This study presents the new version of MODIS/Terra + Aqua Surface Radiation Daily/3-h downward shortwave radiation (DSR) (MCD18A1 V6.2) and photosynthetic active radiation (PAR) (MCD18A2 V6.2) product generated by MODIS adaptive processing system (MODAPS) using the latest version of the science algorithm developed by the NASA MODIS land science team. Key improvements in the new algorithm include using multiple bands covering visible, near-infrared, and shortwave infrared to enhance the capability of characterizing cloud optical characteristics, especially over snow-covered surfaces, and adopting linear interpolation for temporal scaling from instantaneous to 3-hourly retrievals. Comparative validation against MCD18 V6.1 and clouds and the Earth’s radiant energy system synoptic (CERES-SYN) demonstrates that V6.2 significantly improves accuracy at instantaneous, 3-hourly, and daily scales, particularly in snow-covered regions. The root mean square error (RMSE) (relative RMSE: rRMSE) of V6.2 reaches 101.9 W/m2 (18.8%) and 48.4 W/m2 (20.8%) for instantaneous DSR and PAR. The RMSE (rRMSE) reaches 29.9 W/m2 (16.9%) and 14.1 W/m2 (18.4%) for daily DSR and PAR, respectively. Aggregated to 100 km, V6.2 matches CERES-SYN accuracy using only polar-orbiting satellite data. This study also explores the potential for future improvement by integrating geostationary observations to enhance accuracy further.

Solar power is a critical source of renewable energy, offering significant potential to lower greenhouse gas emissions and mitigate climate change. However, the cloud induced-variability of solar radiation reaching the earth’s surface presents a challenge for integrating solar power into the grid (e.g., storage and backup management). The new generation of geostationary satellites such as GOES-16 has become an important data source for large-scale and high temporal frequency solar radiation forecasting. However, no machine-learning-ready dataset has integrated geostationary satellite data with fine-grained solar radiation information to support forecasting model development and benchmarking with consistent metrics. We present SolarCube, a new ML-ready benchmark dataset for solar radiation forecasting. SolarCube covers 19 study areas distributed over multiple continents: North America, South America, Asia, and Oceania. The dataset supports short (i.e., 30 minutes to 6 hours) and long-term (i.e., day-ahead or longer) solar radiation forecasting at both point-level (i.e., specific locations of monitoring stations) and area-level, by processing and integrating data from multiple sources, including geostationary satellite images, physics-derived solar radiation, and ground station observations from different monitoring networks over the globe. We also evaluated a set of forecasting models for point- and image-based time-series data to develop performance benchmarks under different testing scenarios. The dataset is available at https://doi.org/10.5281/zenodo.11498739. A Python library is available to conveniently generate different variations of the dataset based on user needs, along with baseline models at https://github.com/Ruohan-Li/SolarCube.

Land surface temperature (LST) data acquired from satellites are used extensively in studying climate variability. Many researchers have used Moderate Resolution Imaging Spectroradiometer (MODIS) LST to detect the temperature trend, however, its reliability has not been fully investigated. Using in-situ data acquired from 67 stations worldwide, this study examined the reliability of the detected temperature trends and investigated the associated influencing factors. The high-quality MODIS data have an RMSE of 2.44 K and 3.70 K at nighttime and daytime, respectively. However, its trend detection had an RMSE of 0.81 K/decade and 0.98 K/decade at nighttime and daytime, respectively. Clear-sky bias, quality control, LST estimation uncertainties, trend magnitude, and length of time were factors that influenced the detected trends. Filling cloud-covered areas in MODIS data may effectively reduce biases in trend detection.

Surface incident shortwave radiation (ISR) is an important component of the surface radiation budget. We refined the optimization method developed for polar-orbiting satellite data[1]and applied it to estimate ISR from the new generation geostationary Advanced Himawari Imager (AHI) onboard the Himawari-8/9 satellite and Advanced Baseline Imager (ABI) onboard the Geostationary Operational Environmental Satellite-R Series. Validation of the AHI ISR estimation at 2-km resolution showed an$R^{2}$of 0.93, bias of 0.52 W/m2, and RMSE of 106.52 W/m2for instantaneous estimates; an$R^{2}$of 0.95, bias of −0.12 W/m2, and RMSE of 22.49 W/m2for daily mean ISR; and a bias of −0.18 W/m2and RMSE of 7.72 W/m2for monthly mean ISR. Validation of the ABI ISR at 2-km spatial resolution showed an$R^{2}$value of 0.93, bias of 8.71 W/m2, and RMSE of 102.30 W/m2for instantaneous estimates; an$R^{2}$of 0.95, bias of −2.38 W/m2, and RMSE of 27.17 W/m2for daily mean ISR; and a bias of 1.40 W/m2and RMSE of 14.75 W/m2for monthly mean ISR. Our study also demonstrated that AHI and ABI observations have realized much better estimations for hourly and diurnal ISR than previous polar-orbiting satellite data because of their higher frequency of sampling on the atmospheric conditions.

