To mitigate global climate change, the replacement of conventional coal-fired power generation with clean energy sources such as photovoltaics (PV) has become a key strategy. However, solar power output is highly variable because it depends on the amount of sunlight reaching the ground, referred to as Downward Shortwave Radiation (DSR). Accurately forecasting DSR in the short term is therefore critical for the stable integration of large-scale PV systems into urban power grids. Existing methods typically adopt a retrieve-then-forecast paradigm (first deriving physical products, then forecasting them), which performs poorly under rapidly varying atmospheric conditions. We propose a new forecast-then-retrieve framework for short-term DSR prediction based on geostationary satellite observations. Unlike conventional approaches, our method first forecasts the full-spectrum L1B radiances from Himawari-8 for the next 3 hours, and then retrieves DSR values from the predicted radiances. To support this framework, we design AtmoNet, a multimodal network that takes the past 3 hours of Himawari-8 L1B radiances as input and captures spatiotemporal patterns of reflectance, water vapor, and longwave emission. Experiments show that this approach improves DSR prediction accuracy by up to 6.8% in the 1–3 hour forecast window. Moreover, in direct L1B forecasting tasks, AtmoNet outperforms leading models, including the state-of-the-art Swin Transformer, particularly in capturing complex, moisture-driven phenomena such as localized convection and evaporation. By enabling accurate and scalable DSR forecasting, this work supports the stable integration of renewable energy into power grids and has the potential to contribute to global efforts to reduce carbon emissions.

Lunar crater is crucial for estimating the age of the Moon and investigating evolution of both the Moon and the solar system. However, the bottom of the crater is usually uneven, resulting in automatic detection difficult to obtain a complete impact crater. This paper proposes an automated lunar crater detection algorithm, which is robust to the complicated variations of impact craters, based on digital elevation model (DEM) data. The lowest points on the DEM were extracted as the potential centers of craters. The rim of craters was first detected by a max curvature detection method and then completed by the watershed algorithm. The algorithm was applied to the lunar mare and highland and compared with the crater database obtained by Robbins. The results show the algorithm could detect numerous small and complicated types of craters. In the lunar highlands, the algorithm detected 74% of documented impact craters, the newly detected impact craters accounted for 82.79%, and the overall reliability exceeded 81%. In the lunar mare, the algorithm identified over 50% of established craters, more than 66% of detected craters were new, and the overall reliability was greater than 56%. The algorithm calculated the structural attributes and obtained the true rim. Only the DEM data is required, the algorithm is portable to other planets such as Mars.

High-resolution fractional tree cover mapping allows for the representation of spatial details of tree distribution and contributes to forest ecosystem monitoring and modeling. However, the estimation of tree cover at high resolution is challenging, especially in sparsely tree-covered and mountainous areas interfered with background signals (e.g., soil, grass, and water) and terrain shadow. This article presents a fractional tree cover estimation algorithm from Landsat time-series observations at 30-m resolution, with the effects of background and terrain shadow reduced. The seasonal profiles of vegetation index and surface reflectance were constructed from multiyear Landsat data. Three phenological metrics were extracted from the seasonal profiles as input features for tree cover estimation. The training data were collected from the European Space Agency (ESA) WorldCover product and used to calibrate a feedforward neural network model to predict cover fractions. This algorithm extracted the tree cover of major forest types, including boreal, temperate, tropical dry, and moist forests. It also captured the sparse tree cover in areas containing mixtures of tree crowns and grass, bare soil, crop, impervious surface, and water. In two dense montane forest areas, the tree cover fractions on shady and sunny slopes were estimated consistently. The estimation results were evaluated through the reference samples generated from Google submeter-resolution image classification. The values of$R$-squared ($R^{2}$), root-mean-square error (RMSE), and mean absolute error (MAE) reached 0.78, 15.71%, and 11.09%, respectively. The proposed algorithm can be applied to monitor tree cover in spatially fragmented forest areas and sparse forests.

