Validation of non-linear split window algorithm for land surface temperature estimation using Sentinel-3 satellite imagery: Case study; Tehran Province,Iran

In recent years, land surface temperature (LST) has become critical in environmental studies and earth science. Remote sensing technology enables spatiotemporal monitoring of this parameter on large scales. This parameter can be estimated by satellite images with at least one thermal band. Sentinel-3 SLSTR data provide LST products with a spatial resolution of 1 km. In this research, direct and indirect validation procedures were employed to evaluate the Sentinel-3 SLSTR LST products over the study area in different seasons from 2018 to 2019. The validation method was based on the absolute (direct) evaluation of this product with field data and comparison (indirect) evaluation with the MODIS LST product and the estimated LST using the non-linear split-window (NSW) algorithm. Also, two emissivity estimation methods, (1) NDVI thresholding method (NDVI-THM) and (2) classification-based emissivity method (CBEM), were used to estimate the LST using the NSW method according to the two thermal bands of Sentinel-3 images. Then, the accuracy of these methods in estimating LST was evaluated using field data and temporal changes of vegetation, which the NDVI-THM method generated better results. For indirect evaluation between the Sentinel-3 LST product, MODIS LST product, and LST estimated using NSW, four filters based on spatial and temporal separates between pairs of pixels and pixel quality were used to ensure the accuracy and consistency of the compared pairs of a pixel. In general, the accuracy results of the LST products of MODIS and Sentinel-3, and LST estimated using NSW showed a similar trend for LST changes during the seasons. With respect to the two absolute and comparative validations for the Sentinel-3 LST products, summer with the highest values of bias (?1.24 K), standard deviation (StDv = 2.66 K), and RMSE (2.43 K), and winter with the lowest ones (bias of 0.14 K, StDv of 1.13 K, and RMSE of 1.12 K) provided the worst and best results for the seasons in the period of 2018–2019, respectively. According to both absolute and comparative evaluation results, the Sentinel-3 SLSTR LST products provided reliable results for all seasons on a large temporal and spatial scale over our studied area.     

Monitoring of maize lodging using multi-temporal Sentinel-1 SAR data

Lodging is a common phenomenon in maize production, which seriously affects its yield, quality, and mechanical harvesting capacity. With good penetrating power, satellite radar can monitor crop growth even under cloudy weather conditions. In this study, a method based on the change in plant height before and after lodging in maize is proposed to calculate the lodging angle and monitor the lodging degree by using dual-polarization Sentinel-1A data. The results show that the optimal sensitive polarization combinations of maize plant height before and after lodging are VH/VV and VV, respectively. The lodging angle is calculated using the plant height inversion results before and after lodging. The overall accuracy of classifying lodging grade of maize is 67%. The proposed model based on lodging angle could effectively mapped the maize lodging range on a regional scale and classify the lodging grades.     

Estimating canopy LAI and chlorophyll of tropical forest plantation (North India) using Sentinel-2 data

With the free and full access to images from Sentinel-2 satellite, the interest to use this data for quantitative retrieval of vegetation parameters is ever-increasing. LAI and chlorophyll are two key variables which are desired for studying productivity, nutrient and stress status of vegetation. Studies carried out on croplands using simulated Sentinel-2 MSI and parametric approach have identified vegetation indices (VIs) with high sensitivity to LAI and chlorophyll. To test how Sentinel-2 red-edge based VIs perform for retrieval of LAI and Chlorophyll of tropical mixed forest canopies, this study has been performed. The field measurements of LAI and chlorophyll content were recorded in a total of 28 ESUs (Elementary Sampling Units) in Bhakra range in the Tarai Central Forest Division, Uttarakhand (India). The in-situ measurements were statistically correlated with Sentinel-2VIs and strength of correlation was validated using Predicted Residual Error Sum of Squares (PRESS) statistic. Field LAI corrected for foliage clumpiness effect improved correlation of VIs with LAI. Among all VIs tested, Normalized Difference Index (NDI) offered highest positive correlation (R² = 0.79, p < 0.05) with LAI while Red-Edge Chlorophyll Index (RECI) (R² = 0.83, RMSE = 0.24 g/m², p < 0.05) and Simple Ratio (SR) 740/705 (R² = 0.79, RMSE = 0.27 g/m², p < 0.05) were the most closely related to chlorophyll content. VIs with red-edge and NIR combinations offered best results.     

