TM/ETM+ data. International Journal of Applied Earth Observation and Geoinformation, 19, 127–138. Lyon, J.G. (1987). Use of maps, aerial photographs, and other remote sensor data for practical evaluations of hazardous waste sites. Photogrammetric Engineering & Remote Sensing, 53(5), 515–519. Markham, B.L., Storey, J.C., Williams, D.L., Irons, J.R. (2004). Landsat sensor performance: history and current status. IEEE Transactions on Geoscience and Remote Sensing, 42(12), 2691–2694. Noomen, M.F., Smith, K.L., Colls, J.J., Steven, M.D., Skidmore, A.K., Van Der Meer, F.D. (2008). Hyperspectral indices for detecting changes in canopy reflectance as a result of underground natural gas leakage. International Journal of Remote Sensing, 29(20), 5987–6008. Noomen, M.F., van der Werff, H., van der Meer, F.D. (2012). Spectral and spatial indicators of botanical changes caused by long-term hydrocarbon seepage. Ecological Informatics, 8, 55–64. Ottavianelli, G. (2007). Synthetic aperture radar remote sensing for landfill monitoring, Ph.D. Thesis, Cranfield University, United Kingdom. Ou, S.C., Chen, Y., Liou, K.N., Cosh, M., Brutsaert, W. (2002). Satellite remote sensing of land surface temperatures: application of the atmospheric correction method and split-window technique to data of ARM-SGP site. International Journal of Remote Sensing, 23(24), 5177– 5192. Paolini, L., Grings, F., Sobrino, J.A., Jim´enez Mu˜noz, J.C., Karszenbaum, H. (2006). Radiometric correction effects in Landsat multi-date/multi-sensor change detection studies. International Journal of Remote Sensing, 27(4), 685– 704. Philipson, W.R., Barnaba, E.M., Ingram, A., Williams, V.L. (1988). Land-cover monitoring with SPOT for landfill investigations. Photogrammetric Engineering & Remote Sensing, 54(2), 223–228. Pope, P., Eeckhout, E.V., Rofer, C. (1996). Waste site characterization through digital analysis of historical aerial photographs. Photogrammetric Engineering & Remote Sensing, 54(2), 223–228. Richter, R. (1990). A fast atmospheric correction algorithm applied to Landsat TM images. International Journal of Remote Sensing, 11(1), 159–166.

Richter, R. (1996). A spatially adaptive fast atmospheric correction algorithm. International Journal of Remote Sensing, 17(6), 1201–1214. Shaker, A., Faisal, K., El-Ashmawy, N., Yan, W.Y. (2010). Effectiveness of using remote sensing techniques in monitoring landfill sites using multi-temporal Landsat satellite data. Al-Azhar University Engineering Journal, 5(1), 542– 551. Silvestri, S., & Omri, M. (2008). A method for the remote sensing identification of uncontrolled landfills: formulation and validation. International Journal of Remote Sensing, 29(4), 975–989. Slonecker, T., Fisher, G.B., Aiello, D.P., Haack, B. (2010). Visible and infrared remote imaging of hazardous waste: a review. Remote Sensing, 2(11), 2474–2508. Stohr, C., Darmody, R.G., Frank, T.D., Elhance, A.P., Lunetta, R., Worthy, D., O’Connor-Shoresman, K. (1994). Classification of depressions in landfill covers using uncalibrated thermal-infrared imagery. Photogrammetric Engineering & Remote Sensing, 60(8), 1019–1028. Stohr, C., Su, W.J., DuMontelle, P.B., Griffin, R.A. (1987). Remote sensing investigations at a hazardous-waste landfill. Photogrammetric Engineering & Remote Sensing, 53(11), 1555–1563. Themelis, N.J., & Ulloa, P.A. (2007). Methane generation in landfills. Renewable Energy, 32(7), 1243–1257. Well, G.J., Graf, R.J., Forister, L.M. (1994). Investigations of hazardous waste sites using thermal IR and ground penetrating radar. Photogrammetric Engineering & Remote Sensing, 60(8), 999–1005. Yang, K., Zhou, X.-N., Yan, W.-A., Hang, D.-R., Steinmann, P. (2008). Landfills in Jiangsu Province, China, and potential threats for public health: Leachate appraisal and spatial analysis using geographic information system and remote sensing. Waste Management, 28(12), 2750–2757. Yes¸iller, N., Hanson, J.L., Liu, W.-L. (2005). Heat generation in municipal solid waste landfills. Journal of Geotechnical and Geoenvironmental Engineering, 131(11), 1330– 1344. Yes¸iller, N., Hanson, J.L., Oettle, N.K., Liu, W.-L. (2008). Thermal analysis of cover systems in municipal solid waste landfills. Journal of Geotechnical and Geoenvironmental Engineering, 134(11), 1655–1664. Zilioli, E., Gomarasca, M.A., Tomasoni, R. (1992). Application of terrestrial thermography to the detection of waste-disposal sites. Remote Sensing of Environment, 40(2), 153–160.

Analysis of multi-temporal landsat satellite images for monitoring land surface temperature of municipal solid waste disposal sites.

This studypresents a remote sensing application of using time series Landsat satellite images for monitoring the Trail Road and Nepean municipal solid...
10MB Sizes 0 Downloads 5 Views