DEEP LEARNING-BASED RECONSTRUCTION OF SPATIOTEMPORALLY FUSED SATELLITE IMAGES FOR SMART AGRICULTURE APPLICATIONS IN A HETEROGENEOUS AGRICULTURAL REGION

Abstract. Remote sensing offers spatially explicit and temporally continuous observational data of various land surface parameters such as vegetation index, land surface temperature, soil moisture, leaf area index, and evapotranspiration, which can be widely leveraged for various applications at different scales and contexts. One of the main applications is agricultural monitoring, where a smart system based on precision agriculture requires a set of satellite images with a high resolution, both in time and space to capture the phenological stages and fine spatial details, especially in landscapes with various spatial heterogeneity and temporal variation. These requirements sometimes cannot be provided by a single sensor due to the trade-off required between spatial and temporal resolutions and/or the influence of cloud cover. The data availability of new generation multispectral sensors of Landsat-8 (L8) and Sentinel-2 (S2) satellites offers unprecedented options for such applications. Given this, the current study aims to display how the synergistic use of these optical sensors can efficiently support such an application. Herein, this study proposes a deep learning spatiotemporal data fusion method to fill the need for predicting a dense time series of vegetation index with fine spatial resolution. The results show that the developed method creates more accurate fused NDVI time-series data that were able to derive phenological stages and characteristics in single-crop fields, while keeps more spatial details in such a heterogeneous landscape.

Download Full-text

Method for Mapping Rice Fields in Complex Landscape Areas Based on Pre-Trained Convolutional Neural Network from HJ-1 A/B Data

ISPRS International Journal of Geo-Information ◽

10.3390/ijgi7110418 ◽

2018 ◽

Vol 7 (11) ◽

pp. 418 ◽

Cited By ~ 12

Author(s):

Tian Jiang ◽

Xiangnan Liu ◽

Ling Wu

Keyword(s):

Neural Network ◽

Remote Sensing ◽

Time Series ◽

Deep Learning ◽

Convolutional Neural Network ◽

Vegetation Index ◽

Southern China ◽

Series Data ◽

Support Vector ◽

Complex Landscape

Accurate and timely information about rice planting areas is essential for crop yield estimation, global climate change and agricultural resource management. In this study, we present a novel pixel-level classification approach that uses convolutional neural network (CNN) model to extract the features of enhanced vegetation index (EVI) time series curve for classification. The goal is to explore the practicability of deep learning techniques for rice recognition in complex landscape regions, where rice is easily confused with the surroundings, by using mid-resolution remote sensing images. A transfer learning strategy is utilized to fine tune a pre-trained CNN model and obtain the temporal features of the EVI curve. Support vector machine (SVM), a traditional machine learning approach, is also implemented in the experiment. Finally, we evaluate the accuracy of the two models. Results show that our model performs better than SVM, with the overall accuracies being 93.60% and 91.05%, respectively. Therefore, this technique is appropriate for estimating rice planting areas in southern China on the basis of a pre-trained CNN model by using time series data. And more opportunity and potential can be found for crop classification by remote sensing and deep learning technique in the future study.

Download Full-text

Comparison of LAI Estimates from High Resolution Satellite Observations Using Different Biophysical Processors

Biology and Life Sciences Forum ◽

10.3390/iecag2021-09683 ◽

2021 ◽

Vol 3 (1) ◽

pp. 5

Author(s):

Federico Filipponi

Keyword(s):

Time Series ◽

Land Surface ◽

Agricultural Land ◽

Spectral Characteristics ◽

Earth Observation ◽

Estimation Accuracy ◽

Landsat 8 ◽

Biophysical Parameters ◽

Area Index ◽

Satellite Sensors

Earth observation provides timely and spatially explicit information about crop phenology and vegetation dynamics that can support decision making and sustainable agricultural land management. Vegetation spectral indices calculated from optical multispectral satellite sensors have been largely used to monitor vegetation status. In addition, techniques to retrieve biophysical parameters from satellite acquisitions, such as the Leaf Area Index (LAI), have allowed to assimilate Earth observation time series in numerical modeling for the analysis of several land surface processes related to agroecosystem dynamics. More recently, biophysical processors used to estimate biophysical parameters from satellite acquisitions have been calibrated for retrieval from sensors with different high spatial resolution and spectral characteristics. Virtual constellations of satellite sensors allow the generation of denser LAI time series, contributing to improve vegetation phenology estimation accuracy and, consequently, enhancing agroecosystems monitoring capacity. This research study compares LAI estimates over croplands using different biophysical processors from Sentinel-2 MSI and Landsat-8 OLI satellite sensors. The results are used to demonstrate the capacity of virtual satellite constellation to strengthen LAI time series to derive important cropland use information over large areas.

