Progress 06/01/23 to 05/31/24
Outputs
Target Audience:Our research in the fifth year included several aspects: 1) continuing to adjust solar angle effects on Advanced Baseline Imagery (ABI) on GOES-R observations; 2) developing algorithm and computer codes for near real time phenology monitoring from ABI and Visible Infrared Imaging Radiometer Suite (VIIRS) at 500m pixels across the US; 3) calibrating 500m VIIRS-ABI phenometrics to meet the USDA NASS crop progress report; and 4) developing a new algorithm to map crop types in near real time by fusing observations from ABI, VIIRS, and Harmonized Landsat and Sentinel-2 (HLS) data. The results from this research period were presented in four peer-reviewed journal papers (two were published and two were under review in the journal of Remote Sensing of Environment, the best journal in the remote sensing) and presented eight papers to audiences at the International Geoscience and Remote Sensing Symposium (IGARSS) in July 2023, American Geophysical Union (AGU) Fall Meeting in December 2023, AAG Great Plains-Rocky Mountain Division Annual Meeting in October 2023, and AAG Annual Meeting in April 2024. Moreover, we held project team meeting frequently with CO-I from USDA, which provided an opportunity for us to better understand the needs of USDA. Changes/Problems:The algorithm of a geospatial tool has developed for 30m crop progress (corn and soybean) monitoring and is ready for implementation. Because of the ABI surface reflectance product from NASA GoeNex has not been operationally produced in near real time, the actual implementation of the tool will be conducted during the period of our newly funded project. What opportunities for training and professional development has the project provided?This project provided an opportunity to train three PhD students. Specifically, Yu Shen joined this research project in late August 2019, Naeem Abbas Malik started to work on this project in September 2022, and Hoang Khuong Tran worked with part of his time in 2023. The PI held regular group meetings every two weeks with Shen, Malik, and Tran to discuss research progress, and to provide them guidance on solving related problems during his research activities. Moreover, the PI frequently met with them to explain the theories and algorithms in processing satellite data for crop monitoring. During this research period, Mr. Shen has published two peer-reviewed journal articles, submitted one journal article, and presented two papers at international conferences; Mr. Malik presented three papers at national and international conferences; and Mr. Tran developed field-based PhenoCam observations for evaluating satellite-derived crop progress. Further, Dr. Shuai An (a postdoc) joined the team in February 2023 with part of his time. Moreover, Shuai Gao (a postdoc) worked with part of his time in process ABI data. The PI frequently discussed with Drs An and Gao on the computer code development for crop monitoring, satellite data process, and crop phenology analyses. How have the results been disseminated to communities of interest?Results were presented at seven scientific conferences. Four journal articles were published, among which one paper was under revision based on review comments and another was under review. What do you plan to do during the next reporting period to accomplish the goals?Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing Landsat-8 OLI and Sentinel-2 MSI with GOES-R ABI data We have completed the task. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic HLS-ABI EVI2 and climatological crop phenometrics every week We have completed the task. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time We have completed the task. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. We were not able to set up the near real-time implementation for the NASS VegScape, because the ABI surface reflectance product have not been available in near real time from the NASA GeoNEX. This is expected to be completed during the period of our newly funded project.
Impacts
What was accomplished under these goals?
Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing HLS with GOES-R ABI data. (100% Accomplished) We continued to investigate the generation of synthetic time series of daily crop greenness. We added daily VIIRS NBAR (Nadir BRDF (Bidirectional reflectance distribution function)-Adjusted Reflectance) EVI2 time series (500m) to bridge HLS and ABI observations after we recognized that large ABI pixels (>1 km) could not always match well with HLS time series (30m). Thus, we calculated 3-day VIIRS EVI2 time series and fused it with ABI time series to generate synthetic ABI-VIIRS EVI2 time series, which was then fused with HLS time series. The result indicates that ABI-VIIRS-HLS EVI2 time series is effective to monitoring crop growth in the areas with heterogeneous vegetation types, where the corn and soybean were sparsely cultivated. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic HLS-ABI EVI2 and climatological crop phenometrics every week. (100% Accomplished) We continued to refine the algorithm and computer code in order to simulate and project potential EVI2 temporal trajectories. In particular, we employed VIIRS EVI2 time series to bridge HLS and ABI observations for reducing the impact of large ABI pixels (>1 km) on mismatch of 30m HLS crop field. We processed VIIRS time series in four VIIRS tiles (h10v04, h10v05, h11v04, h11v05) of daily VNP43IA4 and VNP43IA4 products from 1 January 2018 to 31 December 2020. The VIIRS time series was fused with ABI time series before the dates of implementing crop monitoring in a weekly basis. The result shows that the ABI-VIIRS-HLS EVI2 time series, in comparing with ABI-HLS data, significantly enhance the capability of mapping corn and soybean and monitoring their phenological events in near real time. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time. (100% Accomplished) We continued to develop and evaluate the algorithm and computer code to produce weekly crop progress in near real time using 30m ABI-VIIRS-HLS EVI2 time series. We also developed two new indices to separate phenological shifts and growth magnitude among different crop types. The first index is termed as "canopy Greenness and Water content index I (GW-I)", which is a ratio of kernel NDVI (Normalized Difference Vegetation Index) to SWIR (Short-Wave InfraRed) to distinguish the phenological shift of different crops. The second index is termed as "canopy Greenness and Water content II" (GW-II), which is a product of kernel NDVI and SWIR to separate growth magnitude of different crops. The GW-I is able to distinguish the phenological dates among different crop types, while GW-II is effective in separate soybean and other crop types. Combining GW-I phenometrics and GW-II growth magnitude allowed us to map corn and soybean in a weekly basis. We tested the algorithm for weekly corn and soybean mapping at 30m fields for nine HLS tiles across the Corn Belt. The result shows the overall accuracy of corn and soybean is higher than 90%. As a result, we was able to produce weekly crop progress for corn and soybean. Moreover, we also developed the algorithm and computer code to produce weekly crop progress at 500m pixels from ABI-VIIRS time series across entire CONUS. Although the spatial resolution is relatively coarser than the 30m field scale, the 500m ABI-VIIRS is able to be implemented with a wide spatial coverage (entire CONUS). However, 30m ABI-VIIRS-HLS EVI2 time series could currently be applicable at local regions because the data size is too large. The crop growth condition at 30m pixels was pot able to investigated because the ABI-HLS time series was not long enough for the extraction of climatological growth condition. Alternatively, we investigated the crop growth conditions at 500m pixels based on weekly MODIS (Moderate Resolution Imaging Spectroradiometer) EVI2 time series from 2000 to 2023. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. (90% Accomplished) We correlated the near real time predictions crop phenometrics at 30m pixels with NASS crop progress (areal percentage) of key phenological stages in corn (planting, emergence, silking, dough, dented, mature, and harvest) and soybean (planting, emergence, blooming, setting pods, dropping leaves, and harvest). The comparison revealed a very good alignment with R2 >0.95 except the two events: greenup vs planted date and senescence onset vs corn dough. We further calibrated the satellite phenometrics at 500m pixels using NASS crop progress. Specifically, 500m pixels with pure crop types were extracted using the 30m crop data layer (CDL) product from 2013-2022. The 500m satellite phenometrics at state levels were correlated to NASS crop progress to build up the correlation models. The result shows that 500m phenometrics were able to predict NASS crop progress with an error less than 10%. Because the ABI surface reflectance product have not been available in near real time from the Geostationary NASA (National Aeronautics and Space Administration) Earth Exchange (GeoNEX) project, we were not able to set up the near real-time implementation for the NASS VegScape. This is expected to be completed during the period of our newly funded project.