Top-of-atmosphere (TOA) albedo is important for studying Earth's energy budget and climate dynamics. As no direct validation can be conducted to determine their accuracies, comprehensive intercomparisons are of significance for robust applications. In this study, five TOA albedo products over land surfaces have been intercompared, including three global TOA albedo products from Advanced Very High Resolution Radiometer (AVHRR) (TAL-AVHRR), Moderate Resolution Imaging Spectroradiometer (MODIS) (TAL-MODIS) and Clouds and the Earth's Radiant Energy System (CERES), the regional product from the Climate Monitoring Satellite Application Facility (CM SAF), and the harmonized product called Diagnosing Earth's Energy Pathways in the Climate system (DEEP-C). Results show that there is good consistence among the five products overall, especially after the year 2000. The differences among these products in the high-latitude regions are relatively larger (~0.2). The differences among TAL-AVHRR, TAL-MODIS and CERES are mostly less than 0.05 while the differences between TAL-AVHRR and DEEP-C before 2000 can be up to 0.1.

Surface shortwave net radiation (SSNR) and surface downward shortwave radiation (DSR) are the two surface shortwave radiation components in earth's radiation budget and the fundamental quantities of energy available at the earth's surface. Although several global radiation products from global circulation models, global reanalyses, and satellite observations have been released, their coarse spatial resolutions and low accuracies limit their application. In this paper, the Global LAnd Surface Satellite (GLASS) DSR product was generated from the Moderate Resolution Imaging Spectroradiometer top-of-atmosphere (TOA) spectral reflectance based on a direct-estimation method. First, the TOA reflectances were derived based on the atmospheric radiative transfer simulations under different solar/view geometries; second, a linear regression relationship between the TOA reflectance and SSNR was developed under various atmospheric conditions and surface properties for different solar/view geometries; third, the coefficients derived from the linear regression were used to compute the SSNR; and finally, the DSR was estimated using the SSNR estimates and broadband albedo at the surface. A 13-year (2003-2015) GLASS DSR product was generated at a 5-km spatial resolution and 1-day temporal resolution. Compared with the ground measurements collected from 525 stations from 2003 to 2005 around the world, the model-computed SSNR (DSR) had an overall bias of 8.82 (3.72) W/m2and a root mean square error of 28.83 (32.84) W/m2at the daily time scale. Moreover, the global land annual mean of the DSR was determined to be 184.8 W/m2with a standard deviation of 0.8 W/m2over a 13-year (2003-2015) period.

Traditional methods for estimating land-surface parameters from remotely sensed data generally focus on a single parameter with a specific spectral region, resulting in physical and spatiotemporal inconsistencies in current satellite products. We recently proposed a unified inversion scheme to estimate a suite of parameters simultaneously from both visible and near-infrared and thermal-infrared MODIS data. In this letter, we implemented this scheme to estimate six time-series parameters [leaf area index, fraction of absorbed photosynthetically active radiation, surface albedo, land-surface emissivity, land-surface temperature (LST), and upwelling longwave radiation (LWUP)] from the Visible Infrared Imaging Radiometer Suite (VIIRS) data. Several components of these schemes are refined, including the incorporation of a snow bidirectional reflectance distribution function model, determination of the best band combination, and better estimation of the snow-covered surface emissivity by accounting for the snow-cover fraction. Validation using the measurements at 12 sites of SURFRAD, CarboEuropeIP, and FLUXNET, and intercomparisons with MODIS and Global Land-Surface Satellite products, are carried out: the retrieved albedo, LST, and LWUP achieved accuracies (R€) of 0.77, 0.96, and 0.95, root mean square errors of 0.06, 2.9 K, and 18.3 W/m2, and biases of 0.01, 0.09 K, and -0.08 W/m2, respectively. The retrieved parameters can achieve comparable or higher accuracy than existing products, which indicates that the unified algorithm can be applied effectively to the VIIRS data with high physical and temporal consistency and accuracy.