Exploring the potential of Sentinel-3 delay Doppler altimetry for enhanced detection of coastal currents along the Northwest Atlantic shelf

In this study, we evaluate Sentinel-3A satellite synthetic aperture radar (SAR) altimeter observations along the Northwest Atlantic coast, spanning the Nova Scotian Shelf, Gulf of Maine, and Mid-Atlantic Bight. Comparisons are made of altimeter sea surface height (SSH) measurements from three different altimeter data processing approaches: fully-focused synthetic aperture radar (FFSAR), un-focused SAR (UFSAR), and conventional low-resolution mode (LRM). Results show that fully-focused SAR data always outperform LRM data and are comparable or slightly better than the nominal un-focused SAR product. SSH measurement noise in both SAR-mode datasets is significantly reduced compared to LRM. FFSAR SSH 20-Hz noise levels, derived from 80-Hz FFSAR data, are lower than 20-Hz UFSAR SSH with 25% noise reduction offshore of 5 km, and 55–70% within 5 km of the coast. The offshore noise improvement is most likely due to the higher native along-track data posting rate (80 Hz for FFSAR, and 20 Hz for UFSAR), while the large coastal improvement indicates an apparent FFSAR data processing advantage approaching the coastlines. FFSAR-derived geostrophic ocean current estimates exhibit the lowest bias and noise when compared to in situ buoy-measured currents. Assessment at short spatial scales of 5–20 km reveals that Sentinel-3A SAR data provide sharper and more realistic measurement of small-scale sea surface slopes associated with expected nearshore coastal currents and small-scale gyre features that are much less well resolved in conventional altimetric LRM data.     

Impact of rain on wave retrieval from Sentinel-1 synthetic aperture radar images in tropical cyclones

The main objective of our work was to investigate the impact of rain on wave observations from C-band (~5.3 GHz) synthetic aperture radar (SAR) in tropical cyclones. In this study, 10 Sentinel-1 SAR images were available from the Satellite Hurricane Observation Campaign, which were taken under cyclonic conditions during the 2016 hurricane season. The third-generation wave model, known as Simulating WAves Nearshore (SWAN) (version 41.31), was used to simulate the wave fields corresponding to these Sentinel-1 SAR images. In addition, rainfall data from the Tropical Rainfall Measuring Mission satellite passing over the spatial coverage of the Sentinel-1 SAR images were collected. The simulated results were validated against significant wave heights (SWHs) from the Jason-2 altimeter and European Centre for Medium-Range Weather Forecasts data, revealing a root mean square error (RMSE) of ~0.5 m with a 0.25 scatter index. Winds retrieved from the VH-polarized Sentinel-1 SAR images using the Sentinel-1 Extra Wide-swath Mode Wind Speed Retrieval Model after Noise Removal were taken as prior information for wave retrieval. It was discovered that rain did indeed affect the SAR wave retrieval, as evidenced by the 3.21-m RMSE of SWHs between the SAR images and the SWAN model, which was obtained for the ~1000 match-ups with raindrops. The raindrops dampened the wave retrieval when the rain rate was < ~5 mm/hr; however, they enhanced wave retrieval for higher rain rates. It was also found that the portion of the rain-induced ring wave with a wave number > 0.05 rad/m (~125 m wavelength) was clearly observed in the SAR-derived wave spectra.     

