ON SURFACE TEMPERATURE AND TEMPORAL CHANGES IN TEMPERATURE A CASE STUDY OF AJMER DISTRICT , RAJASTHAN

An attempt has been made in this research to analyze temporal variations in surface temperature in Ajmer District Rajasthan. The research is carried out to assess the relationship between the land surface temperatures (LST) and land cover (LC) changes both in quantitative and qualitative ways in Ajmer District area using Landsat TM/ETM+ data over the period 1989 to 2013.in this period we used three temporal TM/ETM data 1989, 2001 and 2013. Remote sensing of Land surface temperature (LST) has traditionally used the Normalized Difference Vegetation Index (NDVI) as the indicator of vegetation abundance to estimate the land surface temperature (LST)–vegetation relationship. Unsupervised classification methods have been taken to prepare the LC map. LST is derived from the thermal band of Landsat TM/ETM+ using the calibration of spectral radiance and emissivity correction of remote sensing. NDVI is derived from the NIR & RED Band using image enhancement technique (Indices). Arc-GIS have been utilized for data visualization. This procedure allowed analyzing whether LULC classes match LST classes. However, the results of such overlaying are hard to interpret. LST and LULC maps of these areas give the understanding on how the classes and corresponding LST have changed from one date to the other. Another option is to collect statistical data. it was impossible to calculate linear regression between LULC map and LST map. A solution to that matter is to use Normalized Vegetation Index (NDVI) instead of LULC classification result.


INTRODUCTION
Land surface temperature is an important factor in global changes in estimating radiation budget in heat balance studies and as a control for climate mode.It forms the basic forth application of water and energy balance modeling's(van de grinds et , al 1993).The knowledge's of the surface temperature is important to a range of issue and themes in earth science such as global environment changes, human-environment interaction and more specifically to urban climatology.Surface temperature can be acquired directly from ground measurement but these measurements cannot be generalized since surface temperature varies considerably from one type of surface to another on the other hand satellite measurement are not subject to arbitrary extrapolation .They are area averaged rather than point values and can be acquired a regular temporal basis (Goita and Royer, 1997).
Rise in urbanization (includes both natural increase and migration) and rapid industrialization is one of the major causes of climate changes all over the word.Today the most imperative problem in urban areas is increasing surface temperature because of dramatic alteration of the natural surface as natural vegetation is removed and replace by nonevaporating, non-transpiring surface (e.g.stone, metal, concrete etc.)

STUDY AREA
Ajmer is situated in the geographic centre of Rajasthan and lies about 135 km's south-west of the state capital, Jaipur.Ajmer covering geographical area of 8,481 sq.km. it is surrounded by the Aravalli mountain ranges in the center of Rajasthan state, surrounded by three hill of Aravalli ranges at an average of 486.0 meter above MSL.First we did radiometric correction of the dataset using ENVI 4.8 software in which we used radio metric correction tool for correction the dataset.After this we did Geo-metric correction of all the dataset using Re-projection tool of the ERDAS 9.2

Prepared LU/LC Map Using Masking
It is very helpful to analyze research data and achieve our objective.We did 1 st level classification on the study area.In this we classified the data into major categories.We found that some mixed categories like that: wasteland and settlement, forestland and agriculture etc. mixed in each other.For the solution of this problem we did masking of the mixed categories.We used ERDAS modeler to perform this function.
We have taken unsupervised classification data and in the vector object taken shape file which is masked over the classified data.In the function object taken "either" function for achieve this objective for example "EITHER 6 IF ($n2_settlement_2001=1) OR $n1_recode_ajmer_2001_classi OTHERWISE".

Prepared NDVI
NDVI is to use show the relationship between Land surface temperature and Land use/ Land cover using the linear regression correlation method.Because of the nature of LU/LC it was not possible to calculate the correlation between LU/LC and Land Surface temperature.A solution to that matter is to use Normalized Vegetation Index (NDVI) instead of LULC classification result.

General formula NDVI = (Infra -Red) / (Infra. + Red)
The value is between +1 (vigor) ~ -1 (stress) In other words, NDVI value is related to LULC class.Linear regression between NDVI and LST has been calculated. Where: T is degrees Kelvin CVR1 is cell value as radiance Ε is emissivity (typically 0.95 After all this, Land Surface Temperature Dataset ready to use it.

