Moderate Imaging Research Articles

Remote Sensing Retrieval of Cloud Top Height Using Neural Networks and Data from Cloud-Aerosol Lidar with Orthogonal Polarization.

In order to enhance the retrieval accuracy of cloud top height (CTH) from MODIS data, neural network models were employed based on Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) data. Three types of methods were established using MODIS inputs: cloud parameters, calibrated radiance, and a combination of both. From a statistical standpoint, models with combination inputs demonstrated the best performance, followed by models with calibrated radiance inputs, while models relying solely on calibrated radiance had poorer applicability. This work found that cloud top pressure (CTP) and cloud top temperature played a crucial role in CTH retrieval from MODIS data. However, within the same type of models, there were slight differences in the retrieved results, and these differences were not dependent on the quantity of input parameters. Therefore, the model with fewer inputs using cloud parameters and calibrated radiance was recommended and employed for individual case studies. This model produced results closest to the actual cloud top structure of the typhoon and exhibited similar cloud distribution patterns when compared with the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) CTHs from a climatic statistical perspective. This suggests that the recommended model has good applicability and credibility in CTH retrieval from MODIS images. This work provides a method to improve accurate CTHs from MODIS data for better utilization.

Annual maps of forest cover in the Brazilian Amazon from analyses of PALSAR and MODIS images

Abstract. Many forest cover maps have been generated by using optical and/or microwave images, but these forest cover maps have large area and spatial discrepancies. To date, few studies have assessed forest cover maps in terms of two biophysical parameters used in forest definition: canopy height and canopy coverage. We generated annual forest cover maps from 2007 to 2010 and evergreen forest cover maps from 2000 to 2021 in the Brazilian Amazon using the images from the Phased Array type L-band Synthetic Aperture Radar and the time series images from the Moderate Resolution Imaging Spectroradiometer, using the forest definition of the Food and Agriculture Organization of the United Nations (>5 m tree height and >10 % canopy coverage) as the reference. We used the canopy height and canopy coverage datasets from the Geoscience Laser Altimeter System during 2003–2007 to assess annual forest cover maps from 2007 to 2010 and annual evergreen forest cover maps from 2003 to 2007, and the results show high accuracy of these forest cover and evergreen forest cover maps. These annual forest cover maps and annual evergreen forest cover maps provide data support for the analyses of the causes, processes, and consequences of forest cover changes in the Brazilian Amazon (https://doi.org/10.6084/m9.figshare.21445626; Qin and Xiao, 2022a; https://doi.org/10.6084/m9.figshare.21445590; Qin and Xiao, 2022b).

Contributions of the vegetation index (NDVI) in water quality prediction models in a semi-arid tropical watershed

In this work, a new approach to using NDVI as a predictor of water quality parameters in arid environments is proposed. Our focus was the Araçuaí river basin, which has a predominance of native cerrado vegetation and is subject to seasonal variations in rainfall and vegetation cover. MODIS images (MOD13Q1) from 2000 to 2018 were used to calculate the NDVI of the contributing areas of the water quality monitoring stations and to analyze its relationship with fourteen water quality parameters. The NDVI showed significant seasonality, with high values in the rainy season, and temporal trends of increase in stretches related to the main river. A strong and new relationship was observed between NDVI and six water quality parameters: color, total dissolved solids, total suspended solids, total solids, nitrate and turbidity, this being stronger at the local scale, with better performance at from models that use incremental NDVI, capturing local variations in vegetation cover, instead of regional NDVI, which reflects the general state of vegetation. The results demonstrate the potential of using these indices to develop and improve prediction models of water quality parameters in river basins and to expand the spatial and temporal coverage of water quality monitoring.