Leaf chlorophyll content (LCC) is an important plant physiological trait and is critical for accurate modeling of vegetation photosynthesis over time and space. To date, there is still a lack of a global long time-series dataset of LCC. In this study, we developed an algorithm to retrieve global LCC from MODIS surface reflectance data from 2000–2020. An essential requirement for generating LCC time series is to capture its seasonal dynamics. This issue was addressed by using a matrix system with two pairs of vegetation indices to minimize the impacts of leaf area index and canopy non-photosynthetic material on LCC estimation in different seasons. The matrix system algorithm was applied to Landsat data and MODIS data, respectively. The validation based on Landsat data and ground measurements reveals the algorithm has the ability to catch the seasonal variations of LCC in different plant functional types, and the MODIS-derived LCC shows good agreement with Landsat-upscaled LCC (R2=0.77, RMSE=6.9 μg/cm2). The global 8-day LCC data at 500-m resolution in 2000–2020 was generated using the matrix system from MODIS and presented distinct temporal and spatial variations, which provides a new opportunity for analyzing vegetation physiological dynamics in climate change studies.

Snow covered areas reflect massive solar radiation reaching earth surface, effecting the surface albedo immediately and dynamically. Snow cover is a significant influencing factor of the hydrological cycle, climate change, and the phenology process, playing a considerable role in land cover and climate system [1] . Therefore, snow detection by satellite imagery is recognized as an essential content in a variety of geographic researches.

The maximum and minimum normalized difference vegetation indexes (NDVIs) describe two extremes of vegetation greenness during a predefined period. A maximum NDVI image can be composited easily via the direct selection of the maximum NDVI from multiple observations without the need to mask out cloud or snow. But, a minimum NDVI image cannot be built in a similar manner. In this paper, an approach was proposed to composite the minimum NDVI (the least vegetation greenness) image. The minimum spectral index that consists of the green (555 nm) and SWIR bands (2130 nm) from MODIS data, which was named here as the Brown Vegetation Index (BVI), was taken as a proxy to composite the minimum vegetation NDVI. This composite method performs well on a global scale for the NDVIs that were derived from MODIS land surface reflectance (MOD09A1) products. The BVI-based minimum NDVI was compared with the direct selection of the minimum NDVI after excluding contaminated observations using a refined cloud/snow mask. The comparison shows that the difference for 97% of the minimum NDVI between the two approaches is within the range of ±0.1 NDVI unit. Various potential spectral indices for compositing the minimum NDVI were compared, which demonstrated the BVI-based approach was top rated. Several examples demonstrated that the composited minimum NDVI is valuable and effective for identifying evergreen forests, monsoon forests, and double cropping. The minimum NDVI combined with the maximum NDVI would simplify the way to describe intraannual vegetation changes.

The climatology of dust storms during the nighttime in northern China and Mongolia (33°N-54°N, 73°E-136°E) is characterized and compared with that at daytime at a 1-km resolution based on Aqua/MODIS thermal observations during 2002-2013. The dust was extracted with the dynamic reference brightness temperature differences (DRBTD) dust detection algorithm. The spatial distribution and seasonal and annual variations of dust events are generally similar for day and night, with less dust frequency during nighttime. The major source regions are deserts, including the Tarim Basin, Hexi Corridor, Gobi in Mongolia and northern China, Horqin Sandy Land and Tsaidam Basin, with a maximum frequency occurring in the Tarim Basin. Significant annual and seasonal variations are found for dust events. Low dust frequency was presented during 2003-2005 and 2009-2012, and high dust frequency from 2006-2008 and 2013. More than 43.1% of dust events occur in spring from March to May, with the maximum proportion (up to 20.3%) occurring in April.

The India monsoon is changing under the global climate change, which influence the transformation from dry season to wet season, and then impact the photosynthetic activity and transpiration of tropical and subtropical monsoon forest. The descriptions of this kind of phenology in this area are still lacking, where various shapes of seasonal trajectories (e.g. “U” shape and “V” shape) and the special climate made the extraction of vegetation phenology difficult. However, current curve fitting approaches were inappropriate for simulating all these seasonal trajectories well. In this paper a modified logistic function method was proposed to extract the vegetation phenology (onset of greenness) of India monsoon forest. Due to the fine cloud detection method and the minimum NDVI composited technology, this modified approach had the ability to extract the vegetation phenology of India monsoon forest, which showed significant superiority over the compared six approaches in our previous study. The extracted time series of vegetation phenology by using this modified approach would facilitate the understandings of climate patterns and climate changes in the India monsoon area.

