Downscaling Land Surface Temperature Research Articles

Downscaling Land Surface Temperature from MODIS Dataset with Random Forest Approach over Alpine Vegetated Areas

In this study, we evaluated three different downscaling approaches to enhance spatial resolution of thermal imagery over Alpine vegetated areas. Due to the topographical and land-cover complexity and to the sparse distribution of meteorological stations in the region, the remotely-sensed land surface temperature (LST) at regional scale is of major area of interest for environmental applications. Even though the Moderate Resolution Imaging Spectroradiometer (MODIS) LST fills the gap regarding high temporal resolution and length of the time-series, its spatial resolution is not adequate for mountainous areas. Given this limitation, random forest algorithm for downscaling LST to 250 m spatial resolution was evaluated. This study exploits daily MODIS LST with a spatial resolution of 1 km to obtain sub-pixel information at 250 m spatial resolution. The nonlinear relationship between coarse resolution MODIS LST (CR) and fine resolution (FR) explanatory variables was performed by building three different models including: (i) all pixels (BM), (ii) only pixels with more than 90% of vegetation content (EM1) and (iii) only pixels with 75% threshold of homogeneity for vegetated land-cover classes (EM2). We considered normalized difference vegetation index (NDVI) and digital elevation model (DEM) as predictors. The performances of the thermal downscaling methods were evaluated by the Root Mean Square Error (RMSE) and the Mean Absolute Error (MAE) between the downscaled dataset and Landsat LST. Validation indicated that the error values for vegetation fraction (EM1, EM2) were smaller than for basic modelling (BM). BM model determined averaged RMSE of 2.3 K and MAE of 1.8 K. Enhanced methods (EM1 and EM2) gave slightly better results yielding 2.2 K and 1.7 K for RMSE and MAE, respectively. In contrast to the EMs, BM showed a reduction of 22% and 18% of RMSE and MAE respectively with regard to Landsat and the original MODIS LST. Despite some limitations, mainly due to cloud contamination effect and coarse resolution pixel heterogeneity, random forest downscaling exhibits a large potential for producing improved LST maps.

Downscaling MODIS land surface temperature over a heterogeneous area: An investigation of machine learning techniques, feature selection, and impacts of mixed pixels

Remotely sensed Land Surface Temperature (LST) is of paramount importance in numerous environmental applications. Although, coarse spatial resolution sensors provide frequent LST measurements, their applicability is rather limited for many applications. Downscaling methods are therefore applied to improve the spatial resolution of LST products. A number of Machine Learning (ML) methods have already been used in the LST downscaling studies. Nevertheless, the literature lacks a suitable inter-comparison of different ML methods, as well as the impact of the feature selection process on downscaling results. This study aims at comparing downscaled LST from 1 km daily MODIS LST product (MOD11A1) to 240 m using Random Forest Regression (RFR), Support Vector Regression (SVR), Extreme Learning Machine (ELM) and Temperature Sharpening (TsHARP) approaches with 11 predictor variables at a heterogeneous area in different seasons. In addition, by implementing a feature selection with the Support Vector Machine-Recursive Feature Elimination (SVM-RFE) method, the most important variables were selected and used as inputs of the models. The results were evaluated against the LST derived from Landsat-8 thermal imageries using a split window method, showing that all the ML methods perform well in LST downscaling (with an average RMSE = 2.5 and MAE = 1.74) with marginal differences, outperforming the TsHARP method (RMSE∼ 3.02, MAE∼ 2.19). Among all the methods, ELM required the least computational effort, and when it was combined with SVM-RFE, general efficiency of the downscaling procedure was increased substantially.

