Investigating correlation LST and Vegetation Indices using Landsat images for the warmest month: A Case study of Iași County [307102]

Investigating correlation LST and Vegetation Indices using Landsat images for the warmest month: A [anonimizat], Florian STATESCU

ABSTRACT: In this paper is investigating correlation between land surface temperature and vegetation indices ([anonimizat], [anonimizat] – MSAVI) [anonimizat], for study area. Iași county is considered as study area in this research. Study Area is geographically situated on latitude 46°48'N to 47°35'N and longitude 26°29'E to 28°07'E. Land surface temperature (LST) is used to determine the temperature distribution at the different scale: local, regional and global. Also it’s used in climate change models in particular. Obtaining LST and using them in different analysis is important to determine the problem associated with the environment. A Vegetation Indices (VI) is a spectral transformation what allow reliable spatial and temporal intercomparisons of terrestrial photosynthetic activity and canopy structural variations. It can be observed that VI do not have a [anonimizat]. A lot of factors: [anonimizat], [anonimizat], [anonimizat]. Landsat 5 TM, Landsat 7 ETM and Landsat-8 OLI, all data were used in this study for modeling. Landsat images was taken for august 1994, 2006 and 2016. Preprocessing of Landsat 5/7/8 data stage represent that process that prepare images for subsequent analysis that attempts to compensate/correct for systematic errors. It was observed that the "mean" parameter for LST increased from 1994 to 2016 at approximately 5 ° C. [anonimizat]-[anonimizat]. Many researches indicated that between LST and VI is a linear relationship. It is noted that the R2 [anonimizat] 1994 (i.g.R2= 0.72 [anonimizat]) in 2016 (i.g.R2= 0.58 [anonimizat]). [anonimizat], [anonimizat].

KEY WORDS: [anonimizat], [anonimizat]’s climate since the industrial era starts. Some of the changes occurs due to natural phenomena and anthropogenic activities such as: [anonimizat] (LC-LU), [anonimizat], [anonimizat] (Penny and Kealhofer, 2005).

[anonimizat]-derived (like Landsat-5/7/8) surface temperature data have been utilized for regional and local climate analysis of different scale (Carlson et al. 1977).

Nowadays land surface temperature (LST), [anonimizat] (local/regional/global). Also it’s used in climate change models in particular. LST, [anonimizat]: [anonimizat], [anonimizat]. Obtaining LST and using them in different analysis is important to determine the problem associated with the environment (Orhan et al. 2014).

A Vegetation Indices (VI) is a spectral transformation what allow reliable spatial and temporal intercomparisons of terrestrial photosynthetic activity and canopy structural variations (Huete et al., 2000). It can be observed that VI do not have a standard universal value, research having often shown different results. A lot of factors, according to Bannari, (sensor calibration, brightness, sensor viewing conditions, soil moisture, solar illumination geometry, color and the atmosphere) could seriously affect vegetation indices. Furthermore, in a heterogeneous environment, the study of VI becomes more complex because there is a mixture of vegetation and other ground elements in the pixels. Nevertheless, the choice of a VI to the detriment to another, for whatever application, is really delicate to make. Each environment has its own characteristics and each index is a good indicator of green vegetation in its own right (Bannari et al., 1995). In the field of remote sensing applications, scientists have developed VI for qualitatively and quantitatively evaluating vegetative covers using different spectral measurements.

Many researches indicated that between LST and VI is a linear relationship. According to Carlson, this negative correlation between LST-NDVI is valuable for urban climate studies and another circles of science (Carlson et al., 1994).

In this paper is investigating correlation LST – VI in the warmest month (august) in 1994, 2006 and 2016. As based date was used Landsat 5/7/8 images.

2.Data and methods

2.1. Study Area

Iași county is considered as study area in this research. Study Area is geographically situated on latitude 46°48'N to 47°35'N and longitude 26°29'E to 28°07'E. Neighborings Iași county are Botosani to the north, Neamt to the west, Vaslui to the south and Rep. of Moldova to the east.

2.2. Landsat data

In this paper is investigated corelation LST – V.I. for Iași county using remotely sensed data. The present is focused on the thermal remote sensing application of Landsat satellite data. Landsat 5 TM, Landsat 7 ETM and Landsat-8 OLI, all data were used in this study for modeling. Details of the used data are given in Table 1. Landsat data was offered free by USGS.

Table 1. Landsat data

2.3.Data processing

2.3.1.Image preprocessing

The goal of image preprocessing is to make all of the remote sensing data similar so that imagies can be considered to be taken in same environmental conditions with the same sensors (Hall et al., 1991). To fill the gaps with Landsat 7 images, a specialized toolbox of ArcMap 10.1 was used.