Lagrangian and Eulerian modelling of river plumes in the Great Barrier Reef system, Australia

In this paper, we describe methods to verify the adequacy and accuracy of Lagrangian particles from a Lagrangian model to reproduce the concentrations of a passive tracer from an Eulerian-model in river plumes. The modelling simulates plumes from two major rivers discharging in the Great Barrier Reef (GBR), Australia, under real-world scenarios. The study has been a part of a major project to aid in the protection of the GBR system from the impacts of extreme events and climate change. We employ the Regional Ocean Modelling System (ROMS) activated with its built-in Lagrangian model, and forced with wind fields from global models and recorded river volume discharges. The ROMS-Lagrangian model tracks the Lagrangian particles using the spatially interpolated velocities computed on the Eulerian ROMS three-dimensional (3D) grid. The Lagrangian particles are released in the river in proportion to the measured river volume flux. We apply a novel technique that exploits Voronoi polygon areas to convert Lagrangian particle separation into a concentration field. This facilitates comparison with the passive tracer concentrations driven by the Eulerian velocities computed on the ROMS 3D grid. We evaluate the fate of Lagrangian particles activated on a 4-km grid resolution with those of equivalent Eulerian tracer concentrations. For validation, we compare the Lagrangian particles with the Eulerian passive tracer simulated using a higher model grid resolution of 500-m, and found that the computed Lagrangian particle concentrations showed similar overall patterns compared to the passive tracer concentrations from both the coarse 4-km, and high-resolution 500-m Eulerian models. The spatial extent of the particles was in better agreement with the coarse 4-km model than with the higher resolution 500-m model. The clustering of particles resulted in structures at finer scales than the Eulerian model, over the high particle density areas that compared well against observations. The Lagrangian particles are able to capture the general river plume patterns seen in the satellite MODIS images and the satellite derived Chlorophyll-a. We further compare the online ROMS model with an offline particle tracker, OceanPARCELS and determined the similarity of results from both online and offline methods. Overall, our study demonstrates the viability to use Lagrangian particles to reproduce the Eulerian tracer properties, which can be combined with particle properties include tracer age providing to enhance the guidance on the distribution and concentration of effluent from localised flooding river events.

The Use of EVI Algorithm in Predicting Rice Productivity, Harau District, Lima Puluh Kota Regency, West Sumatra

Using remote sensing data in predicting rice productivity can provide information quickly and easily for relatively large areas. Harau District is a rice center in Lima Puluh Kota Regency and is one of the many rice-producing areas in West Sumatra Province. This study aims to determine a mathematical model to estimate rice productivity in Harau District, Lima Puluh Kota Regency, using the EVI (Enhanced Vegetation Index) algorithm. This study used two methods, field observation and MODIS image data processing (MOD13A1 v006). The results of the analysis obtained a mathematical model to predict rice productivity in Harau District, namely y = 5393.3x4 - 8140.7x3 + 4518.9x2 - 1121.6x + 113.55 where y is the productivity value (tons/ha) and x as the EVI value. The percentage of errors in observation and estimation data was 15.284%, and the results of the observation and estimation data validation test (NSE) were 0.734, a good category.

Predicting Chlorophyll-a Concentration and Harmful Algal Blooms in Lake Okeechobee Using Time-Series MODIS Satellite Imagery and Long Short-Term Memory

Highlights A 10-year dataset of time-series MODIS imagery and in situ Chl-a concentration were curated for Lake Okeechobee. LSTM significantly outperformed KNN, SVR, and RF for Chl-a prediction and subsequent HAB detection. The optimal window length was found to be 13 days with a 4-day temporal resolution for the LSTM model. KNN, SVR, and RF models were not effective at utilizing the temporal dynamics of the input features. Abstract. Harmful algal blooms (HABs) in inland waterbodies are a global concern due to their negative impact on human, animal, and ecosystem health. Chlorophyll-a (Chl-a) concentration is an important water quality parameter for monitoring HABs. While statistical and machine learning (ML) models have been widely studied to predict Chl-a concentration and HABs based on single-time-point satellite data, this work assessed whether long short-term memory (LSTM) can improve both tasks by leveraging temporal features in time-series MODIS satellite images compared to three classical ML models, including k-nearest neighbor (KNN), support vector regression (SVR), and random forest (RF). A dataset of daily MODIS images and monthly in situ Chl-a concentration measurements from 2011 to 2020 was curated for Lake Okeechobee, Florida. A window size of 13 days with a temporal resolution of four days was found to produce the optimal performance for LSTM, which significantly outperformed KNN, SVR, and RF for Chl-a prediction with a root mean square error of 11.95 µg/L, a mean absolute error of 8.55 µg/L, and a R2 value of 0.43. The superior performance of LSTM for Chl-a prediction was likely due to its ability to leverage the temporal dynamics in the features associated with HAB development. The Chl-a predictions were further used to determine HAB events, showing better accuracy and a significantly higher F1 score for LSTM over the other models. The study suggested that combining LSTM with high-temporal-resolution time-series data should be preferred over applying common ML models on time-series or single-time-point remote sensing data for Chl-a and HAB monitoring. Keywords: Cyanobacteria, LSTM, Machine Learning, Remote Sensing, Water Quality.