Publications
- Type:
Journal Articles
Status:
Under Review
Year Published:
2024
Citation:
Shen, Y., Zhang, X., Tran, K.H., Ye, Y., 2024, Near real-time crop mapping at field-scale by blending crop phenometrics with growth magnitude from multiple temporal and spatial satellite observations, Remote Sensing of Environment (under review)
- Type:
Journal Articles
Status:
Under Review
Year Published:
2024
Citation:
Gao, S., Zhang, X., Zhang, H.K., Shen, Y., Roy, D.P., Wang, W., Schaaf, C., 2024, A new constant scattering angle solar geometry definition for normalization of GOES-R ABI reflectance times series to support land surface phenology studies, Remote Sensing of Environment (under revision)
- Type:
Journal Articles
Status:
Published
Year Published:
2024
Citation:
Shen, Y., Zhang, X., Gao S., Zhang, H.K., Schaaf, C., Wang W., Ye, Y., Liu, Y., Tran, K.H., 2024, Analyzing GOES-R ABI BRDF-adjusted EVI2 time series by comparing with VIIRS observations over the CONUS, Remote Sensing of Environment, 302: 113972, https://doi.org/10.1016/j.rse.2023.113972
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2024
Citation:
An, S., Zhang, X., Ye, Y., 2024, The spatially heterogeneous trends and diverse physical environment drivers of the intra-annual maximum greenness rate of global natural vegetation in the past four decades, 2024 AAG Annual Meeting, 1621 April, Honolulu, USA.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Shen, Y., Zhang, X., Gao, S., Ye, Y., Liu, Y., Yang, Z., 2023, Monitoring crop progress at field scales in near-realtime by fusing harmonized Landsat and Sentinel-2 time sereis with geostationary satellite observations, GPRM Annual Meeting 2023 (2023 AAG Great Plains-Rocky Mountain Division Annual Meeting), 6-7 October 2023, Sioux Falls, South Dakota.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Malik, N.A., Zhang, X., 2023, Evaluating MODIS and VIIRS derived spring wheat and rice phenology in comparison with field-based crop progress data, GPRM Annual Meeting 2023 (2023 AAG Great Plains-Rocky Mountain Division Annual Meeting), 6-7 October 2023, Sioux Falls, South Dakota.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Malik, N.A., Zhang, X., 2023, Evaluating MODIS and VIIRS derived corn and soybean phenology in comparison with field-based crop progress data, AGU Fall Meeting 2023, 11-15 December 2023, San Francisco.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Gao, S., Zhang, X., Zhang, H.,Shen, Y., 2023, An optimal angle definition for BRDF normalization of GOES-R ABI reflectance for continental-scale land surface phenology (LSP) mapping, AGU Fall Meeting 2023, 11-15 December 2023, San Francisco.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
An, S., Zhang, X., Ye, Y., 2023, The primary drivers of the senescence rate of terrestrial natural vegetation across the global ecoregions are diverse, AGU Fall Meeting 2023, 11-15 December 2023, San Francisco.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Shen, Y., Zhang, X., Gao, S., Zhang, H., Schaaf, C., Wang, W., Ye, Y., Liu, Y., Tran, K.H., 2023, Investigation of GOES-R ABI EVI2 time series adjusted using different BRDF models, AGU Fall Meeting 2023, 11-15 December 2023, San Francisco
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Malik, N.A., Zhang, X., 2023, Analyzing trend in modis derived crop phenology for corn and soybean in comparison with field-based crop progress data, The International Geoscience and Remote Sensing Symposium (IGARSS), 16-21 July 2023, Pasadena, California.
- Type:
Journal Articles
Status:
Published
Year Published:
2023
Citation:
Shen, Y., Zhang, X., Yang, Z., Ye, Y., Wang, J., Gao, S., Liu, Y., Wang, W., Tran, K.H., Ju, J., 2023, Developing an operational algorithm for near-real-time monitoring of crop progress at field scales by fusing harmonized Landsat and Sentinel-2 time series with geostationary satellite observations, Remote Sensing of Environment, 296: 113729, https://doi.org/10.1016/j.rse.2023.113729
|
Progress 06/01/19 to 05/31/24
Outputs
Target Audience:The results from this research project were published in 8 peer-reviewed journal papers (two were under review, and four were published in the journal of Remote Sensing of Environment, the best journal in the remote sensing) and presented to audiences in 11 conferences. The conferences included the International Geoscience and Remote Sensing Symposium (IGARSS), American Geophysical Union (AGU) Fall Meeting, AAG (American Association of Geographers) Great Plains-Rocky Mountain Division Annual Meeting, and AAG Annual Meeting. Further, the results were also presented at the Digitalization and Informatics Division at the Food and Agriculture Organization (FAO) of the United Nations. Moreover, we held project team meetings every year with scientists from USDA. These meetings provided an opportunity for us to better understand the needs of USDA and showed how the project outputs could serve USDA. The outputs were taken as an example in PI's class. Further, the outputs eventually would be made publicly available for supporting individual farmers or government agencies for adopting management strategies, particularly in supporting and improving NASS' operational agricultural monitoring and assessment of crop progress and condition. Changes/Problems:The algorithm of a geospatial tool has developed for 30m crop progress (corn and soybean) monitoring and is ready for implementation. Because of the ABI surface reflectance product from NASA GoeNex has not been operationally produced in near real time, the actual implementation of the tool will be conducted during the period of our newly funded project. What opportunities for training and professional development has the project provided?This project provided an opportunity to train two PhD students and three postdocs. Specifically, Yu Shen (PhD student) worked on this research project from August 2019-May 2024, and Naeem Abbas Malik (PhD student) started to work on this project from September 2022-May 2024 (with other financial support). The postdocs who worked on this project in part time were Dr. Jianmin Wang from January 2022 to October 2022, Dr. Shuai An from February 2023-May 2025, and Dr. Shuai Gao from February 2024-May 2024. The PI held regular group meeting every two weeks to discuss their research progress, and to provide them guidance on solving related problems during their research activities. Moreover, the PI frequently (every week) met with them separately to explain the theory and algorithm in processing satellite data for crop monitoring. They all improved greatly their knowledge and skills in analyzing seasonal time series of satellite data for crop monitoring. It should be noted that Yu Shen (PhD student), who was the main player in this project, published five peer-reviewed journal articles as the first author (three in the Remote Sensing of Environment) and graduated in May 2024, which was really outstanding. How have the results been disseminated to communities of interest?The results from this research project were published in 8 peer-reviewed journal papers and presented in 11 conferences. The conferences included the International Geoscience and Remote Sensing Symposium (IGARSS), American Geophysical Union (AGU) Fall Meeting, and AAG Annual Meeting. What do you plan to do during the next reporting period to accomplish the goals?