This paper addresses the retrieval of sea surface temperature (SST) and land surface temperature (LST) from the measurements acquired by the Stretched Visible and Infrared Spin Scan Radiometer (SVISSR) on FengYun (FY) 2C/D/E satellites. First, the generalized split-window algorithms for SST and LST retrieval were developed using the moderate spectral resolution atmospheric transmittance algorithm and computer model (MODTRAN) fed with the parameter-adjusted standard model atmospheres. Then, the developed algorithms were applied to SST and LST retrieval from the SVISSR/FY-2C/D measurements in September 2007 and the SVISSR/FY-2E measurements in May 2010 over a large study area (latitude: 10° S-50° N; longitude: 60° E-130° E). Finally, the derived SVISSR/FY-2 SST and LST were, respectively, cross-validated with the MODIS/Terra SST and LST products. The results show the following: 1) the generalized split-window algorithms developed in this work are valid for both SST and LST retrieval from SVISSR/FY-2C/D/E measurements; 2) in contrast to the MODIS/Terra SST and LST products, the errors in nighttime retrieval are usually less than that in daytime retrieval, and the errors in SVISSR/FY-2 SST are generally less than the errors in SVISSR/FY-2 LST; and 3) for both daytime and nighttime, the total errors in SVISSR/FY-2C/D/E LST are, respectively, 0.3 ± 1.6 K, 0.3 ± 1.9 K, and 1.0 ± 1.8 K, while the total errors in SVISSR/FY-2C/D/E SST are, respectively, -0.6 ± 1.3 K, -0.2 ± 1.5 K, and 0.4 ± 1.8 K.

This letter addressed the development of split-window algorithm to estimate land surface temperature (LST) from the measurements acquired by the Visible and Infrared Radiometer on FengYun 3A using radiative transfer modeling experiment with the moderate spectral resolution atmospheric transmittance algorithm and computer model and the SeeBor V5.0 database. To improve the accuracy, the total precipitable water and the mean of land surface emissivities (LSEs) and LST were divided into several subranges. The split-window algorithm was applied to the Northeastern China area (115°E-135°E, 40°N-55°N), and then, the estimated LSTs were cross-validated with the Terra Moderate Resolution Imaging Spectroradiometer (MODIS/Terra) LST and Emissivity (LST/E) V5 products: MOD11C1 V5 and MOD11_L2 V5. The results show that the LSTs in this work are averagely consistent with the MODIS/Terra LST/E V5 products with accuracy better than 1.0 K: The errors are 0.5 ± 0.9 K and 0.0 ± 0.9 K for daytime and nighttime, respectively, when the retrieved LSTs were compared to the MOD11C1 V5 product, while the errors are 0.6 ± 0.9 K and 0.0 ± 0.9 K for daytime and nighttime, respectively, when the results were compared to the MOD11_L2 V5 product.

The Multi-Angle Imaging Spectroradiometer (MISR) is a powerful sensor for leaf area index (LAI) mapping with its simultaneous multi-angle observations. However, the LAI product derived from MISR observations has low temporal resolution, which is unsatisfactory for many applications. This paper presents an algorithm that expands the MISR LAI product to high temporal resolution with the aid of Moderate Resolution Imaging Spectroradiometer (MODIS) data. The algorithm establishes relationships between the MISR LAI and the MODIS red/near-infrared band ratio (simple ratio (SR)) pixel by pixel using coincident data of these two sensors for the past nine years. Using these pixel-based SR-LAI relationships, a new LAI product with the merits of the original MISR product and high temporal resolution is obtained from MODIS surface reflectance. The expanded LAI series was compared with the original MISR and MODIS LAI products, as well as field LAI measurements made at the Baohe and Maoershan forest sites and the Hulunbeier grassland site, to assess the algorithm's performance. The results show that the temporal coverage of the MISR LAI improved from 15.5% to 65.2% in an 8-day composite, and the mean root-mean-square error is 0.74 for the vegetated pixels. This LAI product has similar temporal consistency and seasonal dynamics to the existing MODIS LAI product generated from the main algorithm, but is more robust against the low quality of reflectance inputs. The expanded LAI product differs with field measurements by about 11.5%, with agreement to field observations at all three sites within an accuracy of 0.8 LAI.