Applicability of Downscaling Land Surface Temperature by Using Normalized Difference Sand Index

Land surface temperature (LST) in coarse spatial resolution derived from thermal infrared satellite images has limited use in many remote sensing applications. In this study, we improve our previous approach (multiple remote-sensing index approach of random forest) to downscale LST derived from Landsat 8 and MODIS in an arid oasis - desert ecotone of Zhangye city by designing a normalized difference sand index (NDSI), by the removal of land cover datasets and by the input of SAVI, NDBI and NDWI to downscale LST. Our result demonstrates that NDSI can determine the characteristic of the desert region, and that the distribution of downscaled LST matches those of oasis-desert ecosystems. Relative to the ground observation of HiWATER, our approach also produces relatively satisfactory downscaling results at July 21 (2013), with R2 and root-mean-square error of 0.99 and 1.25 K, respectively. Compared with other methods, our approach demonstrates higher accuracy and minimization of the retrieved Landsat 8 LST in the desert region. Optimal availability occurs in the vegetation and desert region. Our approach is suitable to LST downscaling in all seasons, especially in spring and summer. The model can further be applied in middle-high and middle-low spatial resolutions. The usefulness of the model is relatively satisfactory in the humid region (Nanjing city) but less accurate in the arid region.

Downscaling land surface temperature based on random forest algorithm

Downscaling land surface temperature based on random forest algorithm

Modelling land surface temperature using gamma test coupled wavelet neural network

The climate change has made adverse effects on land surface temperature for many regions of the world. Several climatic studies focused on different downscaling techniques for climatological parameters of different regions. For statistical downscaling of any hydrological parameters, conventional Neural Network Models were used in common. However, it seems that in any modeling study, uncertainty is a vital aspect when making any predictions about the performance. In this paper, Gamma Test is performed to determine the data length selection for training to minimize the uncertainty in model development. Another measure to improve the data quality and model development are wavelet transforms. Hence, Gamma Test with Wavelet decomposed Feedforward Neural Network (GT-WNN) model is developed and tested for downscaled land surface temperature of Patna Urban, Bihar. The results of GT-WNN model are compared with GT-FFNN and conventional Feedforward Neural Network (FFNN) model. The effectiveness of the developed models is illustrated by Root Mean Square Error and Coefficient of Correlation. Results showed that GT-WNN outperformed the GT-FFNN and conventional FFNN in downscaling the land surface temperature. The land surface temperature is forecasted for a period of 2015-2044 with GT-WNN model for Patna Urban in Bihar. In addition, the significance of the probable changes in the land surface temperature is also found through Mann-Kendall (M-K) Test for Summer, Winter, Monsoon and Post Monsoon seasons. Results showed an increasing surface temperature trend for summer and winter seasons and no significant trend for monsoon and post monsoon season over the study area for the period between 2015 and 2044. Overall, the M-K test analysis for the annual data shows an increasing trend in the land surface temperature of Patna Urban.

Spatial Downscaling Land Surface Temperature based on Copula

Spatial Downscaling Land Surface Temperature based on Copula

Downscaling Land Surface Temperature in an Arid Area by Using Multiple Remote Sensing Indices with Random Forest Regression

Many downscaling algorithms have been proposed to address the issue of coarse-resolution land surface temperature (LST) derived from available satellite-borne sensors. However, few studies have focused on improving LST downscaling in arid regions (especially in deserts) because of inaccurate remote sensing LST products. In this study, LST was downscaled by a random forest model between LST and multiple remote sensing indices (such as soil-adjusted vegetation index, normalized multi-band drought index, modified normalized difference water index, and normalized difference building index) in an arid region with an oasis–desert ecotone. The proposed downscaling approach, which involves the selection of remote sensing indices, was evaluated using LST derived from the MODIS LST product of Zhangye City in Heihe Basin. The spatial resolution of MODIS LST was downscaled from 1 km to 500 m. Results of visual and quantitative analyses show that the distribution of downscaled LST matched that of the oasis and desert ecosystem. The lowest (approximately 22 °C) and highest temperatures (higher than 37 °C) were detected in the middle oasis and desert regions, respectively. Furthermore, the proposed approach achieves relatively satisfactory downscaling results, with coefficient of determination and root mean square error of 0.84 and 2.42 °C, respectively. The proposed approach shows higher accuracy and minimization of the MODIS LST in the desert region compared with other methods. Optimal availability occurs in the vegetated region during summer and autumn. In addition, the approach is also efficient and reliable for LST downscaling of Landsat images. Future tasks include reliable LST downscaling in challenging regions.

Downscaling Land Surface Temperature in Complex Regions by Using Multiple Scale Factors with Adaptive Thresholds.