Nothing Reported
Impacts
What was accomplished under these goals?
Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing HLS with GOES-R ABI data. (100% Accomplished) We performed the following tasks to accomplish this objective. First, we devoted a great effort to investing the process of GOES-R ABI data. (1) We investigated the impact of bidirectional reflectance distribution function (BRDF) on diurnal ABI surface reflectance and top of atmosphere (TOA) reflectance. The investigation was conducted based on a set of models including the semi-physical Ross-thick Li-sparse BRDF kernel-driven model and the simplified BRDF model to adjust diurnal ABI reflectance. (2) Because of large uncertainties in currently available ABI surface reflectance products, we proposed a simple and effective way to generate time series ABI EVI2 from TOA reflectance by comparing EVI2 with the corresponding normalized difference vegetation index (NDVI) and the normalized difference water index (NDWI). (3) We further proposed a novel approach, a Constant Scattering Angle (CSA) criterion, to reduce the impacts of ABI solar geometry changes on reflectance for deriving high quality vegetation indices. Second, we developed algorithms and computer codes to generate time series observations from multiple satellites. (1) We developed a new algorithm, Spatiotemporal Shape-Matching Model (SSMM), to fuse high spatial resolution data from the harmonized Landsat and Sentinel-2 (HLS) with high temporal resolution data from daily Visible Infrared Imaging Radiometer Suite (VIIRS), as well as GEOS-R ABI data to generate synthetic time series of daily 30m crop greenness. (2) We evaluated that SSMM algorithm by generating the synthetic 30m time series across 15 land cover types and various degrees of heterogeneity by fusing 500m VIIRS data and 30m HLS observations. (3) We verified the 3-day GOES-R ABI EVI2 across the United States and then fused with HLS to generate high spatiotemporal resolution HLS-ABI EVI2 time series. We evaluated the 30m HLS-ABI EVI2 using the ground-based 30-minute camera (PhenoCam) observations. Third, we added daily VIIRS EVI2 (500m) to bridge HLS and ABI observations becasue large ABI pixels (>1 km) could not always match well with HLS time series (30m). Thus, we first fused 3-day VIIRS EVI2 time series with ABI data to generate synthetic ABI-VIIRS EVI2 time series, which was further fused with HLS time series. The ABI-VIIRS-HLS EVI2 time series was effective to monitoring crop growth, including the areas with heterogeneous vegetation types. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic HLS-ABI EVI2 and climatological crop phenometrics every week. (100% Accomplished) First, we developed algorithms to simulate and project potential EVI2 temporal trajectories of future crop growth using available EVI2 observations from VIIRS data. The algorithm was tested to project the potential 500m EVI2 across the United States for detecting crop phenometics. The compute code was implemented weekly in our local machine to monitor crop growth in near real time with an overall uncertainty less than 10 days. Second, we improved the algorithms to project potential EVI2 temporal trajectories using available EVI2 observations from HLS-ABI data. Specifically, the HLS EVI2 was first fused with ABI EVI2 to generate the 30m HLS-ABI EVI2 for the preceding two years, which could capture various cases of crop rotations or crop-natural vegetation conversions in a local area. The HLS-ABI EVI2 during preceding two years was then considered as climatology, which was used for the prediction of future potential crop growth in a 30m pixel using a modified SSMM algorithm. This algorithm was tested for the prediction of crop growth every week in 2020 across IOWA. Third, we further refined the algorithms for better projecting potential EVI2 temporal trajectories. Specifically, we employed ABI-VIIRS-HLS time series to generate historical time series and then predict the future growth trajectories. The result showed that the ABI-VIIRS-HLS EVI2 time series significantly enhance the capability of mapping corn and soybean and monitoring their phenological events in near real time. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time. (100% Accomplished) First, we developed and evaluated the algorithms to produce weekly crop progress in near real time using 500m VIIRS time series. Compared with the stand phenological detections based on the entire year VIIRS observations (with a latency of one year), the near real time monitoring was able to identify the phenological timing in two weeks ahead with a uncertainty of 6-15 days at various crop growth stages. Second, we further evaluated the algorithm performances by producing weekly crop progress in near real time using 30m HLS-ABI EVI2 time series. We implemented the weekly crop progress monitoring across Iowa in 2020 to perform the operational practices. Third, we improved the algorithms to produce weekly crop progress in near real time using 30m ABI-VIIRS-HLS EVI2 time series. We also developed two new indices to separate phenological shifts and growth magnitude among different crop types, which were canopy Greenness and Water content index I (GW-I) and canopy Greenness and Water content II (GW-II). Combining GW-I phenometrics and GW-II growth magnitude allowed us to map corn and soybean in a weekly basis. We tested the algorithm for weekly corn and soybean mapping at 30m fields for nine HLS tiles across the Corn Belt. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. (90% Accomplished) First, we collected USDA NASS crop progress in 2018 to evaluate 30-m corn and soybean phenometrics detected from historical HLS-VIIRS time series. The HLS-VIIRS phenometrics were separately correlated to the corresponding NASS corn and sybean growth stages. Their statistical regression correlations (R2) were larger than 0.9, indicating the robustness of HLS-VIIRS in crop phenology detections. They were also evaluated using the PhenoCam observations, showing that their mean absolute difference (MAD) was less than 5.3 days in all phenometrics. Second, we correlated the HLS-ABI near real time predictions at 30m pixels with NASS crop progress (areal percentage) of key phenological stages in corn and soybean over Iowa. The HLS-ABI predictions of phenometrics aligned well with corresponding NASS crop phenological stages with slope close to 1 and R2>0.95 for both corn and soybean. Third, we also validated the results of HLS-ABI near real time prediction using PhenoCam time series. The results showed that the near real time predictions were significantly correlated with PhenoCam observations (R2=0.96, P-value <0.001) with an overall MAD of 8 days in all phenometrics. Fourth, we further calibrated the VIIRS phenometrics at 500m pixels using NASS crop progress. Specifically, 500m pixels with pure crop types were extracted using the 30m crop data layer (CDL) product from 2013-2022. The 500m VIIRS phenometrics at state levels were correlated to NASS crop progress to build up the correlation models. The result shows that 500m phenometrics were able to predict NASS crop progress with an error less than 10%. Finally, we tested the computer code in our local machine for monitoring crop progress in near real time. Because the ABI surface reflectance product have not been available in near real time from the Geostationary NASA Earth Exchange (GeoNEX) project, we will set up the near real-time implementation for the NASS VegScape during the period of our newly funded project.