Download Full-text

Assessing the Impact of Satellite Revisit Rate on Estimation of Corn Phenological Transition Timing through Shape Model Fitting

Remote Sensing ◽

10.3390/rs11212558 ◽

2019 ◽

Vol 11 (21) ◽

pp. 2558 ◽

Cited By ~ 2

Author(s):

Emily Myers ◽

John Kerekes ◽

Craig Daughtry ◽

Andrew Russ

Keyword(s):

Time Series ◽

Vegetation Index ◽

Time Series Data ◽

Model Fitting ◽

Series Data ◽

Sigmoid Function ◽

Absolute Deviation ◽

Shape Model ◽

Area Index ◽

The Impact

Agricultural monitoring is an important application of earth-observing satellite systems. In particular, image time-series data are often fit to functions called shape models that are used to derive phenological transition dates or predict yield. This paper aimed to investigate the impact of imaging frequency on model fitting and estimation of corn phenological transition timing. Images (PlanetScope 4-band surface reflectance) and in situ measurements (Soil Plant Analysis Development (SPAD) and leaf area index (LAI)) were collected over a corn field in the mid-Atlantic during the 2018 growing season. Correlation was performed between candidate vegetation indices and SPAD and LAI measurements. The Normalized Difference Vegetation Index (NDVI) was chosen for shape model fitting based on the ground truth correlation and initial fitting results. Plot-average NDVI time-series were cleaned and fit to an asymmetric double sigmoid function, from which the day of year (DOY) of six different function parameters were extracted. These points were related to ground-measured phenological stages. New time-series were then created by removing images from the original time-series, so that average temporal spacing between images ranged from 3 to 24 days. Fitting was performed on the resampled time-series, and phenological transition dates were recalculated. Average range of estimated dates increased by 1 day and average absolute deviation between dates estimated from original and resampled time-series data increased by 1/3 of a day for every day of increase in average revisit interval. In the context of this study, higher imaging frequency led to greater precision in estimates of shape model fitting parameters used to estimate corn phenological transition timing.

Download Full-text

Assessment of different indices from Landsat time series in burned area mapping

10.5194/egusphere-egu2020-8397 ◽

2020 ◽

Author(s):

Jinxiu Liu

Keyword(s):

Time Series ◽

Burkina Faso ◽

Land Surface ◽

Vegetation Index ◽

Landsat Imagery ◽

Burned Area ◽

Free Access ◽

Landsat Images ◽

Area Index ◽

Burned Area Mapping

<p>Fire is recognized as an important land surface disturbance, as it influences terrestrial carbon cycle, climate and biodiversity. Accurate and efficient mapping of burned area is beneficial for social and environmental applications. Remote sensing plays a key role in detecting burned areas and active fires from reginal to global scales. Due to the free access to the Landsat archive, studies using dense time series of Landsat imagery for burned area mapping are appearing and increasing. However, the performance of Landsat time series when using different indices for burned area mapping has not been assessed. In this study, the objective was to identify which indices can detect burned area better when using Landsat time series in savanna area of southern Burkina Faso. We selected Burned Area Index (BAI), Normalized Burned Ratio (NBR), Normalized Difference Vegetation Index (NDVI), Global Environmental Monitoring Index (GEMI) for comparison as they are commonly used indices for burned area detection. The algorithm was based on breakpoint identification and burned pixel detection using harmonic model fitting with different indices Landsat time series. It was tested in savanna area in southern Burkina Faso over 16 years with 281 Landsat images ranging from October 2000 to April 2016.The same reference data was used to evaluate the performance of burned area detection with different indices Landsat time series. The result demonstrated that BAI was the most accurate in burned area detection from Landsat time series, followed by NBR, GEMI and NDVI.</p>