The fraction of absorbed photosynthetically active radiation (FAPAR) is a critical input in many climate and ecological models. The accuracy of satellite FAPAR products directly influences estimates of ecosystem productivity and carbon stocks. The targeted accuracy of FAPAR products is 10% or 0.05 for many applications. However, most current FAPAR products do not meet such requirements, and further improvements are still needed. In this paper, a data fusion scheme based on the multiple resolution tree (MRT) approach is developed to integrate multiple satellite FAPAR estimates at site and regional scales. MRT was chosen because of the superior computational efficiency compared with other fusion methods. The fusion scheme removed the bias in FAPAR estimates and resulted in a 15% increase in the R2and 3% reduction in the root-mean-square error compared with the average of individual FAPAR estimates. The regional-scale fusion filled in the missing values, and provided spatially consistent FAPAR distributions at different resolutions. Overall, MRT can be used to efficiently and accurately generate spatially and temporally continuous FAPAR data across both site and regional scales.

Visible Infrared Imaging Radiometer Suite (VIIRS) land surface albedo (LSA) product is one of the routinely generated VIIRS environmental data records (EDR). A direct estimation approach based on radiative transfer simulation was developed to directly estimate instantaneous LSA value from VIIRS clear-sky top-of-atmosphere (TOA) reflectance. Comprehensive validations and evaluations were performed by comparing VIIRS LSA retrieval with ground measurements, MODIS 16-day mean albedo product, and Landsat high-resolution albedo estimation. Recently, an algorithm framework was also developed to produce a new gridded and temporal-filtered VIIRS LSA product, which is expected to provide spatially and temporally continuous LSA map with improved accuracy, and facilitate the analysis and application of global LSA data.

The availability of multiangular information from the NASA Multi-angle Imaging SpectroRadiometer (MISR) instrument provides an excellent opportunity for the characterization of surface anisotropy, which can be used for improving surface albedo estimation. However, the MISR data have been reported with large uncertainties and data gaps due to inaccurate aerosol estimation and/or cloud masking limiting its otherwise broader applications. To mitigate these issues, two approaches were proposed to estimate land surface albedo directly from surface reflectance (LSA_sfc) and Top-of-Atmosphere reflectance (LSA_toa), respectively. As a further development of the traditional albedo algorithms, this is the first attempt to simultaneously utilize multispectral and multiangular information in surface albedo estimation without any prior constraining information. Validations at AmeriFlux sites show that the proposed algorithms can achieve accuracies similar to that of the MISR product with respective bias and RMSE of 0.004 and 0.032 for LSA_sfc and 0.005 and 0.032 for LSA_toa algorithms. We found that the LSA_toa algorithm can significantly reduce data gaps and provide accurate surface albedo retrievals with two to three times more valid data than the current MISR product. In addition, these approaches can be easily applied to other optical sensors to produce accurate and gap-free clear-sky surface albedo estimations. The results of this paper also highlight the importance of having two to three simultaneous observations with sufficient angular sampling, which can improve albedo accuracy and reduce data gaps.

High-spatial resolution data of the top-ofatmosphere (TOA) reflected shortwave radiation flux are needed to understand the effects of local-scale anthropogenic and natural processes on earth's radiation budget. A previous study developed an algorithm to estimate the TOA instantaneous shortwave component from polar-orbiting Moderate Resolution Imaging Spectroradiometer (MODIS) data. This paper presents a temporal scaling approach to predict daily values of TOA reflected shortwave radiation flux from temporally sparse MODIS observations. Radiative transfer simulation and statistical regression are used to establish the relationship between daily shortwave flux and MODIS spectral reflectance. A comparison between the Terra and Aqua combined MODIS computed data and the Clouds and the Earth's Radiant Energy System SYN1deg product of 1° regional daily shortwave fluxes have a bias of 3.8 W/m2with a root-mean-square error (RMSE) of 13.3 W/m2using data from 2009 over eight subsets across various latitudes. Comparing the regional monthly shortwave fluxes reduces the RMSE to 6.9 W/m2. The longer the averaging period the lower the uncertainty is self-explanatory.

Unlike instantaneous albedo, daily albedo of land surfaces is currently not routinely generated from satellite data, although it is a key input parameter for calculating daily shortwave radiation budget. This paper presents a novel approach to directly retrieve daily mean values of land surface broadband blue-sky albedo from Visible Infrared Imaging Radiometer Suite clear-sky data of apparent reflectance, with the assumption that the atmospheric conditions of the satellite overpass time can represent their daily values. Training data were simulated by atmospheric radiative transfer models, with surface spectra and bidirectional reflectance distribution function data as inputs for four aerosol types and a range of aerosol loadings. Sensitivity analysis was conducted to study the effects of cloud coverage, aerosol, and surface types on retrieval accuracy. Two years of measurements at six Surface Radiation Budget Network and eight Greenland Climate Network stations were used for algorithm validation. Daily albedo of snow-free surfaces can be retrieved with very high accuracy. By excluding far off-nadir observations of snow surfaces, the overall accuracy of retrieving daily albedo has a bias of 0.003 and a root-mean-square error of 0.055.