Publications
- Type:
Journal Articles
Status:
Under Review
Year Published:
2024
Citation:
Gao, S., Zhang, X., Zhang, H.K., Shen, Y., Roy, D.P., Wang, W., Schaaf, C., 2024, A new constant scattering angle solar geometry definition for normalization of GOES-R ABI reflectance times series to support land surface phenology studies, Remote Sensing of Environment (under revision)
- Type:
Journal Articles
Status:
Published
Year Published:
2024
Citation:
Shen, Y., Zhang, X., Gao S., Zhang, H.K., Schaaf, C., Wang W., Ye, Y., Liu, Y., Tran, K.H., 2024, Analyzing GOES-R ABI BRDF-adjusted EVI2 time series by comparing with VIIRS observations over the CONUS, Remote Sensing of Environment, 302: 113972, https://doi.org/10.1016/j.rse.2023.113972
- Type:
Journal Articles
Status:
Under Review
Year Published:
2024
Citation:
Shen, Y., Zhang, X., Tran, K.H., Ye, Y., 2024, Near real-time crop mapping at field-scale by blending crop phenometrics with growth magnitude from multiple temporal and spatial satellite observations, Remote Sensing of Environment (under review)
- Type:
Journal Articles
Status:
Published
Year Published:
2023
Citation:
Shen, Y., Zhang, X., Yang, Z., Ye, Y., Wang, J., Gao, S., Liu, Y., Wang, W., Tran, K.H., Ju, J., 2023, Developing an operational algorithm for near-real-time monitoring of crop progress at field scales by fusing harmonized Landsat and Sentinel-2 time series with geostationary satellite observations, Remote Sensing of Environment, 296: 113729, https://doi.org/10.1016/j.rse.2023.113729
An, S., Zhang, X., Ye, Y., 2024, The spatially heterogeneous trends and diverse physical environment drivers of the intra-annual maximum greenness rate of global natural vegetation in the past four decades, 2024 AAG Annual Meeting, 1621 April, Honolulu, USA. Shen, Y., Zhang, X., Yang, Z., Ye, Y., Wang, J., Gao, S., Liu, Y., Wang, W., Tran, K.H., Ju, J., 2023, Developing an operational algorithm for near-real-time monitoring of crop progress at field scales by fusing harmonized Landsat and Sentinel-2 time series with geostationary satellite observations, Remote Sensing of Environment, 296: 113729, https://doi.org/10.1016/j.rse.2023.113729
An, S., Zhang, X., Ye, Y., 2024, The spatially heterogeneous trends and diverse physical environment drivers of the intra-annual maximum greenness rate of global natural vegetation in the past four decades, 2024 AAG Annual Meeting, 1621 April, Honolulu, USA. Shen, Y., Zhang, X., Yang, Z., Ye, Y., Wang, J., Gao, S., Liu, Y., Wang, W., Tran, K.H., Ju, J., 2023, Developing an operational algorithm for near-real-time monitoring of crop progress at field scales by fusing harmonized Landsat and Sentinel-2 time series with geostationary satellite observations, Remote Sensing of Environment, 296: 113729, https://doi.org/10.1016/j.rse.2023.113729
An, S., Zhang, X., Ye, Y., 2024, The spatially heterogeneous trends and diverse physical environment drivers of the intra-annual maximum greenness rate of global natural vegetation in the past four decades, 2024 AAG Annual Meeting, 1621 April, Honolulu, USA. https://doi.org/10.1016/j.rse.2023.113729
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2024
Citation:
An, S., Zhang, X., Ye, Y., 2024, The spatially heterogeneous trends and diverse physical environment drivers of the intra-annual maximum greenness rate of global natural vegetation in the past four decades, 2024 AAG Annual Meeting, 1621 April, Honolulu, USA.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Shen, Y., Zhang, X., Gao, S., Ye, Y., Liu, Y., Yang, Z., 2023, Monitoring crop progress at field scales in near-realtime by fusing harmonized Landsat and Sentinel-2 time sereis with geostationary satellite observations, GPRM Annual Meeting 2023 (2023 AAG Great Plains-Rocky Mountain Division Annual Meeting), 6-7 October 2023, Sioux Falls, South Dakota.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Malik, N.A., Zhang, X., 2023, Evaluating MODIS and VIIRS derived spring wheat and rice phenology in comparison with field-based crop progress data, GPRM Annual Meeting 2023 (2023 AAG Great Plains-Rocky Mountain Division Annual Meeting), 6-7 October 2023, Sioux Falls, South Dakota.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Malik, N.A., Zhang, X., 2023, Evaluating MODIS and VIIRS derived corn and soybean phenology in comparison with field-based crop progress data, AGU Fall Meeting 2023, 11-15 December 2023, San Francisco.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Gao, S., Zhang, X., Zhang, H.,Shen, Y., 2023, An optimal angle definition for BRDF normalization of GOES-R ABI reflectance for continental-scale land surface phenology (LSP) mapping, AGU Fall Meeting 2023, 11-15 December 2023, San Francisco.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
An, S., Zhang, X., Ye, Y., 2023, The primary drivers of the senescence rate of terrestrial natural vegetation across the global ecoregions are diverse, AGU Fall Meeting 2023, 11-15 December 2023, San Francisco.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Shen, Y., Zhang, X., Gao, S., Zhang, H., Schaaf, C., Wang, W., Ye, Y., Liu, Y., Tran, K.H., 2023, Investigation of GOES-R ABI EVI2 time series adjusted using different BRDF models, AGU Fall Meeting 2023, 11-15 December 2023, San Francisco
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Malik, N.A., Zhang, X., 2023, Analyzing trend in modis derived crop phenology for corn and soybean in comparison with field-based crop progress data, The International Geoscience and Remote Sensing Symposium (IGARSS), 16-21 July 2023, Pasadena, California.