Download Full-text

NEW VEGETATION INDICES FOR FULL AND COMPACT POLARIMETRIC SAR DATA: IN PREPARATION FOR THE RADARSAT CONSTELLATION MISSION (RCM)

ISPRS Annals of Photogrammetry Remote Sensing and Spatial Information Sciences ◽

10.5194/isprs-annals-iv-3-w2-2020-41-2020 ◽

2020 ◽

Vol IV-3/W2-2020 ◽

pp. 41-46

Author(s):

D. Ratha ◽

D. Mandal ◽

S. Dey ◽

A. Bhattacharya ◽

A. Frery ◽

...

Keyword(s):

Time Series ◽

Vegetation Index ◽

Time Series Data ◽

Geodesic Distance ◽

Vegetation Indices ◽

Series Data ◽

Biophysical Parameters ◽

Area Index ◽

Plant Area Index ◽

Sar Data

Abstract. In this paper, we present two radar vegetation indices for full-pol and compact-pol SAR data, respectively. Both are derived using the notion of a geodesic distance between observation and well-known scattering models available in the literature. While the full-pol version depends on a generalized volume scattering model, the compact-pol version uses the ideal depolariser to model the randomness in the vegetation. We have utilized the RADARSAT Constellation Mission (RCM) time-series data from the SAMPVEX16-MB campaign in the Manitoba region of Canada for comparing and assessing the indices in terms of the change in the biophysical parameters as well. The compact-pol data for comparison is simulated from the full-pol RCM time series. Both the indices show better performance at correlating with biophysical parameters such as Plant Area Index (PAI) and Volumetric Water Content (VWC) for wheat and soybean crops, in comparison to the state-of-art Radar Vegetation Index (RVI) of Kim and van Zyl. These indices are timely for the upcoming release of the data from the RCM, which will provide data in both full and compact-pol modes, aimed at better crop monitoring from space.

Download Full-text

Phenological Monitoring of Paddy Crop Using Time Series MODIS Data

Proceedings ◽

10.3390/iecg2019-06205 ◽

2019 ◽

Vol 24 (1) ◽

pp. 19

Author(s):

C. Dineshkumar ◽

S. Nitheshnirmal ◽

Ashutosh Bhardwaj ◽

K. Nivedita Priyadarshini

Keyword(s):

Time Series ◽

Tamil Nadu ◽

Vegetation Index ◽

Google Earth ◽

Paddy Fields ◽

Series Data ◽

Local Minima ◽

Phenological Stages ◽

Local Maxima ◽

Modis Data

Rice is an important staple food crop worldwide, especially in India. Accurate and timely prediction of rice phenology plays a significant role in the management of water resources, administrative planning, and food security. In addition to conventional methods, remotely sensed time series data can provide the necessary estimation of rice phenological stages over a large region. Thus, the present study utilizes the 16-day composite Enhanced Vegetation Index (EVI) product with a spatial resolution of 250 m from the Moderate Resolution Imaging Spectroradiometer (MODIS) to monitor the rice phenological stages over Karur district of Tamil Nadu, India, using the Google Earth Engine (GEE) platform. The rice fields in the study area were classified using the machine learning algorithm in GEE. The ground truth was obtained from the paddy fields during crop production which was used for classifying the paddy grown area. After the classification of paddy fields, local maxima, and local minima present in each pixel of time series, the EVI product was used to determine the paddy growing stages in the study area. The results show that in the initial stage the pixel value of EVI in the paddy field shows local minima (0.23), whereas local maxima (0.41) were obtained during the peak vegetative stage. The results derived from the present study using MODIS data were cross-validated using the field data.