Surface shortwave net radiation (SSNR) is a key component of the surface radiation budget. In this paper, we refined a direct estimation approach to retrieve daily SSNR estimates from combined Terra and Aqua Moderate Resolution Imaging Spectroradiometer (MODIS) data. The retrieved MODIS SSNR estimates were validated against measurements at seven stations of the Surface Radiation Budget Network. We also compared the MODIS retrievals with three existing SSNR products: the Clouds and the Earth's Radiant Energy System (CERES) products, the North American Regional Reanalysis (NARR) data, and the ERA-Interim reanalysis data from the European Centre for Medium-Range Weather Forecasts. MODIS data at 1 km were upscaled to mitigate the mismatch between site measurements and satellite retrievals. Among the four data sets, the aggregated MODIS retrievals agreed best with in situ measurements, with a root-mean-square error (rmse) of 23.1 W/m2and a negative bias of 6.7 W/m2. The CERES products have a slightly larger rmse of 24.2 W/m2and a positive bias of 7.6 W/m2. Both reanalysis data (NARR and ERA-Interim) overestimate daily SSNR and have much larger uncertainties. Monthly satellite SSNR data are more accurate than daily values, and the scaling issue in validating monthly MODIS SSNR retrievals is also less prominent. Averaged with a window size of 23 km, the two MODIS sensors can estimate monthly SSNR with an rmse error of 11.6 W/m2, representing an improvement of 2.4 W/m2over the CERES products.

Maps of high-resolution surface shortwave net radiation (SSNR) are important for resolving differences in the surface energy budget at the ecosystem level. The maps can also bridge the gap between existing coarse-resolution SSNR products and point-based field measurements. This study presents a modified hybrid method to estimate both instantaneous and daily SSNR from Landsat data. SSNR values are directly linked to Landsat top-of-atmosphere reflectance by extensive radiative transfer simulation. Regression coefficients are pre-calculated and stored in a look-up table (LUT). Atmospheric water vapor is a key parameter affecting SSNR, and three methods of treating water vapor are evaluated in this study. Comparison between Landsat retrievals and field measurements at six AmeriFlux sites shows that the hybrid method with water vapor as a dimension of LUT can estimate SSNR with a root mean square error of 77.5 W/m2(instantaneous) and 36.1 W/m2(daily). The method of water vapor correction produces similar results. However, a generic LUT that covers all levels of water vapor results in much larger errors.

Land surface albedo is a key factor in climate change and land surface modeling studies, which affects the surface radiation budget. Many satellite albedo products have been generated during the last several decades. However, due to the problems resulting from the sensor characteristics (spectral bands, spatial and temporal resolutions, etc.) and/or the retrieving procedures, surface albedo estimations from different satellite sensors are inconsistent and often contain gaps, which limit their applications. Many approaches have been developed to generate the complete albedo data set; however, most of them suffer from either the persistent systematic bias of relying on only one data set or the problem of subpixel heterogeneity. In this paper, a data fusion method is prototyped using multiresolution tree (MRT) models to develop spatially and temporally continuous albedo maps from different satellite albedo/reflectance data sets. Data from the Multiangle Imaging Spectroradiometer (MISR), Moderate Resolution Imaging Spectroradiometer (MODIS), and Landsat Thematic Mapper/Enhanced Thematic Mapper Plus are used as examples, at a study area in the north central United States mostly covered by crop, grass, and forest, from June to September 2005. Results show that the MRT data fusion method is capable of integrating the three satellite data sets at different spatial resolutions to fill the gaps and to reduce the inconsistencies between different products. The validation results indicate that the uncertainties of the three satellite products have been reduced significantly through the data fusion procedure. Further efforts are needed to evaluate and improve the current algorithm over other locations, time periods, and land cover types.

The monitoring of top-of-atmosphere (TOA) reflectance time series provides useful information regarding the long-term degradation of satellite sensors. For a precise assessment of sensor degradation, the TOA reflectance time series is usually corrected for surface and atmospheric anisotropy by using bidirectional reflectance models so that the angular effects do not compromise the trend estimates. However, the models sometimes fail to correct the angular effects, particularly for spectral bands that exhibit a large seasonal oscillation due to atmospheric variability. This paper investigates the use of time series algorithms to identify both the angular effects and the atmospheric variability simultaneously in the time domain using their periodical patterns within the time series. Two nonstationary time series algorithms were tested with the Landsat 5 Thematic Mapper time series data acquired over two pseudoinvariant desert sites, the Sonoran and Libyan Deserts, to compute a precise long-term trend of the time series by removing the seasonal variability. The trending results of the time series algorithms were compared to those of the original TOA reflectance time series and those normalized by a widely used bidirectional-reflectance-distribution-function model. The time series results showed an effective removal of seasonal oscillation, caused by angular and atmospheric effects, producing trending results that have a higher statistical significance than other approaches.