Neural network prediction of the topside electron content over the Euro-African sector derived from Swarm-A measurements

This study presents the first prediction results of a neural network model for the vertical total electron content of the topside ionosphere based on Swarm-A measurements. The model was trained on 5 years of Swarm-A data over the Euro-African sector spanning the period 1 January 2014 to 31 December 2018. The Swarm-A data was combined with solar and geomagnetic indices to train the NN model. The Swarm-A data of 1 January to 30 September 2019 was used to test the performance of the neural network. The data was divided into two main categories: most quiet and most disturbed days of each month. Each category was subdivided into two sub-categories according to the Swarm-A trajectory i.e. whether it was ascending or descending in order to accommodate the change in local time when the satellite traverses the poles. Four pairs of neural network models were implemented, the first of each pair having one hidden layer, and the second of each pair having two hidden layers, for the following cases: 1) quiet day-ascending, 2) quiet day-descending, 3) disturbed day-ascending, and 4) disturbed day-descending. The topside vertical total electron content predicted by the neural network models compared well with the measurements by Swarm-A. The model that performed best was the one hidden layer model in the case of quiet days for descending trajectories, with RMSE = 1.20 TECU, R = 0.76. The worst performance occurred during the disturbed descending trajectories where the one hidden layer model had the worst RMSE = 2.12 TECU, (R = 0.54), and the two hidden layer model had the worst correlation coefficient R = 0.47 (RMSE = 1.57).In all cases, the neural network models performed better than the IRI2016 model in predicting the topside total electron content. The NN models presented here is the first such attempt at comparing NN models for the topside VTEC based on Swarm-A measurements.     

Classification of multitemporal Sentinel-2 data for field-level monitoring of rice cropping practices in Taiwan

Cadastral information of rice fields is important for monitoring cropping practices in Taiwan due to official initiatives. Remote sensing based rice monitoring has been a challenge for years because the size of rice fields is small, and crop mapping requires information of crop phenology, relating to spatiotemporal resolution of satellite data. This study aims to develop an approach for mapping rice-growing areas at field level using multi-temporal Sentinel-2 data in Taiwan. The data were processed for 2018, following four main steps: (1) construct time-series Enhanced Vegetation Index (EVI) and Land Surface Water Index (LSWI), (2) noise filtering of the time-series data using wavelet transform, (3) rice crop classification using information of crop phenology, and (4) parcel-based accuracy assessment of the mapping results. The parcel-to-parcel comparisons between mapping results and ground reference data indicated satisfactory results. These findings were confirmed by close agreement between satellite-derived rice area and government’s statistics. Although some factors, including mixed-pixel issues and cloud-cover effects, lowered the mapping accuracies of townships along the coastline, this study has demonstrated the efficacy of using multitemporal Sentinel-2 data to create a reliable database of rice-growing areas over a large and heterogeneous region. Such a quantitative information was important for updating rice crop maps and monitoring cropping practices.     

Integration of Landsat-8 and Sentinel-1 dataset to extract geological lineaments in complex formations of Tepal mountain area,Shahrood, north Iran

Recently, the detection and extraction of geological lineaments have become an essential analytical technique to find relationships between the characteristics and occurrence of hydrogeology, and tectonic studies. The use of remote sensing, with the progressive development of image enhancement techniques, provides an opportunity to produce more reliable and comprehensive lineament maps. In this paper, semi-automatic approach based on Landsat 8 and Sentinel 1 radar data is proposed for lineaments extraction and validation. The combined method of linear filtering and automatic line module ensures a high degree of accuracy resulting in a lineament map. Based on identified lineaments, Sentinel1 is more capable of detecting edges than Landsat8, but the primary orientation lineaments extracted from Landsat8 and Sentinel1 were different. So, by combining band6 of Landsat8, and VV and VH polarization of Sentinel1, the area lineaments were extracted with high accuracy. Rose diagram showed the extracted lineaments' orientation is in good compliance with the region's existing faults. Also, the formations' lineament length density has good consistent with the density of the faults in the geological map.     