Download Full-text

The NDVI-CV Method for Mapping Evergreen Trees in Complex Urban Areas Using Reconstructed Landsat 8 Time-Series Data

Forests ◽

10.3390/f10020139 ◽

2019 ◽

Vol 10 (2) ◽

pp. 139 ◽

Cited By ~ 4

Author(s):

Yingying Yang ◽

Taixia Wu ◽

Shudong Wang ◽

Jing Li ◽

Farhan Muhanmmad

Keyword(s):

Remote Sensing ◽

Time Series ◽

Urban Areas ◽

Vegetation Index ◽

Normalized Difference Vegetation Index ◽

Time Series Data ◽

Series Data ◽

Distribution Data ◽

Landsat 8 ◽

Oxygen Balance

Evergreen trees play a significant role in urban ecological services, such as air purification, carbon and oxygen balance, and temperature and moisture regulation. Remote sensing represents an essential technology for obtaining spatiotemporal distribution data for evergreen trees in cities. However, highly developed subtropical cities, such as Nanjing, China, have serious land fragmentation problems, which greatly increase the difficulty of extracting evergreen trees information and reduce the extraction precision of remote-sensing methods. This paper introduces a normalized difference vegetation index coefficient of variation (NDVI-CV) method to extract evergreen trees from remote-sensing data by combining the annual minimum normalized difference vegetation index (NDVIann-min) with the CV of a Landsat 8 time-series NDVI. To obtain an intra-annual, high-resolution time-series dataset, Landsat 8 cloud-free and partially cloud-free images over a three-year period were collected and reconstructed for the study area. Considering that the characteristic growth of evergreen trees remained nearly unchanged during the phenology cycle, NDVIann-min is the optimal phenological node to separate this information from that of other vegetation types. Furthermore, the CV of time-series NDVI considers all of the phenologically critical phases; therefore, the NDVI-CV method had higher extraction accuracy. As such, the approach presented herein represents a more practical and promising method based on reasonable NDVIann-min and CV thresholds to obtain spatial distribution data for evergreen trees. The experimental verification results indicated a comparable performance since the extraction accuracy of the model was over 85%, which met the classification accuracy requirements. In a cross-validation comparison with other evergreen trees’ extraction methods, the NDVI-CV method showed higher sensitivity and stability.

Download Full-text

The Impact of Non-Photosynthetic Vegetation on LAI Estimation by NDVI in Mixed Grassland

Remote Sensing ◽

10.3390/rs12121979 ◽

2020 ◽

Vol 12 (12) ◽

pp. 1979

Author(s):

Dandan Xu ◽

Deshuai An ◽

Xulin Guo

Keyword(s):

Land Surface ◽

Vegetation Index ◽

Normalized Difference Vegetation Index ◽

Time Series Data ◽

Vegetation Indices ◽

Series Data ◽

Area Index ◽

Plant Area Index ◽

The Impact

Leaf area index (LAI) is widely used for algorithms and modelling in the field of ecology and land surface processes. At a global scale, normalized difference vegetation index (NDVI) products generated by different remote sensing satellites, have provided more than 40 years of time series data for LAI estimation. NDVI saturation issues are reported in agriculture and forest ecosystems at high LAI values, creating a challenge when using NDVI to estimate LAI. However, NDVI saturation is not reported on LAI estimation in grasslands. Previous research implies that non-photosynthetic vegetation (NPV) reduces the accuracy of LAI estimation from NDVI and other vegetation indices. A question arises: is the absence of NDVI saturation in grasslands a result of low LAI value, or is it caused by NPV? This study aims to explore whether there is an NDVI saturation issue in mixed grassland, and how NPV may influence LAI estimation by NDVI. In addition, in-situ measured plant area index (PAI) by sensors that detect light interception through the vegetation canopy (e.g., Li-cor LAI-2000), the most widely used field LAI collection method, might create bias in LAI estimation or validation using NDVI. Thus, this study also aims to quantify the contribution of green vegetation (GV) and NPV on in-situ measured PAI. The results indicate that NDVI saturation (using the portion of NDVI only contributed by GV) exists in grassland at high LAI (LAI threshold is much lower than that reported for other ecosystems in the literature), and that the presence of NPV can override the saturation effects of NDVI used to estimate green LAI. The results also show that GV and NPV in mixed grassland explain, respectively, the 60.33% and 39.67% variation of in-situ measured PAI by LAI-2000.