Surface albedo is one of the key radiation parameters required for modeling of the Earth's energy budget. The future Geostationary Operational Environmental Satellite-R Series (GOES-R) Advanced Baseline Imager (ABI) will provide the observations in several shortwave spectral bands together with both high spatial and temporal resolutions which will carry much angular information for estimating instantaneous surface albedo and bi-directional reflectance. According to these advanced sensor characteristics, we propose an improved algorithm that retrieves surface albedo and aerosol optical depth (AOD) simultaneously. To prototype this algorithm, satellite observations with the similar spectral bands and spatial resolution acquired by MODIS are used. Results show a good agreement between retrieved albedo values and ground measurements from SURFRAD.

The leaf area index (LAI) is a critical variable used to characterize the terrestrial ecosystem and model land surface processes. Remote sensing is an ideal tool for mapping the LAI. However, the quality of current satellite LAI products does not meet the requirements of the user community in terms of estimation accuracy and data consistency. One way to address these issues is to develop LAI integration algorithms that incorporate existing multiple LAI products and prior knowledge. This paper presents a new data integration method based on empirical orthogonal function (EOF) analysis. The proposed EOF integration algorithm can be operated on both fine and coarse spatial resolution to accommodate the problems arising from a large volume of data. Two runs of multivariate EOF analysis are proposed to address the issue of incompatible temporal resolutions among different data sets. Comparisons with high-spatial-resolution LAI reference maps at 12 sites over North America show that the proposed method can improve LAI product accuracy. After data integration, R2increases from 0.75 to 0.81 and the root mean square error (rmse) decreases from 1.04 to 0.71 over moderate-resolution imaging spectroradiometer (MODIS) products. The improvement of R2and rmse over Carbon cYcle and Change in Land Observational Products from an Ensemble of Satellites (CYCLOPES) products is not as significant as that over MODIS products. However, the use of a combination of multiple data sources reduces the bias of the LAI estimate from MODIS's 0.3 and CYCLOPES's -0.2 to -0.1.

Although long-term consistent observations of LAI play a significant role in understanding the Earth system and its change, there is no reliable, consistent and accurate LAI product available due to estimation errors, data gaps and inconsistency among data. This paper presents a method, based on the multiresolution tree (MRT), to integrate satellite LAI data with different resolutions. The results of integrating MODIS and MISR LAI show MRT has the capability of filling data gaps, reducing measurement errors and can also generate consistent results across scales. When compared with optimal interpolation (OI), MRT produces similar results but with significantly improved efficiency.

Gaps and uncertainties within current Moderate Resolution Imaging Spectroradiometer (MODIS) land products limit their application in understanding land surface dynamics. This paper employs a revised iterative strategy based on multi-channel singular spectrum analysis (MSSA) to fill the gaps in MODIS land products. Our results on MODIS leaf area index (LAI) products show MSSA has the ability to use both spatial and temporal information to fill gaps within MODIS LAI products to produce spatially and temporally smoother datasets.

A previous algorithm developed to estimate photosynthetically active radiation (PAR) from satellite observations is improved and extended to retrieve surface albedo as well. The revised algorithm was validated with GOES-12 data over the North America continent. We used measurements from six SURFRAD and six FLUXNET sites in 2003 to validate the GOES PAR and albedo retrievals. The results show that GOES instantaneous PAR retrievals have a mean deviation of 175.0 mu mol s-1m-2(20.4% in relative value), a bias of 15.4 mu mol s-1m-2(2.0% in relative value), and a correlation coefficient of 0.881 for the sites under all-sky conditions. GOES daytime averaged PAR has a mean deviation of 91.3 NASA mol s-1m-2(11.8% in relative value), a bias of 14.1 mu mol s-1m-2(2.0% in relative value), and a correlation coefficient of 0.931 for all the sites under all sky conditions. Compared with the albedo measurements at the six FLUXNET sites, GOES albedo retrievals in 0.4-0.7 mum have a mean deviation of 0.017, a bias of 0.002 and a correlation coefficient of 0.773, while the corresponding MODIS visible albedo product has a mean deviation of 0.021, a bias of -0.028 and a correlation coefficient of 0.600.