Fire is one of the most contributive factors to forest ecosystem disturbances. Many researches on burnt area detection have been carried out. This paper discusses the change of forest characters before and after fires in different phenophases and several burnt scar detection methods are implemented to confirm the influence of phenophases on forest burnt scar detection based on multi-temporal Moderate Resolution Imaging Spectroradiometer(MODIS) data. Preliminary results show that burnt area detection methods based on time series MODIS vegetation indices(VI) products carry out reliable result while mapping burned scar caused by fire that occurred on and after maturity days, but do not give very fine performance while detecting burnt scar caused by fire occurred before green-up days, except in the case that the fire is fatal to forest.

A global multi-decade long-term Leaf area index (LAI) record from remote sensing measurements is required for global change modeling and analysis. The dataset is generated through the fusion of MODIS and historical AVHRR data based on global pixel-based AVHRR SR MODIS LAI relationship which is established with AVHRR data and LAI derived form high quality MODIS observations in the overlapping years from 2000 to 2006. Based on this relationship, historical AVHRR retrieval from 1981 to 2006 is constrained with high quality MODIS observations. The comparisons of the results from 2000 to 2006 show high consistency between the derived AVHRR and MODIS LAI dataset. The derived AVHRR and MODIS LAI were directly validated with 120 field measurements in global 36 sites with various vegetation types, with RMSE of 0.91 and 1.01, respectively.

With increasing land degradation and deforestation, dust storms have been an essential factor of air pollution and global climate and biogeochemical cycle. A new index, Thermal Infrared Integrated Dust Index (TIIDI), has been developed to detect dust over land and ocean. This dust detection algorithm is based on brightness temperature difference of four thermal infrared channels, including BTD3.7-11, BTD8.6-11 and BTD12-11. BTD12-11 is mainly used to identify cloud, BTD8.6-11 is used as an indicator of airborne dust and surface sand, and BTD3.7-11 is used to separate dark surface and represent the intensity of dust storm. The algorithm is applied to monitor atmospheric dust storms over various landcover types, including ocean, dark object (vegetation), and bright surface such as desert by using MODIS data. The results show that the TIIDI could distinguish mineral dust from cloud and land surface over bright surface and ocean. The index provides a dust indicator both day and night for global apply.

Using MODIS data over China for the 2000-2008 period, we performed a statistical analysis of the apparent reflectance difference of two blue bands (BRD) between the 469nm and 415nm over ocean and land. The results showed that the BRD gives a discrimination of heavy biomass burned smoke and dust from others phenomena such as cloud, snow/ice, haze and clear-sky surface. It indicates that the MODIS blue bands contrast provides a potentially simple approach for identification of heavy smoke and dust. This method was applied to detect dust storms over several major dust sources region, including the Gobi Desert, the Red Sea and Afghanistan, and fire smoke over Southeastern Siberia and Alaska. The results showed that comparisons of “blue contrast” are consistent with the aerosol thickness retrieval.

The visible/SWIR band ratio is important for aerosol retrieval. In this presentation, a pixel-based ratio is estimated for MODIS data. It is assumed that the ratio is constant in a pixel in same season in ten years. The 2000-2009 MODIS data are used to infer the pixel for each two months. All clear pixels are selected and then the Rayleigh effects are corrected. The most 5%clearest pixels are used to calculate the ratio for each pixel. The spatial and temporal distribution of ratio is shown. And these pixel-based ratio are used to estimated the aerosol. The results indicated this land reflectances ratio is suitable for aerosol retrieval.