|
Progress 06/01/22 to 05/31/23
Outputs
Target Audience:Our research in the fourth year included several aspects: 1) developing algorithm and computer codes for near real time phenology monitoring from Advanced Baseline Imagery (ABI) on GOES-R observations and Harmonized Landsat and Sentinel-2 (HLS) data; 2) analyzing spatial variations of HLS-ABI-derived near real time crop progress in corn and soybean fields over Iowa; 3) comparing crop phenometrics from HLS-ABI near-real time monitoring with standard phenology detections, corn and soybean progress in PhenoCam sites, and NASS measurements of crop progresses; and 4) investigating the effect of bi-directional reflectance distribution function on ABI observations. The results were presented in two peer-reviewed journal papers (one was submitted to Remote Sensing of Environment, the best journal in the remote sensing, and another one is under internal review) and presented to audiences at the American Geophysical Union (AGU) Fall Meeting in December 2022, and the Digitalization and Informatics Division at the Food and Agriculture Organization (FAO) of the United Nations in April 2022. Moreover, we held project team meeting twice with scientists from USDA, which provided an opportunity for us to better understand the needs of USDA. Changes/Problems:We actually started this project in August 2019 after a PhD student (Yu Shen) joined our team, although the project was funded to start on 06/01/2019. A PhD student (Naeem Abbas Malik) joined in September 2022. Because of the impact of COVID-19, the potential post-doctoral student was not able to travel to the US. As a result, the position of part time post-doctoral student was started in 2022, which delayed the progress of this project to some degree. Moreover, one of or Co-PI, Dr. Emmanuel Byamukama, left SDSU in January 2022, which could affect our field observations. To reduce this impact, the field data will be replaced using PhenoCam observations. Besides, Dr. Shuai An joined this team in February 2023 to analyze crop phenology. What opportunities for training and professional development has the project provided?This project provided an opportunity to train two graduate students. Specifically, one PhD student (Yu Shen) joined this research project in late August 2019 and another PhD student (Naeem Abbas Malik) started to work on this project in September 2022. The PI held regular meetings every two weeks with Mr. Shen and Mr. Malik to discuss research progress, and to provide them guidance on solving related problems during his research activities. Moreover, the PI frequently met with them to explain the theories and algorithms in processing satellite data for crop monitoring. As a result, Mr. Shen has published two peer-reviewed journal articles, submitted one journal article, and presented three papers at international conferences. Further, Dr. Jianmin Wang, as a postdoc, started to work on this project in January 2022 and left in October 2022. Dr. Shuai An joined the team in February 2023. The PI frequently discussed with Dr. Wang and Dr. Shuai on the computer code development for crop monitoring and crop phenology analyses. How have the results been disseminated to communities of interest?Results were presented at three scientific conferences. Four journal articles were published, one paper was under review and another was almost ready to be submitted What do you plan to do during the next reporting period to accomplish the goals?Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing Landsat-8 OLI and Sentinel-2 MSI with GOES-R ABI data We have completed the task. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic HLS-ABI EVI2 and climatological crop phenometrics every week We will improve the computer code to reduce the computing time for simulation although the algorithm has completed. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time We will extend the implementation of real time monitoring across the entire study area (ND, SD, NE, MN, and IA) and evaluate spatial and temporal pattern in corn and soybean progress. We will further investigate the crop condition from near real time HLS-ABI EVI2 data. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. We will continue to investigate the crop progress using PhenoCam observations, calibrate the HLS-ABI observations with USDA NASS crop progress, and work with USDA NASS to distribute the product.
Impacts
What was accomplished under these goals?
Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing HLS with GOES-R ABI data. (100% Accomplished) We continued to investigate the generation of synthetic time series of daily crop greenness. 1) We investigated the effect of bi-directional reflectance distribution function (BRDF) on GOES-R ABI EVI2 (two-band enhanced vegetation index) by comparing three BRDF models. 2) We analyzed the influences of ABI view zenith angle (VZA) on EVI2 across different ecosystems. The results indicated that ABI EVI2 has much higher opportunities to product high quality temporal observations than VIIRS (Visible Infrared Imaging Radiometer Suite) data. The ABI EVI2 can be adjusted well using BRDF models based on observations within 3 or 5 days, while VIIRS EVI2 is adjusted with 16-day observations. Moreover, ABI EVI2 is barely affected by VZA in croplands although the effect is evident in other ecosystems. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic HLS-ABI EVI2 and climatological crop phenometrics every week. (90% Accomplished) We continued to improve the algorithm and compute code to simulate and project potential EVI2 temporal trajectories using available EVI2 observations from HLS-ABI data. We investigated a new approach to replace the climatological phenometrics to avoid the uncertainties caused by crop rotation. In particular, the HLS EVI2 was fused with ABI EVI2 first to generate the 30m HLS-ABI EVI2 for the preceding two years, which could represent various cases of crop rotations or crop-natural vegetation conversions in a local area. The HLS-ABI EVI2 during preceding two years was considered as climatology and used for the prediction of future potential crop growth in a 30m pixels. This algorithm was tested for the prediction of crop (corn and soybean) growth every week in 2020 across IOWA. The result showed the robust of the updated algorithm. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time. (75% Accomplished) We continued to develop and evaluate the algorithm and computer code to produce weekly crop progress in near real time using 30m HLS-ABI EVI2 time series. After completing the computer code, we implemented the weekly crop progress monitoring across Iowa in 2020 to perform the operational practices. To match with the temporal observations in NASS crop progress, the algorithm implementation was started from 1 March 2020 to 15 December 2020 on a weekly basis. Although the algorithm was able to predict all crop phenological events for an entire growing cycle in each weekly implementation, the short-term prediction did not record events detected more than 28 days ahead to avoid large uncertainties. The results of near real time prediction were validated using two different datasets: (1) standard phenology detections (from HLS-ABI EVI2 observations during July 2019-June 2021), and (2) PhenoCam time series. The results showed that the near real time predictions (1) produced similar crop progress to the standard detections with a mean absolute difference (MAD) less than 10 days in majority pixels; 2) were significantly correlated with PhenoCam observations (R2=0.96, P-value <0.001) with the overall MAD in six phenometrics of 8 days. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. (70% Accomplished) We correlated the HLS-ABI near real time predictions with NASS crop progress (areal percentage) of key phenological stages in corn (planting, emergence, silking, dough, dented, mature, and harvest) and soybean (planting, emergence, blooming, setting pods, dropping leaves, and harvest) over Iowa. The weekly cumulative areal percentage for each individual HLS-ABI real time prediction in 2020 was separately calculated over corn and soybean fields. It was then correlated to NASS crop progress. The HLS-ABI predictions of four phenometrics (greenup onset, maturity onset, mid-senescence phase, and dormancy onset) aligned well correspondingly with four NASS crop phenological stages (emergence, silking/blooming, mature/dropping leaves, and harvest) with slope close to 1 and R2>0.95 for both corn and soybean. The HLS-ABI greenup onset closely tracked the NASS emerged dates with 3 and 5 days ahead for corn and soybean, respectively. The HLS-ABI maturity onset agrees well with the corn silking stage and soybean blooming stage with 4 and 3 days ahead. The HLS-ABI mid-senescence phase is 4 and 2 days earlier than corn mature and soybean dropping leave stages, respectively. The HLS-ABI dormancy onset closely follows the NASS harvesting timing with 1 day and 3 days earlier on average.