An Artificial Neural Network based approach for estimation of rain intensity from spectral moments of a Doppler Weather Radar

By using a Doppler Weather Radar (DWR) at Shriharikota (13.66°N & 80.23°E), an Artificial Neural Network (ANN) based technique is proposed to improve the accuracy of rain intensity estimation. Three spectral moments of a Doppler spectra are utilized as an input data to an ANN. Rain intensity, as measured by the tipping bucket rain gauges around the DWR station, are considered as a target values for the given inputs. Rain intensity as estimated by the developed ANN model is validated by the rain gauges measurements. With the help of a developed technique, reasonable improvement in the estimation of rain intensity is observed. By using the developed technique, root mean square error and bias are reduced in the range of 34–18% and 17–3% respectively, compared to Z–R approach.     

TS-InSAR analysis for monitoring ground deformation in Lanzhou New District,the loess Plateau of China,from 2017 to 2019

Large-scale land creation projects involving the cutting of mountains to infill gullies for construction have been carried out in Lanzhou New District (LZND). However, there is an urgent need for comprehensive and detailed research on the spatiotemporal evolution of ground deformation in LZND. Based on Sentinel-1A SAR data, combined with the urban geological background, the ground deformation in LZND from 2017 to 2019 was analysed. Two independent, multi-temporal techniques, persistent scatterers interferometry (PS-InSAR) and the small baseline subset (SBAS-InSAR), were used to calculate the deformation time series, and the results were cross-verified. The time series-monitoring results of the SBAS and PS calculations exhibited strong consistency in LZND and verified the high reliability of the experimental results. The results showed the whole surface of the LZND from March 2017 to October 2019 maintained stability, and the deformation rate was primarily in the range of ?10 to 10 mm/year. However, ground deformation in the Xicha area was evident. The maximum annual deformation rates monitored by SBAS-InSAR and PS-InSAR were ?52.48 mm/year and ?56.35 mm/year, respectively. The most typical deformation areas include the built-up area and the land creation area. The surface subsidence area was concentrated in the filling area. The ground deformation range of LZND kept expanding and accelerating from 2017 to 2019. Land creation, urban construction, geology and precipitation were the primary factors contributing to local severe ground deformation. The results of this study provide reference for the regional urban planning in LZND.     

Validity assessment of SAMOSA retracking for fully-focused SAR altimeter waveforms

We demonstrate in this work how we can take advantage of known unfocused SAR (UF-SAR) retracking methods (e.g. the physical SAMOSA model) for retracking of fully-focused SAR (FF-SAR) waveforms. Our insights are an important step towards consistent observations of sea surface height, significant wave height and backscatter coefficient (wind speed) with both UF-SAR and FF-SAR. This is of particular interest for SAR altimetry in the coastal zone, since coastal clutter may be filtered out more efficiently in the high-resolution FF-SAR waveform data, which has the potential to improve data quality. We implemented a multi-mission FF-SAR altimetry processor for Sentinel-3 (S3) and Sentinel-6 Michael Freilich (S6), using a back-projection algorithm, and analysed ocean waveform statistics compared to multilooked UF-SAR. We find for Sentinel-3 that the averaged power waveforms of UF-SAR and FF-SAR over ocean are virtually identical, while for Sentinel-6 the FF-SAR power waveforms better resemble the UF-SAR zero-Doppler beam. We can explain and model the similarities and differences in the data via theoretical considerations of the waveform integrals. These findings suggest to use the existing UF-SAR SAMOSA model for retracking S3 FF-SAR waveforms but the SAMOSA zero-Doppler beam model for S6 FF-SAR waveforms, instead. Testing the outlined approach over short track segments, we obtain range biases between UF-SAR and FF-SAR lower than 2 mm and significant wave height biases lower than 5 cm.     

