Worldwide continuous gap-filled MODIS land surface temperature dataset

Abstract

Satellite land surface temperature (LST) is vital for climatological and environmental studies. However, LST datasets are not continuous in time and space mainly due to cloud cover. Here we combine LST with Climate Forecast System Version 2 (CFSv2) modeled temperatures to derive a continuous gap filled global LST dataset at a spatial resolution of 1 km. Temporal Fourier analysis is used to derive the seasonality (climatology) on a pixel-by-pixel basis, for LST and CFSv2 temperatures. Gaps are filled by adding the CFSv2 temperature anomaly to climatological LST. The accuracy is evaluated in nine regions across the globe using cloud-free LST (mean values: R2 = 0.93, Root Mean Square Error (RMSE) = 2.7 °C, Mean Absolute Error (MAE) = 2.1 °C). The provided dataset contains day, night, and daily mean LST for the Eastern Mediterranean. We provide a Google Earth Engine code and a web app that generates gap filled LST in any part of the world, alongside a pixel-based evaluation of the data in terms of MAE, RMSE and Pearson’s r.

A global seamless 1 km resolution daily land surface temperature dataset (2003–2020)

“…In this study, the average RMSE is 1.83 and 1.28 • in the urban and surrounding areas for daytime and nighttime, respectively (Table 1). The gap-filled LSTs based on the data fusion method implemented on GEE (Shiff et al, 2021) were also evaluated at the global scale, but the mean RMSE is 2.7 • , higher than that of this study. The accuracies of other seamless LST datasets were generally evaluated based on a limited number of in situ LST observations (Zhang et al, 2019;Zhou et al, 2017), which are not exactly the same as satellite LSTs (Hong et al, 2021), and the evaluation in these studies are not directly comparable with our study.…”

“…The systematic differences between neighboring regions with the use of different gap-filling techniques in the hybrid method may lead to boundary effects (Li et al, 2018a). The data fusion method implemented on GEE (Shiff et al, 2021) directly filled the missing values in MODIS LST using the estimated LST values without consideration of the spatial continuity, which might lead to boundary effects. The seamless LST data produced by might also contain boundary effects since different regression methods were used to reconstruct the missing values according to the number of valid pixels.…”

“…Recently, Zhan et al (2021) pro-duced a global 1 km LST dataset (2003LST dataset ( -2019, but only a daily average of LST was included. Shiff et al (2021) developed a Google Earth Engine (GEE) code and a web app for generating 1 km gap-filled LST by using Climate Forecast System Version 2 (CFSv2) modeled air temperature and MODIS LST data, but they did not consider the naturally spatial variation in LST. A global daily minimum and maximum LST dataset with reasonable spatial pattern that can be used for a variety of studies and applications by scientists and practitioners such as city planners and water resources managers is still not available.…”

Global spatiotemporally continuous MODIS land surface temperature dataset

“…The LST at a specific time and date can be regarded as consisting of two parts: one is the long-term mean of the temperature at that time (climatological temperature), and the other is the deviation from that climatological temperature due to the weather (anomaly temperature) 17 . First, we calculate the climatological temperatures of the ideal, clear-sky satellite and the -ERA5-Land LSTs by using the reconstructed ideal clear-sky MODIS LST and ERA5-land LST data from 2002 to 2020: where is the climatological temperature of the ideal, clear-sky satellite on day i of the year, is the mean of the reconstructed ideal clear-sky MODIS LST on day i of each year from 2002 to 2020, is the climatological temperature of the ERA5-Land LST on day i of the year, and is the mean of the ERA5-Land LST on day i of each year from 2002 to 2020.…”

“…Field and in-situ measurements are not easily affected by weather or other factors, and LSTs can be obtained continuously over time. However, the usefulness of such data is poor when the field stations are sparsely distributed 17 . Most model reanalysis datasets, such as the Modern-Era Retrospective Analysis for Research and Applications (MERRA) dataset, National Center for Environmental Prediction (NCEP) products, and the European Center for Medium-Range Weather Forecasts (ECMWF) Reanalysis product ERA-Interim 18 , can provide spatiotemporally continuous LSTs at a global scale.…”