Publications
- Type:
Journal Articles
Status:
Other
Year Published:
2023
Citation:
Shen, Y., Zhang, X., Gao, S., Zhang, H., Wang, W. 2023. Analyzing BRDF-adjusted EVI2 using the Geostationary Satellite and VIIRS observations in the COUNS Region. International Journal of Applied Earth Observation and Geoinformation (in preparation)
- Type:
Journal Articles
Status:
Under Review
Year Published:
2023
Citation:
Shen, Y., Zhang, X., Yang, Z., Ye, Y., Wang, J., Gao, S., Liu, Y., Wang, W., Tran, K.H., Ju, J. 2023. Developing an operational algorithm for near-real-time monitoring of crop progress at field scales by fusing harmonized Landsat and Sentinel-2 time series with geostationary satellite observations. Remote Sensing of Environment (under review)
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
Shen, Y., Zhang, X., Yang, Z., Wang, J., Ye, Y., Wang, W. 2022. A novel algorithm for near real time crop Progress Monitoring at Field Scales by fusing observations from harmonized Landsat and Sentinel-2 and geostationary satellites. AGU Fall Meeting 2022. 12-16 December. Chicago, IL.
- Type:
Other
Status:
Other
Year Published:
2022
Citation:
Zhang, X. 2022. Operational monitoring of land surface phenology from polar-orbiting and geostationary satellites in support of sustainable agriculture, forestry, and rangeland management. Invited presentation to the Digitalization and Informatics Division at the Food and Agriculture Organization (FAO) of the United Nations. 27 April. Virtual.
- Type:
Journal Articles
Status:
Published
Year Published:
2022
Citation:
Shen, Y., Zhang, X., Yang, Z. 2022. Mapping corn and soybean phenometrics at field scales over the United States corn belt by fusing time series of Landsat 8 and Sentinel-2 data with VIIRS data. ISPRS Journal of Photogrammetry and Remote Sensing. 186, 55-69. https://doi.org/10.1016/j.isprsjprs.2022.01.023
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2021
Citation:
Shen, Y., and Zhang, X. 2021. B15I-1548 fusing new generation geostationary satellite observations with Landsat-8 and Sentinel-2 time series for monitoring land surface phenology. AGU Fall Meeting. Dec 13-17. New Orleans, LA and Virtual.
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Gao, F., and Zhang, X. 2021. Mapping crop phenology in near real-time using satellite remote sensing: Challenges and opportunities. Journal of Remote Sensing. Volume 2021, Article ID 8379391. https://doi.org/10.34133/2021/8379391
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Shen, Y., Zhang, X., Wang, W., Nemani, R., Ye, Y., and Wang, J. 2021. Fusing geostationary satellite observations with harmonized Landsat-8 and Sentinel-2 time series for monitoring field-scale land surface phenology. Remote Sensing. 13(21), 4465. https://doi.org/10.3390/rs13214465
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Zhang, X., Gao, F., Wang, J., and Ye, Y. 2021. Investigation of a spatiotemporal shape-matching model for generating synthetic high spatiotemporal resolution time series of satellite data. International Journal of Applied Earth Observation and Geoinformation. 104. https://doi.org/10.1016/j.jag.2021.102545
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2020
Citation:
Shen, Y., Zhang, X., Yang, Z. 2020. Mapping crop phenological metrics at field scales by fusing time series of VIIRS and HLS over the United States corn belt. AGU Fall Meeting. 1-17 December. Online.
|
Progress 06/01/21 to 05/31/22
Outputs
Target Audience:Our research in the third year included several aspects: 1) Comparing the satellite observations of field scale (30m) corn and soybean progress data with NASS reports based on field observations for the 2018 growing season across the Corn belt; 2) analyzing high spatial and temporal resolution time series of vegetation index fused from geostationary satellites with Landsat-8 and Sentinel-2 observations for field scale phenology detection; and 3) developing algorithm and computer codes for near-real time phenology monitoring. The results were published in four peer-reviewed articles and presented to audiences at the American Geophysical Union (AGU) Fall Meeting in December 2021. Moreover, we held project team meeting twice with scientists from USDA, which provided an opportunity for us to understand better the needs of USDA. Changes/Problems:We actually started this project in August 2019 after a PhD student joined our team, although the project was funded to start on 06/01/2019. Because of the impact of COVID-19, the potential post-doctoral student was not able to travel to the US. As a result, the position of part time post-doctoral student was started in 2022, which delayed the progress of this project to some degree. Moreover, one of or Co-PI, Dr. Emmanuel Byamukama, left SDSU in January 2022, which could affect our field observations. To reduce this impact, the field data will be replaced using PhenoCam observations. What opportunities for training and professional development has the project provided?This project provided an opportunity to train one graduate student. Specifically, a PhD student (Yu Shen) joined this research project in late August 2019. The PI held regular meetings every two weeks to discuss Mr. Shen's course work and research progress, and to provide him guidance on solving related problems during his research activities. Moreover, the PI frequently met with Mr. Shen to explain the theories and algorithms in processing satellite data for crop monitoring. As a result, Mr. Shen has published two peer-reviewed journal articles and presented one paper at international conference. Further, Dr. Jianmin Wang worked as a postdoc on this project with a 10% effort. The PI also frequently discussed with Dr. Wang on the computer code development for crop monitoring. How have the results been disseminated to communities of interest?Results were presented at a scientific conference. Four journal articles were published. What do you plan to do during the next reporting period to accomplish the goals?Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing Landsat-8 OLI and Sentinel-2 MSI with GOES-R ABI data We will continue to investigate the 10-minute GOES-R ABI observations and replace the TOA reflectance using surface reflectance. Currently, the ABI surface reflectance product is still under improvement in NASA AMS. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic OLI-MSI-ABI EVI2 and climatological crop phenometrics every week We will keep working on the enhancement of both algorithms and computer codes for simulating temporal trajectory of crop greenness development, which is expected to reduce the uncertainty in detecting crop phenometrics. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time We will continue to modify our computer code to automatically detect weekly crop progress using HLS and ABI time series in the western Corn Belt and evaluate the accuracy by comparing with field observations. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. We will continue to investigate the crop progress using PhenoCam observations.