Exploiting the Sentinel-3 tandem phase dataset and azimuth oversampling to better characterize the sensitivity of SAR altimeter sea surface height to long ocean waves

Being the very first SAR mode altimeter tandem phase, the Sentinel-3 A/B tandem phase has provided an unprecedented opportunity to better characterize the sensitivity of SAR altimetry retrievals to high-frequency processes, such as long ocean waves. In this paper, we show that for some sea-state conditions, that are still to be precisely characterized, long ocean waves are responsible for high-frequency (spatial and temporal) coherent Sea Level Anomaly (SLA) signals. It is found that the peak wavelength corresponds to the dominant swell wavelength. Furthermore, the short time lag between S3-A and S3-B acquisitions allows performing cross-spectral analyses that reveal phase shifts consistent with waves travelling according to the wave dispersion relation. It is also demonstrated that the classical 20 Hz sampling frequency is insufficient to properly sample most swell-induced SLA signals and that aliasing can generate errors over the entire frequency spectrum, including at long wavelengths. These results advocate for the use of azimuth oversampling (40 Hz or 80 Hz). Low-pass filtering should be applied prior to any down-sampling to 20 Hz, in order to prevent long-wavelength errors induced by spectral leakage.     

A multilayer perceptron and multiclass support vector machine based high accuracy technique for daily rainfall estimation from MSG SEVIRI data

The current paper introduces a new multilayer perceptron (MLP) and support vector machine (SVM) based approach to improve daily rainfall estimation from the Meteosat Second Generation (MSG) data. In this study, the precipitation is first detected and classified into convective and stratiform rain by two MLP models, and then four multi-class SVM algorithms were used for daily rainfall estimation. Relevant spectral and textural input features of the developed algorithms were derived from the spectral MSG SEVIRI radiometer channels. The models were trained using radar rainfall data set colected over north Algeria. Validation of the proposed daily rainfall estimation technique was performed by rain gauge network data set recorded over north Algeria. Thus, several statistical scores were calculated, such as correlation coefficient (r), root mean square error (RMSE), mean error (Bias), and mean absolute error (MAE). The findings given by: (r = 0.97, bias = 0.31 mm, RMSE = 2.20 mm and MAE = 1.07 mm), showed a quite satisfactory relationship between the estimation and the respective observed daily precipitation. Moreover, the comparison of the results with those of two advanced techniques based on random forests (RF) and weighted ‘k’ nearest neighbor (WkNN) showed higher accuracy obtained by the proposed model.     

2021 Turkey mega forest Fires: Biodiversity measurements of the IUCN Red List wildlife mammals in Sentinel-2 based burned areas

Mega wildfires are one of the environmental disasters worldwide. This study evaluates the pre-fire species diversity and the indirect effects, including habitat loss for one of the largest wildfires in Manavgat (Antalya-Turkey) in 2021, with a two-step methodology. Here, (1) burnt areas in the Manavgat district (2021) were detected with remote sensing data from Sentinel-2A by delta Normalized Burn Ratio calculation for a selected area in Google Earth Engine, and (2) mammals' habitat vector data of International Union for Conservation of Nature (IUCN) Red List were integrated into Habitat and Biodiversity modelling of Terrset to analyze the alpha, beta, gamma diversity and Range Restriction Index for the wildfire region. In the total 4210 km² study area, 696 km² of the area was damaged by different fire severity; also, there were 56 mammal species' habitats here. These species include bats (i.e. Nyctalus leisleri), felids (i.e. Felis chaus), rodents (i.e. Rattus norvegicus) and large mammals (i.e. Ursus arctos). 88 % of these species are in IUCN's Least Concern category. The remaining classes are Near Threatened (3.7 %) and Vulnerable (7.4 %). This study also indicated that the burnt area's species richness does not totally consist of endemic species. Therefore, pre-fire species richness analyses of this study can be a base for further studies about the species' post-fire activity and occupancy.Furthermore, the 2021 mega wildfires show us the necessity of wildfire monitoring and risk studies in the entire Mediterranean ecosystem to evaluate the risks to the Sustainable Development Goals. Therefore, post-fire spatial data, fire migration monitorization, and recording of the species' activities should be performed temporally. In this way, the ability of wildlife's recovering, and the direct and indirect effects of the fire will be examined for ecosystems in the long term.     