An integrated method to develop high temporal Landsat NDVI using free and open-source software (FOSS) GIS application

Landsat image series are long-term times series observing the earth’s surface at a relatively fine spatial resolution of 30 m on the visible and infrared channels, which can be used to monitor vegetation dynamics using band ratios such as NDVI. However, due to cloud cover, having Landsat NDVI at a high temporal resolution is impossible. This is especially found in high-humidity regions such as Indonesia, particularly in the rainy season. Whereas some studies, such as long-term phenological observations and detailed land use/land cover change detection, may require Landsat NDVI with a high temporal resolution. One of the higher temporal resolution images with similar spectral characteristics to Landsat is MODIS, which can be used to fill the gap; however, it may have a coarser spatial resolution. Thus, some methods have been developed to blend Landsat and MODIS images to create images with both Landsat spatial detail and MODIS high temporal resolution. Unfortunately, the blending methods are usually not straightforward, where complex steps are required to find and combine several techniques with advanced computational skills. This study, therefore, aims to develop an integrated method to create high temporal Landsat NDVI using free and open-source software (FOSS) GIS application, which is easily accessible and applicable.

Vegetation response to droughts: The case of northern Italy

AbstractDrought is a normal and recurrent climate feature, but one of the most important climate hazards that can produce negative effects on natural and socioeconomic systems. The northern part of Italy is historically rich in water resources and a very fertile and productive area. Recently, however, drought events increased their influence on the hydrological behaviour of the main rivers in this region. In this work, we analyse the spatial distribution of drought events and their effects on vegetation greenness in northern Italy during the 2000–2020 period, using MODIS images at 1 km spatial resolution. Correlation maps between fields of bi‐weekly vegetation indices (normalized difference vegetation index [NDVI] and enhanced vegetation index [EVI]) and drought indices (standardized precipitation index [SPI] and standardized precipitation evapotranspiration index [SPEI]) were estimated. To generate the SPEI index, 150 daily precipitation and temperature ground station series were collected, aggregated at bi‐weekly scale, reconstructed, homogenized and spatialized at 1 km resolution by mean of Universal Kriging with auxiliary variables. The NDVI and EVI indices (MOD13A2) and the land surface temperature (LST; MOD11A2) were collected from MODIS images and quality checked, removing pixels with clouds and low‐quality data. Drought events were detected by means of the SPEI and SPI at medium (12‐months) and long (24 and 36 months) timescales. The events were characterized by drought intensity, duration and percentage of affected area. For each pixel, the Pearson's correlation between NDVI/EVI and SPEI/SPI at different timescales was estimated. On the basis of the correlation maps and the CORINE Land Cover 2020 (MCD12Q1), we estimated the drought impact on different vegetation types as the percentage of vegetation undergoing drought stress. The investigation on drought indices has revealed nine main summer and spring events in the study period. The events of summer 2003 and 2017 were especially remarkable, with respectively 50% and 40% of the total area under drought according to the SPEI index at 12 months. The correlation between drought and vegetation indices revealed that in the Po Valley and Tuscan hills, the vegetation communities perturbed by agriculture are those that are most affected by droughts, with 13% of vegetation under drought on long timescales. This study also indicated that in the last two decades, the main trigger of droughts in northern Italy was the lack of precipitation rather than pure temperature increase.

Extraction the Spatial Distribution of Mangroves in the Same Month Based on Images Reconstructed with the FSDAF Model