Impacts
What was accomplished under these goals?
Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing Landsat-8 OLI and Sentinel-2 MSI with GOES-R ABI data. (95% Accomplished) We continued to investigate the generation of synthetic time series of daily crop greenness. 1) We verified the 3-day time series GOES-R ABI EVI2 (two-band enhanced vegetation index) from observations in a 10-minute interval across the United States. 2) We evaluated our newly developed data fusion algorithm of Spatiotemporal Shape-Matching Model (SSMM), which was used to generate the synthetic 30m time series across 15 land cover types and various degrees of heterogeneity by fusing 500m Visible Infrared Imaging Radiometer Suite (VIIRS) data and 30m harmonized Landsat-8 OLI and Sentinel-2 MSI (HLS) observations. 3) We analyzed phenology detections at 30m pixels by fusing HLS and ABI EVI2 observations to generate high spatiotemporal resolution HLS-ABI time series. The results indicated that ABI EVI2 time series was robust in tracking the temporal shape of vegetation growing cycles. It was successfully used to fill the cloud-contaminated observations in the HLS time series with the SSMM algorithm, leading to the generation of synthetic time series of daily 30m HLS-ABI crop greenness. Compared with ground-based 30-minute camera (PhenoCam) observations in the states of Wisconsin and Michigan, the RMSE (Root Mean Square Error) in green-up onset was 27 days from HLS, 13 days from ABI, and 4 days from HLS-ABI. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic OLI-MSI-ABI EVI2 and climatological crop phenometrics every week. (75% Accomplished) We continued to improve the algorithm and compute code to simulate and project potential EVI2 temporal trajectories using available EVI2 observations from VIIRS data. The improved algorithm was tested to project the potential 500m EVI2 time series across the United States for the detection of crop phenometics, which was implemented weekly. The projected EVI2 time series was able to monitor crop growth in near real time with an overall uncertainty less than 10 days, although the uncertainty varied with crop growth stage. The developed algorithm was also set up to evaluate the simulation of 30m EVI2 time series separately from the fused VIIRS and HLS time series, as well as ABI and HLS time in the Middle Western US. The preliminary implementation demonstrated that the 30m crop growth trajectory was reasonably produced at a weekly base. The accuracy of the performance will be investigated by comparing with simulated values with actual observations. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time. (60% Accomplished) We continued to develop and evaluate the algorithm and computer code to produce weekly crop progress in near real time using 500m VIIRS time series. Compared with the stand phenological detections based on the entire year VIIRS observations (with a latency of one year), the near real time monitoring was able to identify the phenological timing in two weeks ahead with a difference of 6-15 days at various vegetation growing stages and in two weeks late with a difference less than 8 days except for the senescence stage. We also investigated the capability of producing 30-m time series from fused ABI and HLS time series. The computer code has been able to produce reasonable results of crop progress by visually investigating the spatial pattern. Statistical analyses and evaluations are under way and the algorithm improvement is likely needed. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. (50% Accomplished) We collected USDA NASS crop progress in 2018 that was used to evaluate 30-m corn and soybean phenometrics detected from historical HLS-VIIRS time series. The HLS-VIIRS six phenometrics (greenup onset, mid-greenup phase, maturity onset, senescence onset, mid-senescence phase, and dormancy onset) were separately correlated to NASS corn growth stages (planting, emergence, silking, dough, dented, mature, and harvest) and soybean growth stages (planting, emergence, blooming, setting pods, dropping leaves, and harvest). Although their systematical bias varied with crop growth stages and states, the statistical regression correlations (R2) were larger than 0.9, indicating the robustness of HLS-VIIRS in crop phenology detections. We also collected PhenoCam network observations every 30 minutes to evaluate the 30m crop phenology derived from historical HLS-VIIRS time series as well as HLS-ABI time series. The result indicated that their mean absolute difference was less than 5.3 days in all six phenometrics and their mean systematic bias varied from 1.5 to 3.5 days.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Shen, Y., Zhang, X., Wang, W., Nemani, R., Ye, Y., and Wang, J. 2021. Fusing geostationary satellite observations with harmonized Landsat-8 and Sentinel-2 time series for monitoring field-scale land surface phenology. Remote Sensing. 13(21), 4465. https://doi.org/10.3390/rs13214465
- Type:
Journal Articles
Status:
Published
Year Published:
2022
Citation:
Shen, Y., Zhang, X., Yang, Z. 2022. Mapping corn and soybean phenometrics at field scales over the United States corn belt by fusing time series of Landsat 8 and Sentinel-2 data with VIIRS data. ISPRS Journal of Photogrammetry and Remote Sensing. 186, 55-69. https://doi.org/10.1016/j.isprsjprs.2022.01.023
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Zhang, X., Gao, F., Wang, J., and Ye, Y. 2021. Investigation of a spatiotemporal shape-matching model for generating synthetic high spatiotemporal resolution time series of satellite data. International Journal of Applied Earth Observation and Geoinformation. 104. https://doi.org/10.1016/j.jag.2021.102545
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2021
Citation:
Shen, Y., and Zhang, X. 2021. B15I-1548 fusing new generation geostationary satellite observations with Landsat-8 and Sentinel-2 time series for monitoring land surface phenology. AGU Fall Meeting. Dec 13-17. New Orleans, LA and Virtual.