An approach to multi-temporal MODIS image analysis using image classification and segmentation

This paper discusses an approach for river mapping and flood evaluation based on multi-temporal time series analysis of satellite images utilizing pixel spectral information for image classification and region-based segmentation for extracting water-covered regions. Analysis of MODIS satellite images is applied in three stages: before flood, during flood and after flood. Water regions are extracted from the MODIS images using image classification (based on spectral information) and image segmentation (based on spatial information). Multi-temporal MODIS images from “normal” (non-flood) and flood time-periods are processed in two steps. In the first step, image classifiers such as Support Vector Machines (SVM) and Artificial Neural Networks (ANN) separate the image pixels into water and non-water groups based on their spectral features. The classified image is then segmented using spatial features of the water pixels to remove the misclassified water. From the results obtained, we evaluate the performance of the method and conclude that the use of image classification (SVM and ANN) and region-based image segmentation is an accurate and reliable approach for the extraction of water-covered regions.     

An improved cyclic multi model-eXtreme gradient boosting (CMM-XGBoost) forecasting algorithm on the GNSS vertical time series

The study of GNSS vertical coordinate time series forecasting is helpful for monitoring the crustal plate movement, dam or bridge deformation monitoring, and global or regional coordinate system maintenance. The eXtreme Gradient Boosting (XGBoost) algorithm is a machine learning algorithm that can evaluate features, and it has a great potential and stability for long-span time series forecasting. This study proposes a multi-model combined forecasting method based on the XGBoost algorithm. The method constitutes a new time series as features through the fitting and forecasting results of the forecasting model. The XGBoost model is then used for forecasting. In addition, this method can obtain higher precision forecasting results through circulation. To verify the performance of the forecasting method, 1095 epochs of data in the Up coordinate of 16 GNSS stations are selected for the forecasting test. Compared with the CNN-LSTM model, the experimental results of our forecasting method show that the mean absolute error (MAE) values are reduced by 30.23 %～52.50 % and the root mean square error (RMSE) values are reduced by 31.92 %～54.33 %. The forecasting results have higher accuracy and are highly correlated to the original time series, which can better forecast the vertical movement of the GNSS stations. Therefore, the forecasting method can be applied to the up component of the GNSS coordinate time series.     

High-performance altimeter Doppler processing for measuring sea level height under varying sea state conditions

This paper describes an innovative method for processing nadir altimeter data acquired in Synthetic Aperture Radar (SAR) mode, enhancing the system performances over open ocean. Similarly to the current SAR data processing scheme, the so-called LR-RMC (Low Resolution with Range Migration Correction) method, originally designed by Phalippou and Demeester (2011), includes Doppler beam forming, Doppler shift correction and range correction. In LR-RMC, however, an alternative and less complex averaging (stacking) operation is used so that all the Doppler beams produced in a radar cycle (4 bursts of 64 beams for the open-burst Sentinel-3-mode altimeter) are incoherently combined to form a multi-beam echo. In that manner, contrarily to the narrow-band SAR technique, the LR-RMC processing enlarges the effective footprint to average out the effects of surface waves and particularly those from small sub-mesoscale structures (<1 km) that are known to impact SAR-mode performances. On the other hand, the number of averaged beams is as high as in current SAR-mode processing, thus providing a noise reduction at least equally good. The LR-RMC method has the added benefit of reducing the incoherent integration time with respect to the SAR-mode processing (50 ms compared to 2.5 s) limiting possible surface movement effects. By processing one year of Sentinel-3A SRAL SAR-mode data using the LR-RMC method, it is shown that the swell impact on the SAR altimeter performances is totally removed and that an improvement of 10–50% is obtained in the measurement noise of the sea surface height and significant wave height with respect to SAR mode. Additionally, observational capabilities over the middle scales are enhanced potentially allowing the ocean mesoscale features to be retrieved and observations assimilated more usefully in ocean models.     