Mangroves have extremely high economic and ecological value. Through remote sensing, the spatial distribution of and spatiotemporal changes in mangroves can be accurately obtained, providing data support for the sustainable development of coastal wetlands. However, due to the cloudy and rainy conditions in the growing areas of mangroves, there are relatively few high-quality image data available, resulting in a time difference between regional mosaic images, with a maximum difference of several months, which has a certain impact on accuracy when extracting the spatial distribution of mangroves in some regions. At present, most regional mangrove research has ignored the impact of the time difference between mosaic images, which not only leads to inaccurate monitoring results of mangroves’ spatial distribution and dynamic changes but also limits the frequency of monitoring of regional mangrove dynamic changes to an annual scale, making it difficult to achieve more refined time scales. Based on this, this study takes the coastal mangrove distribution area in China as the research area, uses Landsat 8 and MODIS images as basic data, reconstructs the January 2021 images of the research area based on the FSDAF model, and uses a random forest algorithm to extract the spatial distribution of mangrove forests and analyze the landscape pattern. The results showed that the fused image based on the FSDAF model was highly similar to the validation image, with an R value of 0.85, showing a significant positive correlation, indicating that the fused image could replace the original image for mangrove extraction in the same month. The overall accuracy of the spatial distribution extraction of mangroves based on the fused image was 89.97%. The high sample separation and spectral curve changes highly similar to the validation image indicate that the fused image can more accurately obtain the spatial distribution of mangroves. Compared to the original image, the fused image based on the FSDAF model is closer to the validation image, and the fused image can reflect the changes in mangroves in time series, thus achieving accurate acquisition of dynamic change information in a short time span. It provides data and methodological support for future monitoring of dynamic changes in large-scale mangroves. The total area of mangroves in China in January 2021 based on the fused image was 27,122.4 ha, of which Guangdong had the largest mangrove area, with 12,098.34 ha, while Macao had the smallest mangrove area of only 16.74 ha. At the same time, the mangroves in Guangdong and Guangxi had a high degree of fragmentation and were severely disturbed, requiring strengthened protection efforts, while the mangroves in Hong Kong, Zhejiang, and Macao had regular shapes, benefiting from local active artificial restoration.

Assessment of satellite images terrestrial surface temperature and WVP using GNSS radio occultation data

Abstract Land Surface Temperature (LST) and Water Vapor Pressure (WVP) contour maps can be produced using cameras aboard satellites, for instance, under the name “Remote Sensing (RS)”. Satellite image observations should be verified before using based on a reliable data. Global Navigation Satellite System Radio Occultation (GNSS-RO) method is observing accurate Earth atmosphere parameters continuously. In the present research, LST and WVP differences between Landsat 8 (LC08), Sentinel-3 (S3), and MODIS (Terra and Aqua) images and GNSS-RO are assessed in Egypt depending on the satellites operating periods and data availability during the years from 2015 to 2020. Statistically, S3 and Terra have insignificant differences with RO temperature with an average bias of 3.48 °C and 1.47 °C, respectively, but LC08 and Aqua have significant differences with it. For WVP, Aqua and LC08 have insignificant differences with an average bias of 0.02 kg/m2 and 2.31 kg/m2, respectively, but S3 and Terra have significant differences with RO observations. When comparing LC08 LST data to other satellites, it was found that there were insignificant differences between LC08 and S3 as well as Terra. However, significant differences were observed when comparing LC08 LST data to Aqua. Additionally, significant differences were noted when comparing LC08 WVP data to other satellites. In response to these differences, improvement models have been developed to enhance the estimation of terrestrial data through remote sensing, particularly for satellites that exhibited significant disparities when compared to reference observations (RO).

Cross-Radiometric Calibration and NDVI Application Comparison of FY-4A/AGRI Based on Aqua-MODIS

To enhance the accuracy and stability of FY-4A/AGRI detection data, the MODIS, with highly accurate onboard calibration, is selected as the reference sensor for cross-radiation calibration calculations. The following are the data selection conditions: full considered time, observation geometries, field angles, cloud cover, etc. FY-4A/AGRI and Aqua-MODIS image data are selected as matching sample region locations, where the time difference between the observations for the same ground object is less than 15 min, the satellite zenith angle is less than 30°, and the field angle difference is less than 0.01. The 245 collected reflectance spectral curves are convolved with the spectral response functions of the two sensors, and the spectral band adjustment factors of the corresponding bands are calculated for spectral correction purposes. The cross-calibration coefficients for the red and near-infrared bands are calculated by linearly fitting the simulated top of the atmosphere reflectance values and digital number values from the AGRI sensor in a homogeneous area. In this paper, 16 cross-calibration calculations are performed on FY-4A/AGRI image data from August 2018 to September 2020, and the results are compared with the original calibration coefficients to test the feasibility of the proposed method. Additionally, 31 cross-calibration calculations are performed on image data from October 2020 to December 2022 to study the resulting AGRI sensor quality and performance changes. The NDVI of the FY-4A/AGRI image data was calculated before and after the cross-radiometric calibration using the maximum synthesis method. Additionally, the NDVI of the MODIS image data was compared and analyzed from three aspects: time, space, and the change trend. The results show that the spectral band adjustment factor calculated using the reflectance spectral curves of the ground objects in this paper can effectively correct for the spectral differences between the two sensors. Sixteen cross-calibration coefficients are less than 5.2% different from the original calibration coefficients, which fully proves the feasibility of the method used in this paper. All of the cross-calibration results show that the AGRI sensors have a certain degree of attenuation in the red and near-infrared bands, and the annual attenuation rates are approximately 1.37% and 2.55%, respectively. Cross-radiometric calibration has further improved the quality of the NDVI in FY-4A/AGRI imagery, enhancing the precision of its data application.