Preprocessing of Landsat 5/7/8 data stage represent that process that prepare images for subsequent analysis that attempts to compensate/correct for systematic errors. The images are subjected to several corrections like radiometric and atmospheric. In order to be used by some image processing aplication, the 90m resolution TIRs bands were resampled to "fit" to 30m spatial. Resampling is used to "keep" the original pixel values in the resampled images nearest neighbor (miningeology.blogspot).

Radiometric correction is done to decrease/correct errors in the digital numbers (DNs) of images. The process of eliminating the effects of the atmosphere to obtain surface reflectance values represents the atmospheric correction. Atmospheric correction can significantly enhance the interpretability and usage of images. Perfectly for this process would be to have knowledge about the aerosol properties and the atmospheric conditions at the time the remote sensing data was acquired (miningeology.blogspot).

2.3.2. Vegetation Indices (VI)

2.3.2.1.Normalized Difference Vegetation Index (NDVI)

This index is a numerical indicator, according to Rouse, that uses the visible and near-infrared (NIR) bands of the electromagnetic spectrum. Was adopted to analyze remote sensing measurements and assess whether the target being observed contains live green vegetation or not (John Rouse, 1973). The NDVI algorithm subtracts the red (RED) reflectance values from the near-infrared (NIR) and divides it by the sum of that bands (John Rouse, 1973).

NDVI=

2.3.2.2.Enhanced Vegetation Index (EVI)

EVI is composed of: “C” – coefficients for atmospheric resistance, “L” – to adjust for canopy background, values from the blue band (B). These enrichments allow for index calculation as a ratio between the R and NIR values, while reducing: atmospheric noise, background noise and saturation in most cases (https://landsat.usgs.gov). The enhanced vegetation index (EVI) is an 'optimized' vegetation index projected to enhance the vegetation signal with improved sensitivity in high biomass regions and „upgraded“ vegetation monitoring through a decoupling of the baldachin background signal and a minimize in atmosphere influences (Huete et al., 2002).

EVI = G * ((NIR – R) / (NIR + C1 * R – C2 * B + L))

Where, G=2.5, C1=6, C2=7.5 and L=1

2.3.2.3. Modified Soil Adjusted Vegetation Index (MSAVI)

MSAVI is determined as a ratio between the R and NIR values with an inductive L function applied to maximize reduction of soil effects on the vegetation signal (https://landsat.usgs.gov). MSAVI is soil adjusted vegetation indices that look for to address some of the limitation of NDVI when applied to areas with a high degree of exposed soil surface. Qi et al. in 1994 developed the MSAVI.

MSAVI=(2 * NIR + 1 – sqrt ((2 * NIR + 1)2 – 8 * (NIR – R))) / 2

2.3.3.Method Selection for Estimating LST

To determine the land surface temperature from the Landsat 5/7 thermal infrared band data, must convert digital numbers (DNs) of sensors to spectral radiance using equation (Chander and Groeneveld, 2009).

Lλ= ×(Qcal-Qcalmin)+Lminl

Where:

• Lλ = the cell value as radiance – expressed in (W/(m2sr μm))

• Qcal = the quantized calibrated digital number

• Qcalmin = the minimum quantized calibrated pixel value

• Qcalmax = the maximum quantized calibrated pixel value

• LMINλ = the spectral radiance scales for Qcalmin

• LMAXλ = the spectral radiance scales for Qcalmax

To estimate the LST from thermal infrared (TIR) band data of Landsat-8 OLI, digital numbers (DN) of sensors were transformed to spectral radiance using equation (Barsi et al., 2014).

Lλ =ML×Qcal+AL-Qi

Where:

•𝑀𝐿= rescaling factor

•𝑄cal = the Band 10/11 image

•𝐴𝐿=the band-specific additive rescaling factor

• 𝑂𝑖 = the correction for Band 10/11

Spectral radiance is converted to brightness temperature by assuming the earth of surface is a black body (Chander et al, 2009; Coll et al, 2010):

BT=

Where:

• BT = the brightness temperature

• Lλ = the cell value as radiance

• K1, K2 = Calibration constant of Landsat 8 calibration

There are algorithms applied to transfer Brightness Temperature (BT) for LST:

LST=BT/[1+(λ(BT)/ρ) × ln(LSE)]

Where: , λ(11.5µm) = the wavelength of emitted radiance: ρ = h × c/σ = 1.438 × 10−2 mK, σ is the Stefan–Boltzmann constant, h is Planck’s constant, c is the velocity of light, and LSE=the land surface emissivity. NDVI was used to extract LSE, which is an adjustable parameter in correcting LST in the next step. Values of LSE were calculated based on PV (Jiménez-Muñoz et al. 2014).

