Gap-Free LST Generation for MODIS/Terra LST Product Using a Random Forest-Based Reconstruction Method

Abstract

Land surface temperature (LST) is a crucial input parameter in the study of land surface water and energy budgets at local and global scales. Because of cloud obstruction, there are many gaps in thermal infrared remote sensing LST products. To fill these gaps, an improved LST reconstruction method for cloud-covered pixels was proposed by building a linking model for the moderate resolution imaging spectroradiometer (MODIS) LST with other surface variables with a random forest regression method. The accumulated solar radiation from sunrise to satellite overpass collected from the surface solar irradiance product of the Feng Yun-4A geostationary satellite was used to represent the impact of cloud cover on LST. With the proposed method, time-series gap-free LST products were generated for Chongqing City as an example. The visual assessment indicated that the reconstructed gap-free LST images can sufficiently capture the LST spatial pattern associated with surface topography and land cover conditions. Additionally, the validation with in situ observations revealed that the reconstructed cloud-covered LSTs have similar performance as the LSTs on clear-sky days, with the correlation coefficients of 0.92 and 0.89, respectively. The unbiased root mean squared error was 2.63 K. In general, the validation work confirmed the good performance of this approach and its good potential for regional application.

Spatial-Multitemporal Analysis of Heatwaves in Thailand: Discrepancies between In-Situ Air Temperature and Remote Sensing-Derived Land Surface Temperature

“…The quality of LST data is often limited by cloud contamination, as noted by [52], particularly in heatwave assessment studies that require daily data inputs rather than instantaneous or averaged values. In this study, LST prediction analysis was conducted using the RF model, yielding compelling results similar to those found in [57]. This success is attributed to the employment of both temporal and spatial variables as significant predictors.…”

“…However, widespread data gaps in LST product retrieval due to cloud cover affecting over 60% of global MODIS LST datasets. Therefore, to overcome the restriction of missing values resulting from clouds, a range of research has focused on developing reconstruction methods [57][58][59][60][61][62][63], such as a random forest (RF) machine learning algorithm to estimate uncompleted LST data. While LST correlates highly with T air , their magnitudes and temporal behaviors exhibit substantial heterogeneity [64].…”

“…Studies by [57,109] offer validation for our utilization of MODIS-LST data and RF models. Notably, our study places added emphasis on factors both temporal (such as Day of Year) and spatial (such as elevation), showcasing their paramount importance in predicting T air and LST with heightened accuracy.…”

Integrating Remote Sensing and Ground-Based Data for Enhanced Spatial–Temporal Analysis of Heatwaves: A Machine Learning Approach

“…Furthermore, studies by [47,48] have shown that thermal satellite-derived LST is a crucial parameter for identifying heatwaves and understanding the consequences of extreme heat. Additionally, LST plays a vital role as a key input in the study of land surface water and energy budgets at both local and global scales [49]. LST, recognized as one of the essential climate variables (ECVs) by the World Meteorological Organization (WMO), is a key indicator for both climate change and land surface processes.…”

Reconstruction of Gap-Free Land Surface Temperature at a 100 m Spatial Resolution from Multidimensional Data: A Case in Wuhan, China

“…Their method achieved a correlation coefficient (R 2 ) of 0.94 with a good visual performance. Xiao et al [ 32 ] constructed an RF regression model to reconstruct a time series gap-free LST in Chongqing City, and the results showed that R 2 value between the reconstructed LST and in situ LST measurements reached 0.89. Although machine learning models have good effects with respect to evaluation metrics and visual performance, their results are limited by the quality of the training dataset.…”