Delineation of Rice Yield and Production in Combination of Crop Model and Remote Sensing

The study was to simulate rice yield in various places and outline a rice yield map for the study area using GIS, remote sensing, and a rice model. As a case study, the data were collected on the climate, soil characteristics, and rice cropping status in the AnGiang province (Southern of the Mekong River Delta in Viet Nam). The AquaCrop model was used to predict rice yield. The MODIS image delineated the rice cropping status based on spatial and temporal NDVI values. The results of the yield simulation are then put together with information about where the rice was planted, the weather, and the properties of the soil to make a map of the yield distribution. Finally, the outcomes are verified and contrasted with the statistical findings in the last step. The rice yield was predicted and compared with actual 1 and 6 percent rice yields. The anticipated rice yield map was established for the Winter-Spring cropping season 2012-2013 and the Summer-Autumn and Autumn–Winter cropping seasons 2013. Rice production and yield distribution can be divided into two major areas. The alluvial soil area produces significantly more rice than the LongXuyen quadrangle area because of the difference in soil and weather conditions. Rice yield simulation and delineation combining remote sensing and crop models is a good approach for yield prediction and better agricultural management strategy development in a country or region. The accuracy of the results depends on the quality of the input data, such as soil weather and remote sensing.
 Keywords: GIS, MODIS, cropping season, rice yield

A Spatiotemporal Fusion Model of Land Surface Temperature Based on Pixel Long Time-Series Regression: Expanding Inputs for Efficient Generation of Robust Fused Results

Spatiotemporal fusion technology effectively improves the spatial and temporal resolution of remote sensing data by fusing data from different sources. Based on the strong time-series correlation of pixels at different scales (average Pearson correlation coefficients > 0.95), a new long time-series spatiotemporal fusion model (LOTSFM) is proposed for land surface temperature data. The model is distinguished by the following attributes: it employs an extended input framework to sidestep selection biases and enhance result stability while also integrating Julian Day for estimating sensor difference term variations at each pixel location. From 2013 to 2022, 79 pairs of Landsat8/9 and MODIS images were collected as extended inputs. Multiple rounds of cross-validation were conducted in Beijing, Shanghai, and Guangzhou with an all-round performance assessment (APA), and the average root-mean-square error (RMSE) was 1.60 °C, 2.16 °C and 1.71 °C, respectively, which proved the regional versatility of LOTSFM. The validity of the sensor difference estimation based on Julian days was verified, and the RMSE accuracy significantly improved (p < 0.05). The accuracy and time consumption of five different fusion models were compared, which proved that LOTSFM has stable accuracy performance and a fast fusion process. Therefore, LOTSFM can provide higher spatiotemporal resolution (30 m) land surface temperature research data for the evolution of urban thermal environments and has great application potential in monitoring anthropogenic heat pollution and extreme thermal phenomena.