Random Sampling of the study area
In order to establish the relationship between LST and corresponding land use/cover type, 70 sample points were selected randomly from different land use/cover types in the study area.About 10-10 samples were selected from each land use/cover types, e.g., Forestland, Agriculture, Fallow land, Water body, industry, Settlement and Wasteland.The thermal characteristics of these 70 locations are studied across the all temporal dataset.

Relationship between LULC and LST Using Statistic
The average temperature of each land use/cover category for each temporal LST was calculated using the formula of simple arithmetic mean.

Relationship between LULC and LST by the linear regression between LST & LU/LC (NDVI)
NDVI values are related to LULC class.Linear regression between NDVI and LST has been calculated using the random sampling values of different temporal LST and their corresponding NDVI values it is helpful to investigate the spatial pattern of LST.

Visual Interpretation between LST and LU/LC
A focus area with significant LULC transition was chosen for each Temporal Dataset.LST and LULC maps of these areas give the understanding on how the classes and corresponding LST slices have changed from one date to the other.Arc-GIS have been utilized due to its better capabilities for data visualization.

Comparison done by the Visual Interpretation
Same procedure follows for the comparison by the visual interpretation.

Comparison done by the Statistic
The average temperature of each Land use/Land cover category for each temporal LST maps was calculated using the formula of simple arithmetic mean.After this compared each land use/Land cover category minimum, maximum and average temperature of the other temporal LST maps category and calculate the minimum, maximum and average temperature changes.

RESULT & DISCUSSION
Within the frame work of research objectives, the following result and discussion were deriving.First of all, we deals with the Land use/ land cover result 1989, 2001 and 2013 respectively.

Land Use/Land Cover Classification Result and Accuracy Assessment
Differences between classification results and reference data are known as classification errors.One of the most common methods of classification accuracy assessment is error matrix or confusion matrix.This matrix contains a category comparison of relationship between known, ground-truth data and classification results for the same category.

Relationship on the basis of Linear Regression method of 1989, 2001 and 2013 year
Due to the nature of LULC map, it was impossible to calculate linear regression between LULC map and LST map.A solution is to that matter is to use Normalized Vegetation Index (NDVI) instead of LULC classification result.
We performed linear regression method between LST map and their corresponding NDVI using the corresponding random sampling values of 1989, 2001 and 2013 year datasets of each LU/LC Categories with their corresponding NDVI values.We observed that, in year 1989 all these LU/LC categories are having negative correlation with LST and NDVI except water body.In year 2001, we also observed that, all these LU/LC       In order to relate Land Surface temperature with other Land Surface Temperature, both LST map and other corresponding LST maps were imported to Arc-GIS.This procedure allowed analyzing whether LULC classes match LST classes.However, the results of such overlaying are hard to interpret.Instead, a focus area with significant LULC transition was chosen for each category.In 1989 to 2001 through the visual interpretation we observed that the temperature of both LST and their corresponding LST increased in such a manner and 2001 to 2013 visual interpretation we observed that the temperature of both LST and their corresponding LST increases and some portion of the corresponding LST is decrease in such a manner.This is only way to understand the changes in temperature between different temporal datasets.It means that trough this we got only qualitative information not quantitative information of the different temporal datasets.

RECOMMENDATIONS:
 Coarser resolution satellite dataset did not give much information of the surface temperature related to heterogeneous and complex urban areas.Hence, aerial thermal remote sensing (with high resolution spatial resolution) is are visible option to help in realistic assessment of surface temperature and characteristic of heterogeneous and complex urban areas related to it.Added to that aerial thermal remote sensing has temporal flexibility times to give the most normalized characteristic of the area and thus reduce the apparent bias due to solar and atmosphere distortion.


Given the finding that surface temperature is negative correlated to vegetation; it is advisable to afforest regions that are currently available for planning, especially in highly built up areas and barren land.This will in turn reduced the surface temperature of the region and thus to a certain extent influence the micro-climate of the region.


Further research could include utilizing more effective classification method, such as neural network or support vector machine and decision tree control classification method to reduce the classification errors and their vector machines to reduce the classification errors and their influence on the final result.

REFERENCES:
A Figure 2.1: -Location map of Ajmer District

3. 5
Converting Landsat TM and ETM+ thermal bands to temperature It is possible to convert these DNs to degrees Kelvin using a two (or optionally three) step process.The first step is to convert the DNs to radiance values using the bias and gain values.a)Spectral Radiance Scaling Method the formula used in this process is as follows: cell value as radiance QCAL = digital number LMINλ = spectral radiance scales to QCALMIN LMAXλ = spectral radiance scales to QCALMAX QCALMIN = the minimum quantized calibrated pixel value (typically = 1) QCALMAX= the maximum quantized calibrated pixel value.b) Convert Radiance to Kelvin the formula to convert radiance to temperature without atmospheric correction is:

Figure
Figure 4.1: -Visual Interpretation of LST and their corresponding LU/LC of 1989, 2001 and 2013 year.