Many downscaling algorithms have been proposed to address the issue of coarse-resolution land surface temperature (LST) derived from available satellite-borne sensors. However, few studies have focused on improving LST downscaling in urban areas with several mixed surface types. In this study, LST was downscaled by a multiple linear regression model between LST and multiple scale factors in mixed areas with three or four surface types. The correlation coefficients (CCs) between LST and the scale factors were used to assess the importance of the scale factors within a moving window. CC thresholds determined which factors participated in the fitting of the regression equation. The proposed downscaling approach, which involves an adaptive selection of the scale factors, was evaluated using the LST derived from four Landsat 8 thermal imageries of Nanjing City in different seasons. Results of the visual and quantitative analyses show that the proposed approach achieves relatively satisfactory downscaling results on 11 August, with coefficient of determination and root-mean-square error of 0.87 and 1.13 °C, respectively. Relative to other approaches, our approach shows the similar accuracy and the availability in all seasons. The best (worst) availability occurred in the region of vegetation (water). Thus, the approach is an efficient and reliable LST downscaling method. Future tasks include reliable LST downscaling in challenging regions and the application of our model in middle and low spatial resolutions.

Improving the Downscaling of Diurnal Land Surface Temperatures Using the Annual Cycle Parameters as Disaggregation Kernels

The downscaling of geostationary diurnal thermal data can ease the lack of land surface temperature (LST) datasets that combine high spatial and temporal resolution. However, the downscaling of diurnal LST data is more demanding than single scenes. This is because the spatiotemporal interrelationships of the original LST data have to be preserved and accurately reproduced by the downscaled LST (DLST) data. To that end, LST disaggregation kernels/predictors that provide information about the spatial distribution of LST during different times of a day can prove especially useful. Such LST predictors are the LST Annual Cycle Parameters (ACPs). In this work, multitemporal ACPs are employed for downscaling daytime and nighttime ~4 km geostationary LST from SEVIRI (Spinning Enhanced Visible and Infrared Imager) down to 1 km. The overall goal is to assess if the use of the ACPs can improve the estimation of the diurnal range of DLST (daytime DLST minus nighttime DLST). The evaluation is performed by comparing the DLST diurnal range maps with reference data from MODIS (Moderate Imaging Spectroradiometer) and also with data retrieved from a modified version of the TsHARP (Thermal Sharpening) algorithm. The results suggest that the ACPs increase the downscaling performance, improve the estimation of diurnal DLST range and produce more accurate spatial patterns.

Downscaling land surface temperatures at regional scales with random forest regression

Abstract Environmental monitoring with satellite data is facilitated by frequent observations at a fine spatial scale. As land surface temperature (LST) is one environmental key variable, we implemented a random forest (RF) regression approach to increase the spatial resolution of LST maps from ~ 1 km, routinely available in daily repetition from the Moderate Resolution Imaging Spectroradiometer (MODIS), to ~ 250 m. LST was downscaled based on its relationship to topographic variables derived from digital elevation data of the Shuttle Radar Topography Mission (SRTM), land cover data (MODIS product MCD12Q1), and surface reflectances in the visible red and near-infrared, which both are provided with the MODIS/Terra daily product MOD09GQ at ~ 250 m resolution. The approach was tested for a complex landscape in the Eastern Mediterranean, the Jordan River Region, with LST fields from aggregated Landsat-7 ETM + and MODIS (MODIS/Terra LST product MOD11A1) data; as reference at the finer scale, we used Landsat-7 derived LST data. For the ideal-case scenario with both degraded and reference values from the same sensor (Landsat-7 ETM +), root mean square errors (RMSE) of downscaled LSTs ranged from 1.02 K to 1.43 K for six different acquisition dates. When compared to the widely-adopted and in parallel applied TsHARP sharpening method, that is based on the relationship between NDVI and LST, downscaling accuracy with RF improved up to 19%. Applying the RF approach to MODIS LST products yielded RMSEs from 1.41 K to 1.92 K, whereas the TsHARP method and also a uniform disaggregation by resampling provided only slightly worse results. For the real MODIS LST product, downscaling with RF was affected by lower thermal contrasts in the image data that hindered an adequate training to reproduce temperature variations at the finer scale of ~ 250 m. In this context, we assume the LST product of the Visible Infrared Imaging Radiometer Suite (VIIRS) instrument (as a successor of MODIS onboard the SUOMI NPP platform) to be a better candidate for downscaling, as it provides a spatial resolution of ~ 750 m. One advantage of the RF approach is that predictor datasets can easily be adapted to data availability. In an extended RF approach with all Landsat spectral bands, downscaling results for formerly aggregated Landsat data improved distinctly and now ranged from 0.98 K to 1.33 K. This approach is also promising for the downscaling of real MODIS or VIIRS LST data as it may be combined with already available reflective data fusion models that are able to blend Landsat data and sensor data with a coarse spatial resolution (given spectral bandwidths corresponding to Landsat) to generate temporally dense synthetic Landsat time series.