A deep transfer learning framework for mapping high spatiotemporal resolution LAI

Leaf area index (LAI) is an important variable for characterizing vegetation structure. Contemporary satellite-based LAI products with moderate spatial resolution, such as those derived from the MODIS observations, offer unique opportunities for large-scale monitoring but are insufficient for resolving heterogeneous landscapes. Although high-resolution satellite observations can derive detailed LAI maps, the low revisit frequency and the presence of cloud cover disrupt the temporal continuity of these high-resolution data, thereby leading to the fact that most current LAI inversion models for high spatial resolution data are based on single pixel and date without fully utilizing temporal information. Moreover, LAI estimation models trained solely using satellite products or simulations are also impeded by the inconsistencies between field and satellite LAI exist owing to local atmospheric, soil, and canopy conditions, while in-situ LAI measurements are sparse and insufficient for large-scale missions. To address these challenges, this study proposed a new framework based on deep transfer learning, which includes three key features that contribute to its high performance. Firstly, a Bi-directional Long Short-Term Memory (Bi-LSTM) model is pre-trained using MODIS reflectance and MODIS LAI products to capture the general non-linear relationship between reflectance and LAI and incorporate temporal dependencies as prior information to reduce uncertainties associated with ill-posed inversion problems and noises. Secondly, this pre-trained Bi-LSTM is transferred from satellite to field by fine-tuning with sparse in-situ LAI measurements to overcome issues arising from local inconsistencies. Thirdly, reconstructed Landsat time-series images via fusing MODIS and Landsat reflectance images are used as inputs of the Bi-LSTM to generate high-quality Landsat LAI products at both high spatial and temporal resolutions. To validate the proposed approach, field LAI measurements were collected at nine locations across the contiguous U.S. from 2000 to 2018, including three land cover types: croplands, grasslands, and forest. Quantitative assessments demonstrate that the Bi-LSTM outperforms three benchmarks, including a PROSAIL-based Look-Up Table (LUT) method, a random forest-based LAI retrieval and MODIS LAI (MCD15A3H), exhibiting lower RMSE and higher R2 in most cases. Additionally, the Bi-LSTM predictions yield lower random fluctuations than estimations from the LUT·, random forest and MODIS LAI, indicating the higher robustness of the proposed framework. The findings of this study highlight the value of transfer learning in estimation of vegetation biophysical parameters, which involves pre-training using sufficient existing satellite products to produce a generalized model and transferring knowledge from in-situ measurements to bridge gaps between satellite and field. By leveraging advanced transfer learning techniques and multi-source and multi-scale data, the proposed framework enables the production of long-term LAI maps at fine resolutions, facilitating downstream applications in regions characterized by high spatial heterogeneity.

Combining Satellite Optical and Radar Image Data for Streamflow Estimation Using a Machine Learning Method

River water surface extent can be extracted from optical and radar satellite images; this is useful for estimating streamflow from space. The radiation characteristics of open water from the visible and microwave bands are different and provide independent information. In this study, for the purpose of improving streamflow estimation from space for data-sparse regions, a method that combines satellite optical and radar images data for streamflow estimation using a machine learning technique was proposed. The method was demonstratedthrough a case study in the river segment upstream of the Ganzi gauging station on the Yalong River, China. Utilizing the support vector regression (SVR) model, the feasibility of different combinations of water surface area derived from Sentinel-1 synthetic aperture radar images (AREA_SAR), modified normalized difference water index derived from Landsat 8 images (MNDWI), and reflectance ratios between NIR and SWIR channels derived from MODIS images (RNIR/RSWIR) for streamflow estimation were evaluated through three experiments. In Experiment I, three models using AREA_SAR (Model 1), MNDWI (Model 2), and a combination of AREA_SAR and MNDWI (Model 3) were built; the mean relative error (MRE) and mean absolute error (MAE) of streamflow estimates corresponding to the SVR model using both AREA_SAR and MNDWI (Model 3) were 0.19 and 31.6 m3/s for the testing dataset, respectively, and were lower than two models using AREA_SAR (Model 1) or MNDWI (Model 2) solely as inputs. In Experiment II, three models with AREA_SAR (Model 4), RNIR/RSWIR (Model 5), and a combination of AREA_SAR and RNIR/RSWIR (Model 6) as inputs were developed; the MRE and MAE for the model using AREA_SAR and RNIR/RSWIR (Model 6) were 0.25 and 56.5 m3/s, respectively, which outperformed the two models treating AREA_SAR (Model 4) or MNDWI (Model 5) as single types of inputs. In Experiment III, three models using AREA_SAR (Model 7), MNDWI, and RNIR/RSWIR (Model 8) and the combination of AREA_SAR, MNDWI and RNIR/RSWIR (Model 9) were built; combining all three types of satellite observations (Model 9) exhibited the highest accuracy, for which the MRE and MAE were 0.18 and 18.4 m3/s, respectively. The results of all three experiments demonstrated that integrating optical and microwave observations could improve the accuracy of streamflow estimates using a data-driven model; the proposed method has great potential for near-real-time estimations of flood magnitude or to reconstruct past variations in streamflow using historical satellite images in data-sparse regions.