FigureFigure 4
Figure 4.5:-Correlation between NDVI and LST of the Agriculture of the 1989 year negative correlation with LST and NDVI except water body.In year 2013, above describe condition occur.These entire LU/LC categories are having negative correlation with LST and NDVI except water body.It means that each category like; Fallow land, Wasteland, Settlement, industry severity are increased then the NDVI values of this corresponding category is decreases it means temperature of these categories increase.In case of Forestland and Agriculture categories Quality are increases than the NDVI values of this corresponding category is increased it means temperature of these categories decreases.In case of water body positive correlation occur with LST and NDVI maps.When the NDVI values are increases then the temperature values also increases.In the water NDVI values are increases it means quality of the water is decreases because pure water having -0 NDVI it shows that when the NDVI values of the water body increases then the quality of the water is decrease and result is that the temperature of the Water body increase is increase then the NDVI values of this corresponding category is decrease it means temperature is decrease.So, water body one and only categories which is having positive correlation with LST and NDVI maps.
Figure 4.9: -Show the 1989 and 2001 LST Maps and Changes in Temperature Range Figure: -2001 LST Figure:-2013 LST Figure 4.10: -2001 and 2013 LST Maps and Changes in Temperature Range We observed that land Surface Temperature of overall study area is decrease -2.6299 ˚C from 2001 to 2013 year.Region is that in year 2001 rainfall is low.Temperature is going on high on this year.
7: -Average Temperature of 1989, 2013 & Changes in Temperature from 1989 to 2013.Finally we observed that overall study area temperature increase since 1989 to 2013 year.All the LU/LC categories temperature is increases except industry.4.3.2Land Surface Temperature changes between different temporal Land Surface Temperature data show by the Visual interpretation.

Figure 4 .
Figure 4.12: -Visually shows the changes in temperature since 1989 to 2013 year .K.Tiwari, M.Kudrat and S.K. Bhan, "Vegetation Cover Classification in Sariska National Park and Surroundings" Journal of Indian Society of Remote Sensing, Vol.18, No.3, 1990, Pages 43-51 Abdussalam Ahmed Ibrahim, Using Remote Sensing Technique (NDVI) for Monitoring Vegetation Degradation in semi-arid lands and its relationship to precipitation: Case

.5 Change Rate of LU/LC Since 2001 to 2013 Table 4.2: -Change rate of Land use/ Land cover from 2001to 2013 4.1.6 Change Rate of LU/LC from 1989 to 2013 Year Table 4.3: -Change rate of Land use/ Land cover from 1989 to 2013
These changes in Land use and Land cover category effect the Land Surface Temperature pattern on the study area from 1989 to 2013.

4.2 Land Surface Temperature and LU/LC Relationship 4.2.1 Using the Visual Interpretation Technique
A focus area with significant LULC transition was chosen for each Temporal Dataset.LST and LULC maps of these areas give the understanding on how the classes and corresponding LST slices have changed from one date to the other.
When LU/LC of the area changes then the temperature of the area also changes.Water body has a lowest temperature in the study area, fallow land has a highest temperature in the study area and Industry and settlement have a moderate temperature.

Year 2013 Figure 4.4:-Land surface temperature of different LU/LC of Landsat8 band 10(2013)
We found the Sequenced of the LU/LC categories which are responsible to increment in temperature of the LST maps in uniform manner show; Water body, Forestland, Agriculture, Industry, Settlement, Wasteland and Fallow land.This sequence of LU/LC Categories is responsible for sustainable increment in temperature.It means that Land Surface Temperature distribution pattern is depending on the LU/LC categories.Where are the water body, Forestland and Agriculture situated temperature of the area is lower in comparison of other LU/LC categories.

Table 4 .
We observed that land Surface Temperature of overall study area is increase 5.5825 ˚C from 1989 to 2001 year.

2001 Ave. 2013 Ave.
Land Surface Temperature of overall study area of Different land use and land cover categories is changed which shown in table 4.6.

.1.3 LST Changes from 1989 to 2013
We observed that land Surface Temperature of overall study area is increase 2.9526 Degree in Celsius From 1989 to 2013 year.