Estimation of Air Temperatures for the Urban Agglomeration of Athens with the Use of Satellite Data

Changing air temperature trends within urban regions deserve careful monitoring as they may reflect modifications in the thermal environment, including the development of an urban heat island. Air temperature fields need to be dense in order the state of the thermal environment to be adequately assessed; yet in most cases, the networks of ground measuring stations are sparse. This paper attempts to define the relationship between downscaled land surface temperature (LST) at resolution 1 km as deduced from MSG-SEVIRI satellite images, and air temperature (Tair) in the urban agglomeration of Athens, for varying land cover types. Polynomial regression and artificial neural networks are used to estimate Tair from LST at a particular time, whereas the LST values for several hours before are also used. In this way, the “memory” of the surface materials is taken into consideration, practically reflecting the thermal inertia associated with land cover. For urban stations, an average R2 of 0.85 and an RMSE of 1.0-1.2˚C was achieved for the majority of the examined time period, an indication of both the capacity of the methodology to define Tair fields in the area under consideration as well as of the fact that LST is the controlling parameter for Tair. The parametric relations as extracted from the above methodology are in principle applicable for a specific station, as they depend on the land cover and the associated land surface characteristics. They may be also used for stations in areas with similar land cover and in the same climatic zone.

Regression-Kriging Technique to Downscale Satellite-Derived Land Surface Temperature in Heterogeneous Agricultural Landscape

The study introduces regression-kriging technique to downscale land surface temperature (LST) over a heterogeneous landscape of India and compares it with other models. All models are initially tested on aggregated 960 m Landsat LST and downscaled to 240 m ( ${\bf RMSE} = {\bf 0}.{\bf 45}^{\circ} {\rm{C}}$ ) and 120 m ( ${\bf RMSE} = {\bf 0}.{\bf 68}^{\circ} {\rm{C}}$ ) resolution. Finally, the MODIS LST is downscaled to 250 m ( ${\bf RMSE} = {\bf 0}.{\bf 7}^{\circ} {\rm{C}}$ ) resolution. The proposed model outperformed compared to other models. The qualitative assessment reveals appearance of box-like structure is not seen in the proposed model-derived LST due to reconstruction of residual .

Downscaling of thermal images over urban areas using the land surface temperature–impervious percentage relationship

Intensive expansion and densification of urban areas decreases environmental quality and quality of urban life as exemplified by the urban heat island effect. For this reason, thermal information is becoming an increasingly important data source for integration in urban studies. It is expected that future spaceborne thermal sensors will provide data at appropriate spatial and temporal resolutions for urban studies. Until they become operational, research has to rely on downscaling algorithms increasing the spatial resolution of relatively coarse resolution thermal images albeit having a high temporal resolution. Existing downscaling algorithms, however, have been developed for sharpening images over rural and natural areas, resulting in large errors when applied to urban areas. The objective of this study is to adapt the DisTrad method for downscaling land surface temperature (LST) over urban areas using the relationship between LST and impervious percentage. The proposed approach is evaluated by sharpening aggregated LST derived from Landsat 7 ETM+ imagery collected over the city of Dublin on May 24th 2001. The new approach shows improved downscaling results over urban areas for all evaluated resolutions, especially in an environment with mixed land cover. The adapted DisTrad approach was most successful at a resolution of 480m, resulting in a correlation of R2=0.84 with an observed image at the same resolution. Furthermore, sharpening using the adapted DisTrad approach was able to preserve the spatial autocorrelation present in urban environments. The unmixing performance of the adapted DisTrad approach improves with decreasing resolution due to the fact that the functional relationship between LST and impervious percentage was defined at coarse resolutions.