An improved deep learning network for AOD retrieving from remote sensing imagery focusing on sub-pixel cloud

ABSTRACT Following the success of MODIS, several widely used algorithms have been developed for different satellite sensors to provide global aerosol optical depth (AOD) products. Despite the progress made in improving the accuracy of satellite-derived AOD products, the presence of sub-pixel clouds and the corresponding cloud shadows still significantly degrade AOD products. This is due to the difficulty in identifying sub-pixel clouds, as they are hardly identified, which inevitably leads to the overestimation of AOD. To overcome these conundrums, we proposed an improved deep learning network for retrieving AOD from remote sensing imagery focusing on sub-pixel clouds especially and we call it the Sub-Pixel AOD network (SPAODnet). Two specific improvements considering sub-pixel clouds have been made; a spatial adaptive bilateral filter is applied to top-of-atmosphere (TOA) reflectance images for removing the noise induced by sub-pixel clouds and the corresponding shadows at the first place and channel attention mechanism is added into the convolutional neural network to further emphasize the relationship between the uncontaminated pixels and the ground measured AOD from AERONET sites. In addition, a compositive loss function, Huber loss, is used to further improve the accuracy of retrieved AOD. The SPAODnet model is trained by using ten AERONET sites within Beijing-Tianjin-Hebei (BTH) region in China, along with their corresponding MODIS images from 2011 to 2020; Subsequently, the trained network is applied over the whole BTH region and the AOD images over the BTH region from 2011 ~ 2020 are retrieved. Based on a comprehensive validation with ground measurements, the MODIS products, and the AOD retrieved from the other neural network, the proposed network does significantly improve the overall accuracy, the spatial resolution, and the spatial coverage of the AOD, especially for cases with sub-pixel clouds and cloud shadows.

Crop monitoring and biomass estimation based on downscaled remote sensing data in AquaCrop model (case study: Qazvin Plain, Iran).

Ensuring food security requires monitoring crop growth conditions prior to harvesting. This research encompass three stages: remote sensing, crop modeling, and establishing a connection between the two. Besides, Fusion satellite images have been implemented to monitor and investigate fodder corn in three study areas. In the remote sensing stage, after implementing the downscaling algorithm and producing the leaf area index (LAI) time series, the results were compared with the estimated values from Landsat 8 and MODIS images, which were associated with overestimation in all cases. Furthermore, the results exhibited statistically significance with R2 > 95% and P-value < 0.05. The AquaCrop model was first calibrated and implemented in the crop modeling section at each growth stage based on the observational data measured in each field. The accuracy of the simulated model was checked according to the results of statistical indicators with high accuracy (NRMSE = 10% and RMSE = 0.03 (Ton/ha)) at significant level (95%) and was associated with underestimation. Using the SVM decision support algorithm, the relationship between downscaled LAI and calibrated CC (crop canopy) was estimated. This relationship was generated with 70% of the data and subsequently validated using the remaining 30% (R2 = 0.99, NRMSE = 0.01). On this basis, CC values were predicted. Finally, biomass values were compared with observed biomass values. According to the results of statistical indicators (RMSE = 0.19 (Ton/ha), NRMSE = 0.01, R2 = 0.96, P-value < 0.05), biomass estimation was highly accurate. These results demonstrate the reliability and accuracy of the model and the proposed method for simulating and estimating biomass before harvesting.