LSE=0.004PV+0.986

Where, Pv is the proportion of vegetation, based on a normalized NDVI value of each pixel.

PV=()2

3.Results and discussions

Table 2 shows LST data statistics and table 3 shows VI statistics. It is noticeable that parmeter “mean” for LST has increased from 1994 to 2016 about 5oC. In 05.08.1994 the maximum temperature was 37 oC, and in 2016 was 37 oC. Analyzing the data from Table 3, it can be assumed that the built-up area increased for the Iasi county, while the area occupied by dense vegetation has decreased.

Figure 1 shows LST and VI maps, also presents the charts for LST-VI correlation.

Table 2.LST data statistics

Table 3.VI data statistics

Figure 1.LST, VI maps and LST-VI correlation

R2

4. Conclusions

In this paper was studying relationship LST-VI using Landsat 5/7/8 data, that were preprocessing. It is noticeable that parameter “mean” for LST has increased from 1994 to 2016 about 5oC. Analyzing the data from VI, it can be assumed that the built-up area increased for the Iasi county, while the area occupied by dense vegetation has decreased.

R2

In conclusion, these correlation can be used to study vegetation health, drought damage, and areas where Urban Heat Island can occur.

References

A. Bannari , D. Morin , F. Bonn, A. R. Huete, 1995, A review of vegetation indices Journal Remote Sensing Reviews Volume 13, 1995 – Issue 1-2,

A. Huete, K. Didan, T. Miura, E.P. Rodriguez, X. Gao, L.G. Ferreira, 2002, Overview of the radiometric and biophysical performance of the MODIS vegetation indices, Remote Sensing of Environment 83 (2002) pp. 195 – 213

Barsi J. C., Tu Q., Davidson E. H., (2014). General approach for in vivo recovery of cell type-specific effector gene sets. Genome Res. Vol. 24, pp. 860-868.

Carlson, B.E., A.A. Lacis, and W.B. Rossow, 1994: Belt-zone variations in the Jovian cloud structure. J. Geophys. Res., 99, pp. 14623-14658.

Carlson, J. Augustine, and F. E. Boland, 1977, Potential application of satellite temperature measurements in the analysis of land use over urban areas, Bulletin of the American Meteorological Society, vol. 58, pp. 1301–1303.

Chander G., Markham B. L., Helder D. L. (2009). Summary of current radiometric calibration coefficients for Landsat MSS, TM, ETM+, and EO-1 ALI sensors. Remote Sensing of Environment, vol. 113, pp. 893–903.

Coll M., Piroddi C., Steenbeek J., Kaschner K., Ben Rais Lasram F., Aguzzi J., et al. (2010). The biodiversity of the Mediterranean Sea: estimates, patterns and threats. PLoS ONE 5:e11842.

D. Penny, L Kealhofer, 2005, Microfossil evidence of land-use intensification in north Thailand. Journal of Archaeological Science 32 (1), pp. 69-82

G. Chander, D.P. Groeneveld, 2009, Intra-annual NDVI validation of the Landsat 5 TM radiometric calibration, International Journal of Remote Sensing, Volume:30 Issue:6, pp.1621-1628

Hall F.G., Strebel D.E., Nickeson J.E., Goetz S.J., (1991). Radiometric Rectification Toward a common Radiometric response among Multidate, Multisensor images. Remote sense Enviroment, Vol. 35, pp. 11-27.

Huete A., Didan K., Miura T., Rodriguez P., (2000). Validation of the MODIS vegetation indices over a global set of test sites: preliminary result. University of Arizona.

J. Qi, A. Chehbouni, A.R. Huete,Y.H. Kerr, S. Sorooshian, (1994). A modified soil adjusted vegetation index. Remote Sensing of Environment, vol.48, no.2, pp.119–126.

Jiménez-Muñoz J.C., Sobrino J., Skokovic D., Mattar C., Cristóbal J., (2014). Land Surface Temperature Retrieval Methods From Landsat-8 Thermal Infrared Sensor Data, IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, Vol. 11, No. 10, pp. 1840-1843.

NASA/GSFCT Type II Report, Greenbelt, MD, USA.

Osman Orhan, Semih Ekercin, Filiz Dadaser-Celik, (2014). Use of Landsat Land Surface Temperature and Vegetation Indices for Monitoring Drought in the Salt Lake Basin Area, Turkey. The Scientific World Journal, Vol.(2014), pp.1-11.

Rouse, J. W. (1973) Monitoring the vernal advancement and retrogradation of natural vegetation.

***https://landsat.usgs.gov

***miningeology.blogspot