Downscaling Land Surface Temperature in an Urban Area: A Case Study for Hamburg, Germany

Monitoring of (surface) urban heat islands (UHI) is possible through satellite remote sensing of the land surface temperature (LST). Previous UHI studies are based on medium and high spatial resolution images, which are in the best-case scenario available about four times per day. This is not adequate for monitoring diurnal UHI development. High temporal resolution LST data (a few measurements per hour) over a whole city can be acquired by instruments onboard geostationary satellites. In northern Germany, geostationary LST data are available in pixels sized 3,300 by 6,700 m. For UHI monitoring, this resolution is too coarse, it should be comparable instead to the width of a building block: usually not more than 100 m. Thus, an LST downscaling is proposed that enhances the spatial resolution by a factor of about 2,000, which is much higher than in any previous study. The case study presented here (Hamburg, Germany) yields promising results. The latter, available every 15 min in 100 m spatial resolution, showed a high explained variance (R2: 0.71) and a relatively low root mean square error (RMSE: 2.2 K). For lower resolutions the downscaling scheme performs even better (R2: 0.80, RMSE: 1.8 K for 500 m; R2: 0.82, RMSE: 1.6 K for 1,000 m).

Downscaling land surface temperatures with multi-spectral and multi-resolution images

Land surface temperature (LST) plays an important role in many fields. However, the limited spatial resolution of current thermal sensors impedes the utilization of LSTs. Based on a theoretical framework of thermal sharpening, this report presents an Enhanced Generalized Theoretical Framework (EGTF) to downscale LSTs using multi-spectral (MS) and multi-resolution images. MS proxy-sharpening and LST downscaling are combined under EGTF. Simulated images upscaled from Enhanced Thematic Mapper Plus (ETM+) and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data are produced for indirect validations. Validation of MS proxy-sharpening shows that EGTF is better than the Gram-Schmidt (GS) and the Principle Component (PC) methods, yielding a lower root mean square error (RMSE) and ERGAS (erreur relative globale adimensionnelle de synthèse) and, thus, maintaining higher spectral similarity. For LST downscaling, validations show that EGTF has a higher accuracy than the Unmixing-Based Image Fusion (UBIF) method and indicate that the proxy-sharpening process improves the accuracy of downscaled LSTs. Further discussions regarding the selection of the moving-window size (MWS) demonstrate that the MWS could be determined by the range in a semi-variance analysis of scaling factor images.

Downscaling land surface temperature for urban heat island diurnal cycle analysis

Urban heat island (UHI) is a phenomenon of high spatial and temporal variabilities. It can develop during night or daytime. UHI monitoring is possible through thermal satellite remote sensing of land surface temperature (LST). LST over large areas (size of a city) can be measured at high temporal resolution merely from instruments on-board geostationary satellites. These can cover the diurnal cycle as they provide data even every 5 min (SEVIRI rapid scanning). Sensors on-board the geostationary satellites have, however, poor spatial resolution. Using high spatial resolution is in many regions most important because LST is a spatially inhomogeneous parameter especially in urban areas. UHI characteristics are correlated with the land cover and micro-relief parameters. These are often available in a higher spatial resolution (e.g. NDVI and EVI). Thus, we employed them to enhance the spatial resolution of the SEVIRI LST over central Europe – using moving window analysis – to 1000 m spatial resolution and temporal resolution of 15 min. For each SEVIRI pixel a multiple regression was run on the low resolution data. Regression equation was then used on the high resolution data in order to estimate LST of high spatial and temporal resolutions. The validation over urban areas showed that the downscaled SEVIRI LST is comparable with the MODIS LST with an average root mean square error of 2.5 K. The obtained results make possible to analyse the diurnal cycle of UHI.