A new method for improving the performance of an ionospheric model developed by multi-instrument measurements based on artificial neural network

There are remarkable ionospheric discrepancies between space-borne (COSMIC) measurements and ground-based (ionosonde) observations, the discrepancies could decrease the accuracies of the ionospheric model developed by multi-source data seriously. To reduce the discrepancies between two observational systems, the peak frequency (foF2) and peak height (hmF2) derived from the COSMIC and ionosonde data are used to develop the ionospheric models by an artificial neural network (ANN) method, respectively. The averaged root-mean-square errors (RMSEs) of COSPF (COSMIC peak frequency model), COSPH (COSMIC peak height model), IONOPF (Ionosonde peak frequency model) and IONOPH (Ionosonde peak height model) are 0.58 MHz, 19.59 km, 0.92 MHz and 23.40 km, respectively. The results indicate that the discrepancies between these models are dependent on universal time, geographic latitude and seasons. The peak frequencies measured by COSMIC are generally larger than ionosonde’s observations in the nighttime or middle-latitudes with the amplitude of lower than 25%, while the averaged peak height derived from COSMIC is smaller than ionosonde’s data in the polar regions. The differences between ANN-based maps and references show that the discrepancies between two ionospheric detecting techniques are proportional to the intensity of solar radiation. Besides, a new method based on the ANN technique is proposed to reduce the discrepancies for improving ionospheric models developed by multiple measurements, the results indicate that the RMSEs of ANN models optimized by the method are 14–25% lower than the models without the application of the method. Furthermore, the ionospheric model built by the multiple measurements with the application of the method is more powerful in capturing the ionospheric dynamic physics features, such as equatorial ionization, Weddell Sea, mid-latitude summer nighttime and winter anomalies. In conclusion, the new method is significant in improving the accuracy and physical characteristics of an ionospheric model based on multi-source observations.     

Temporal extrapolation of TEC using WNN during 2007–2018 and comparison against GIM,IRI2016 and Kriging

In this paper, a new method of temporal extrapolation of the ionosphere total electron content (TEC) is proposed. Using 3-layer wavelet neural networks (WNNs) and particle swarm optimization (PSO) training algorithm, TEC time series are modeled. The TEC temporal variations for next times are extrapolated with the help of training model. To evaluate the proposed model, observations of Tehran GNSS station (35.69°N, 51.33°E) from 2007 to 2018 are used. The efficiency of the proposed model has been evaluated in both low and high solar activity periods. All observations of the 2015 and 2018 have been removed from the training step to test the proposed model. On the other hand, observations of these 2 years are not used in network training. According to the F10.7, the 2015 has high solar activity and the 2018 has quiet conditions. The results of the proposed model are compared with the global ionosphere maps (GIMs) as a traditional ionosphere model, international reference ionosphere 2016 (IRI2016), Kriging and artificial neural network (ANN) models. The root mean square error (RMSE), bias, dVTEC = |VTEC_GPS ? VTEC_Model| and correlation coefficient are used to assess the accuracy of the proposed method. Also, for more accurate evaluation, a single-frequency precise point positioning (PPP) approach is used. According to the results of 2015, the maximum values of the RMSE for the WNN, ANN, Kriging, GIM and IRI2016 models are 5.49, 6.02, 6.34, 6.19 and 13.60 TECU, respectively. Also, the maximum values of the RMSE at 2018 for the WNN, ANN, Kriging, GIM and IRI2016 models are 2.47, 2.49, 2.50, 4.36 and 6.01 TECU, respectively. Comparing the results of the bias and correlation coefficient shows the higher accuracy of the proposed model in quiet and severe solar activity periods. The PPP analysis with the WNN model also shows an improvement of 1 to 12 mm in coordinate components. The results of the analyzes of this paper show that the WNN is a reliable, accurate and fast model for predicting the behavior of the ionosphere in different solar conditions.