Leaf area index (LAI) is an essential parameter for monitoring crop growth dynamic. An algorithm, which is based on physical model and neural networks to derive crop LAI from MODIS land surface reflectance, is presented. This algorithm utilizes the directional reflectances instead of the BRDF normalized data to avoid complex BRDF normalization and the error from it. The estimated LAI is compared with existing LAI products. Results show that it is consistent with MODIS (RMSE = 0.4994) and CYCLOPES (RMSE = 0.6658) LAI products in temporal and spatial patterns. The algorithm is validated against ground measurements of annual crop LAI in 2004 in Hengshui, China. The neural network derived LAI could represent the spatial pattern of the field LAI. However, all these LAI products are lower than field measurements. It would be suggested that the physical model should be modified to adapt to the dense crop in Northern China.

Much attention has been focused on the assimilation of satellite data and products into land surface processes. In this letter, a variational data assimilation scheme is developed based on the weak-constraint concept. It assimilates surface skin temperature into a simple land surface model for the estimation of turbulent fluxes. An automatic differentiation technique is used to derive the adjoint codes to evaluate the gradient of the cost function. After the construction of this assimilation system, numerical experiments are conducted to test its performance with different model errors, and the comparison is also made with the strong constraint scheme. The results show that the land surface turbulent fluxes can be retrieved with highly satisfactory accuracy.

This letter reports a statistical method to estimate detector-dependent systematic error in Moderate Resolution Imaging Spectroradiometer (MODIS) thermal infrared (TIR) Bands 20-25 and 27-36. There exist scan-to-scan overlapped pixels in MODIS data. By analyzing a sufficiently large amount of those most overlapped pixels, the systematic error of each detector in the TIR bands can be estimated. The results show that the Aqua MODIS data are generally better than the Terra MODIS data in 160 MODIS TIR detectors. There are no detector-dependent systematic errors in Bands 31 and 32 for both Terra and Aqua MODIS data. The maximum detector errors are 3.00 K in Band 21 of Terra and -8.15 K in that of Aqua for brightness temperatures of more than 250 K

The minimum band reflectance based compositing method is a main compositing method for retrieval of land surface reflectance from multi-temporal observations. But the shadow effect is difficult to be removed. In this paper, a new method that can identify and remove the shadow pixels from multi-temporal data is proposed. It compares the least two pixels to determine the possibility of the first pixel as shadow. If the first pixel is shadow, it is discarded. The shadow pixels can he removed from the multi-temporal observations after iteration of this procedure two or three times. It has been tested with MODIS data and the results showed it could operate effectively over different land cover types.

The burned area is an important parameters for modelling the carbon cycles. The remote sensing tenology is a only way to monitor it at large scale region. In general, two temporal vegetation index difference was used to detect the burned area. But this technology is difficult to be applied to large-scale region owing to the BRDF effect, atmospheric contamination, geolocation errors, phenological changes, vegetation regrowth and others. In this paper, a new approach was proposed to detect the burned area using MODIS data, which is based on the vector-change technology, and combines the MODIS 500 and 250 meter resolution bands data to find 250 meter resolution burned area. The method adequately uses the spectral and multi-spatial resolution character of MODIS data that can resist the noise pollution and improve the detection accuracy. The detection results are very corresponding with the visual interpretation under different background. Since the method needs no prior knowledge, it could also be applied in large region scale. Based on this algorithm, the burned area dataset covering all China from 2000 to 2004 were produced. © 2005 IEEE.

A change vector based method for extracting flooded area from multi-temporal ENVISAT ASAR image is presented in this paper. The high resolution and multi-polarisation modes of ENVISAT ASAR make itself a useful tool for flood monitoring. The vector change method, which has been used extensively to detect the change of land surface in optical remote sensing, can take advantage of the reference images information recorded during, the non-flooded period so it is more robust than other change detection approaches. This technology has been applied to monitor flood in Dongting Lake, China using six temporal ASAR data from June to October 2004.