A novel total phosphorus concentration retrieval method based on two-line classification in lakes and reservoirs across China

Phosphorus is widely recognized as a nutrient that restricts growth and is the primary contributor to eutrophication in 80 % of water bodies. Consequently, the Chinese government has consistently prioritized monitoring and controlling total phosphorus (TP) levels. The remote estimation of TP in lakes and reservoirs at a national scale is a challenging task due to TP being a non-optically active parameter. Currently, there is a lack of developed TP inversion models specifically designed for lakes and reservoirs in China. For solving this problem, a novel two-line classification method drawn on scatter plots based on the natural logarithm of TP (Ln(TP)) and B33/B9 was proposed and used to classify 1211 measured samples obtained from field cruises in 105 lakes and reservoirs across China from 2012 to 2022 into three categories, Class 1, Class 2, and Class 3. Results demonstrate that the proposed classification method has the ability to enhance the correlation between Ln(TP) and 43 basic potential single band and band combinations. Specifically, the correlation range improved from (−0.31,0.15) to (−0.77,0.24) in Class 1, (−0.81, 0.36) in Class 2, and (−0.74, 0.52) in Class 3. Additionally, the classification method also improved the correlation range between Ln(TP) and 820 band ratios, from (−0.32, 0.32) to (−0.83, 0.82) in Class 1, (−0.86, 0.86) in Class 2, and (−0.86, 0.86) in Class 3. These datasets were subsequently utilized as input for eXtreme Gradient Boosting (XGBoost) models. Finally, well performing XGBoost models in Class 1 (R2 = 0.76, RMSE = 0.3, MAPE = 12 %), Class 2 (R2 = 0.84, RMSE = 0.49, MAPE = 38 %), and Class 3 (R2 = 0.74, RMSE = 0.46, MAPE = 14 %) were used to map TP of 563 large lakes and reservoirs (≥20 km2) across China using MODIS images from 2005, 2010, 2015, and 2020. This study presents a novel approach for estimating non-optically active parameters through remote sensing on a national scale.

Introducing a novel dust source identification method based on edge points and paths extracted from integration of time-series MODIS products

In this paper, a novel method for identifying dust sources in the Middle East using multi-temporal MODIS images was introduced. This method was utilized to identify the sources of 14 Sand and Dust Storms (SDSs) events that occurred between 2008 and 2012 in Ahwaz city, in southwestern Iran. For this purpose, the SDS extents were detected using the proposed Thermal Infrared (TIR) algorithm consisting of MODIS TIR bands 20, 31, and 32, and the Brightness Temperatures Difference between the two bands 31 and 32. The true color RGB MODIS image of the study area showed that the proposed SDS detection algorithm was highly efficient in separating these storms. In the next step, potential dust sources were obtained based on the proposed algorithm, which included the main sub-steps of separating the major SDS extents and calculating the centroids and the edges of each, computing the angles between each point of the obtained edges with the centroid of the corresponding SDS extent, and applying two thresholds to separate potential sources. The dust source map was produced by overlaying all dilated potential dust sources and classifying them based on the number of overlapped layers. This map illustrated that dust sources were scattered in eastern Syria, northwestern and southeastern Iraq, and southwestern Iran. Interpretation of the results demonstrated that the degradation of vegetation covers and reducing the width of rivers and dry lake-beds were the main characteristics of these sources. From the results of this research, it was concluded that the proposed novel method is a promising and efficient algorithm for identifying dust sources.

Long-Range Transport Analysis Based on Eastern Atmospheric Circulation and Its Impact on the Dust Event over Moldavia, Romania in August 2022

During the second half of August 2022, a dust intrusion event occurred when dust that originated in the dry regions of the Kalmyk steppe (located in Russia, northeast of the Black Sea, north of Georgia, and northwest of the Caspian Sea) and the Precaspian plain was transported over the eastern region of Romania. The arid soil found in these areas can be attributed to an extended period of intense drought, with notable instances occurring in 2002, 2003, 2015, and 2018. This situation was further intensified by heatwaves experienced in May and June of 2022. The dust event was captured in MODIS images. In addition, smoke trains originating from fires in the north of the Azov were detected, but these did not reach Romania. Optical parameters from AERONET were used to confirm the dust event. To determine the trajectory of the particles, the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model was used in this paper. The ensemble median model was used to highlight the presence and concentration of dust in the eastern part of Romania. Aerosols were detected between 0 and 4 km, according to radar and ceilometer data from the REXDAN cloud remote sensing facility in Galați, Romania. This dust intrusion event was the result of the dominant easterly circulation caused by the extension of the East European High to the northeast of the continent, which transported the dust towards the eastern part of Romania for more than 2 days. Moreover, the torrential rains between 22 and 24 August did not clear the atmosphere of dust, since the intense easterly circulation kept carrying the dust into the Moldavian area.