Download Full-text

A Data-Intensive Approach to Address Food Sustainability: Integrating Optic and Microwave Satellite Imagery for Developing Long-Term Global Cropping Intensity and Sowing Month from 2001 to 2015

Sustainability ◽

10.3390/su12083227 ◽

2020 ◽

Vol 12 (8) ◽

pp. 3227 ◽

Cited By ~ 3

Author(s):

Anjar Dimara Sakti ◽

Wataru Takeuchi

Keyword(s):

Time Series ◽

Land Surface ◽

Vegetation Index ◽

Time Series Data ◽

Global Analysis ◽

Local Level ◽

Series Data ◽

Pixel Resolution ◽

Cropping Intensity

It is necessary to develop a sustainable food production system to ensure future food security around the globe. Cropping intensity and sowing month are two essential parameters for analyzing the food–water–climate tradeoff as food sustainability indicators. This study presents a global-scale analysis of cropping intensity and sowing month from 2000 to 2015, divided into three groups of years. The study methodology integrates the satellite-derived normalized vegetation index (NDVI) of 16-day composite Moderate Resolution Imaging Spectroradiometer (MODIS) and daily land-surface-water coverage (LSWC) data obtained from The Advanced Microwave Scanning Radiometer (AMSR-E/2) in 1-km aggregate pixel resolution. A fast Fourier transform was applied to normalize the MODIS NDVI time-series data. By using advanced methods with intensive optic and microwave time-series data, this study set out to anticipate potential dynamic changes in global cropland activity over 15 years representing the Millennium Development Goal period. These products are the first global datasets that provide information on crop activities in 15-year data derived from optic and microwave satellite data. The results show that in 2000–2005, the total global double-crop intensity was 7.1 million km2, which increased to 8.3 million km2 in 2006–2010, and then to approximately 8.6 million km2 in 2011–2015. In the same periods, global triple-crop agriculture showed a rapid positive growth from 0.73 to 1.12 and then 1.28 million km2, respectively. The results show that Asia dominated double- and triple-crop growth, while showcasing the expansion of single-cropping area in Africa. The finer spatial resolution, combined with a long-term global analysis, means that this methodology has the potential to be applied in several sustainability studies, from global- to local-level perspectives.

Download Full-text

Tree Water Status in Apple Orchards Measured by Means of Land Surface Temperature and Vegetation Index (LST–NDVI) Trapezoidal Space Derived from Landsat 8 Satellite Images

Sustainability ◽

10.3390/su12010070 ◽

2019 ◽

Vol 12 (1) ◽

pp. 70 ◽

Cited By ~ 1

Author(s):

Mohammad Zare ◽

Katrin Drastig ◽

Manuela Zude-Sasse

Keyword(s):

Surface Temperature ◽

Land Surface Temperature ◽

Land Surface ◽

Satellite Images ◽

Vegetation Index ◽

Irrigation System ◽

Water Status ◽

Irrigation Management ◽

Landsat 8 ◽

Apple Orchards

In this study, the split window (SW) method was applied for land surface temperature (LST) retrieval using Landsat 8 in two apple orchards (Glindow, Altlandsberg). Four images were acquired during high demand of irrigation water from July to August 2018. After pre-processing images, the normalized difference vegetation index (NDVI) and LST were calculated by red, NIR, and thermal bands. The results were validated by interpolated infrared thermometer (IRT) measurements using the inverse distance weighting (IDW) method. In the next step, the temperature vegetation index (TVDI) was calculated based on the trapezoidal NDVI/LST space to determine the water status of apple trees in the case studies. Results show good agreement between interpolated LST using IRT measurements and remotely sensed LST calculation using SW in all satellite overpasses, where the absolute mean error was between 0.08 to 4.00 K and root mean square error (RMSE) values ranged between 0.71 and 4.23 K. The TVDI spatial distribution indicated that the trees suffered from water stress on 7 and 23 July and 8 August 2018 in Glindow apple orchard with the mean value of 0.69, 0.57, and 0.73, whereas in the Altlandsberg orchard on 17 August, the irrigation system compensated the water deficit as indicated by the TVDI value of 0.34. Moreover, a negative correlation between TVDI and vegetation water content (VWC) with correlation coefficient (r) of −0.81 was observed. The corresponding r for LST and VWC was equal to −0.89, which shows the inverse relation between water status and temperature-based indices. The results indicate that the LST and/or TVDI calculation using the proposed methods can be effectively applied for monitoring tree water status and support irrigation management in orchards using Landsat 8 satellite images without requiring ground measurements.

Download Full-text