عنوان مقاله [English]
Biophysical properties of vegetation, temperature and surface moisture are key parameters to control and evaluate the physical and chemical processes of the land surface. Biophysical properties can be used to monitor various applications such as urban heat island, climate change and drought. Access to timely information and awareness of the changes in land’s biophysical properties are required in integrated management and sustainable development. Erath surface biophysical properties can be successful in being retrieved using different types of remote sensing data. From 1970s, remote sensing data provides unique information in surveying dynamic phenomena. Remote sensing imagery provides repetitive data of wide distant area.
Remote sensing sensors collect Earth surface data at different part of electromagnetic spectrum, i.e. optic, thermal and microwave. As a result, the same phenomenon may provide different responses depending on the radiation’s wavelength. These responses are complementary and the joint use of them offers more reliable information. That is the reason, multi-sensor approaches gain more attention. Multi-spectral optical sensors such as Landsat have been widely used in earth surface studies; however, their applications are limited mainly in the presence of smoke, fog and clouds. In contrary, radar sensors (e.g. Synthetic Aperture Radar, SAR) operate well even in cloudy sky. SAR sensors are sensitive to the moisture content and structure (shape, direction, roughness) of the surface. Therefore, the main purpose of this study is to evaluate the efficiency of radar bands for extracting surface biophysical properties.
Materials and methods
For the purpose of comprehensive study, three different areas with different types of land cover were considered as the study areas. The first study area, located in the east of Ardebil, encompasses bare land. The second study area is located in the southeast of Ardebil with agricultural land use. The third study area in Mazandaran province is around Noor city. Land cover is dense natural forest.
Landsat-8 and Sentinel-1 satellite images dated on 2019 were acquired. The Landsat-8 image has already been geo-referenced with UTM coordinates system, zone 39. The coordinate system for Sentinel images is WGS84 ellipsoid. We used the GRD product which has VV and VH polarizations.
In this study, pre-processing steps were done to prepare images including atmospheric correction (Landsat images) and geometric (Sentinel images). FLAASH algorithm has been used for atmospheric correction. Next, spectral indices were computed from Landsat visible and infrared bands to represent surface biophysical properties. Single-channel algorithm is used to calculate surface temperature. Multiple linear regression was applied to model surface biophysical properties by the help of Sentinel polarimetric bands. Finally, based on these models, surface biophysical properties maps were driven.
Results and discussion
In this study, 18 spectral indices were extracted from Landsat image. It should be noted that some of these indices are normalized and some are not normalized, so for all indices to be comparable the values of all indices were set to zero and one normal.
Case study 1
In the first study area, radar’s backscattering values showed more significant relationships with LST, NDBI and IPVI indices, while there were weak relationships among radar’s backscattering values with MNDWI, GDVI and SR indices. High coefficient of determination between radar responses and LST values could be justified by the effect of soil moisture on soil temperature and radar backscattering, as well. That’s why, radar responses can predict LST values.
Case study 2
The investigation of the relationship among spectral indices and radar bands shows that radar bands have high potential to extract biophysical properties in this region. Among 18 spectral indices, EVI, MTVI1 and MTVI2 indices were highly correlated with the radar bands. The LST, SGI and SR indices showed the weakest correlations with radar bands. This indicates, LST, SGI and SR could not be predicted by backscattering values in agricultural land.
Case study 3
In the third study area, MNDWI showed a high correlation with radar responses. MNDWI index was first developed to study the amount of water available to represent vegetation health. Radar bands are also highly sensitive to moisture content. Therefore, it is not surprising that a high correlation between the radar bands and MNDWI index were reported for the third study area, as it is a high moisture forest area. The lowest correlation was observed between radar bands with NDBI spectral index with correlation coefficient of 0.418. The reason for this low value is the nature of the study area, because the study area covers with dense vegetation and the NDBI index has been developed to extract the built-up area.
Remote sensing technology provides valuable information in recognition of patterns and changes of the biophysical properties of the earth surface. Optical data have good spatial, spectral and radiometric resolution and have been used in a variety of applications. However, optical data is not available in all seasons because the presence of smoke, fog, clouds limits their availability. In contrast to optical data, SAR sensors have the ability to acquire data in all weather conditions. Therefore, the main objective of this study was to investigate the capability of radar data to extract biophysical properties of the land surface. The results showed that radar bands have a high capability to extract surface biophysical properties, so radar data can be considered as a good alternative especially when optical data is not available. Considering the proper relation of spectral indices with the targets’ responses in radar bands, the results of this study can further be used in different environmental applications such as heat island, evapotranspiration and coastline extraction.
Based on the findings of the current study, it is recommended that future researches investigate the efficiency of full polarization images in comparison with existing spectral indices to extract biophysical properties of the earth's surface. The full polarized radar images also allow the calculation of radar indices based on the degree of backscattering values in different polarimetric bands. In addition, given the low saturation level of spectral indices (in comparison with radar, responses) and the loss of sensitivity of these indices to phenomena changes, it is highly helpful to investigate the relationship of biophysical properties with radar bands in such a situation.