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Gao, F., and Zhang, X. 2021. Mapping crop phenology in near real-time using satellite remote sensing: Challenges and opportunities. Journal of Remote Sensing. Volume 2021, Article ID 8379391. https://doi.org/10.34133/2021/8379391
|
Progress 06/01/20 to 05/31/21
Outputs
Target Audience:Our research in the second year compared the satellite observations of field scale (30m) crop progress data with NASS reports based on field observations for the 2018 growing season across North Dakota, South Dakota, Nebraska, Minnesota, and Iowa. The comparison demonstrates the capability of satellite monitoring of crop progress at the field scale. The results were presented to audiences at the American Geophysical Union (AGU) Fall Meeting in December 2020, and were used to prepare a manuscript for a peer-reviewed journal. Moreover, we held project team meeting twice with scientists from USDA, which provided an opportunity for us to understand better the needs of USDA. Changes/Problems:We actually started this project in August 2019 after a PhD student joined our team, although the project was funded to start on 06/01/2019. Because of the impact of COVID-19, the potential post-doctoral student was not able to travel to the US. As a result, the position of part time post-doctoral student has not been filled, which impacted the progress of this project to some degree. What opportunities for training and professional development has the project provided?This project provided an opportunity to train one graduate student. Specifically, a PhD student (Yu Shen) joined this research project in late August 2019. The PI held regular meeting every two weeks to discuss Mr. Shen's course work and research progress, and to provide him guidance on solving related problems during his research activities. Moreover, the PI frequently met with Mr. Shen to explain the theories and algorithms in processing satellite data for crop monitoring. As a result, Mr. Shen has greatly improved his knowledge and skills in analyzing seasonal time series of satellite data for crop monitoring. How have the results been disseminated to communities of interest?Results were presented at a scientific conference. One journal article was published and a second is under review. What do you plan to do during the next reporting period to accomplish the goals?Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing Landsat-8 OLI and Sentinel-2 MSI with GOES-R ABI data We will continue to investigate the daily angularly-corrected EVI2 from GOES-R ABI TOA reflectance and surface reflectance. Once the ABI surface reflectance product is improved in NASA AMS, we will use it to enhance the EVI2 time series of crop growth. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic OLI-MSI-ABI EVI2 and climatological crop phenometrics every week We will enhance algorithms and computer codes for simulating temporal trajectory of crop greenness development, which will reduce the uncertainty in detecting crop phenometrics. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time We will complete a computer code to automatically detect weekly crop progress using HLS and ABI time series in South Dakota and evaluate the accuracy by comparing with field observations. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. We will conduct field experimental observations and continue to investigate the crop progress using PhenoCam observations.
Impacts
What was accomplished under these goals?
Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing Landsat-8 OLI and Sentinel-2 MSI with GOES-R ABI data. (85% Accomplished) We performed the following tasks to accomplish this objective. (1) We developed a new algorithm, Spatiotemporal Shape-Matching Model (SSMM), to fuse high spatial resolution data [Landsat-8 OLI and Sentinel-2 MSI at 30m pixels, called harmonized Landsat and Sentinel-2 (HLS) data] with high temporal resolution data [daily Moderate Resolution Imaging Spectroradiometer (MODIS) and the Visible Infrared Imaging Radiometer Suite (VIIRS) observations], as well as GEOS-R ABI data to generate synthetic time series of daily 30m crop greenness. This algorithm hypothesizes that the temporal shape of EVI2 at a finer resolution pixel is able to match to a coarser resolution pixel within a local area, although their magnitude and phase could be mismatched. Thus, the shape of the selected VIIRS (or MODIS) or ABI EVI2 time series can be used to dynamically match to the given HLS EVI2 time series to generate a high spatiotemporal resolution time series. We developed computer code (C programming and Perl scripts) and started to process 3-day HLS-VIIRS time series for each 30m pixel. This algorithm was used to establish climatological phenometrics of corn and soybean in North Dakota, South Dakota, Nebraska, Minnesota, and Iowa. (2) We further investigated the impact of bidirectional reflectance distribution function (BRDF) on diurnal ABI surface reflectance and top of atmosphere (TOA) reflectance (15 minute). The investigation was conducted based on a set of models including the semi-physical Ross-thick Li-sparse BRDF kernel-driven model and the simplified BRDF model to adjust diurnal ABI reflectance. Our results indicate that currently available ABI surface reflectance products contain large uncertainties, including cloud mask, and that BRDF models are ineffective in adjusting BRDF effects. Thus, we generated time series ABI EVI2 from TOA reflectance using the following method: (a) selecting EVI2 in a 15-minute interval if the corresponding normalized difference vegetation index (NDVI) is larger than normalized difference water index (NDWI) (to reduce impacts from snow, cloud, and moisture) and EVI2<0.9NDVI (to reduce abnormal value); and (b) selecting 90th percentile EVI2 during a 3-day period. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic OLI-MSI-ABI EVI2 and climatological crop phenometrics every week. (40% Accomplished) We continued to develop the algorithm and compute code to simulate and project potential EVI2 temporal trajectories using available EVI2 observations from VIIRS data. The algorithm was able to project the potential EVI2 for the detection of crop phenometics. Because the uncertainty of current algorithms was relatively large in crop phenometrics, we started to further enhance the algorithm. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time. (20% Accomplished) We developed and tested an algorithm and computer code to produce weekly crop progress in near real time using VIIRS (500m) time series. The majority of the computer code was completed and tested. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. (15% Accomplished) We started to collect USDA NASS crop progress data and PhenoCam network observations to evaluate the 30m crop phenology derived from historical HLS-VIIRS time series. Further, we conducted field observations of growth conditions of corn and soybean around Brookings, SD for year 2020.
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2020
Citation:
Shen, Y., Zhang, X., Yang, Z. 2020. (GC023-0008) Mapping Crop Phenological Metrics at Field Scales by Fusing Time Series of VIIRS and HLS over the United States Corn Belt, AGU Fall Meeting. Dec 1-17. (virtual)
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Gao, F., Zhang, X. 2021. Mapping Crop Phenology in Near Real-Time Using Satellite Remote Sensing: Challenges and Opportunities. Journal of Remote Sensing. Volume 2021, Article ID 8379391, https://doi.org/10.34133/2021/8379391
- Type:
Journal Articles
Status:
Submitted
Year Published:
2021
Citation:
Zhang, X., Gao, F., Wang, J., Ye, Y. 2021. Investigation of a Spatiotemporal Shape-Matching Model for Generating Synthetic High Spatiotemporal Resolution Time Series of Satellite Data. International Journal of Applied Earth Observation and Geoinformation. (under review)
|
Progress 06/01/19 to 05/31/20
Outputs
Target Audience:Our research in the first year generated field scale (30m) crop progress data for the 2018 growing season across North Dakota, South Dakota, Nebraska, Minnesota, and Iowa. This research output will provide improved information for crop monitoring communities of NASS regarding satellite capability to monitor crop progress at the field scale. The results will be presented in a peer-reviewed journal article to deliver the science-based knowledge to broad audiences. Changes/Problems:We actually started this project in August 2019 after a PhD student joined our team, although the project was funded to start on 06/01/2019. A post-doctorate student will not be available until summer 2020. Moreover, the COVID-19 is likely to impact our planed field observations however growth on a set of selected. What opportunities for training and professional development has the project provided?This project provided an opportunity to train one PhD student. Specifically, a PhD student (Yu Shen) joined this research project in late August 2019. The PI held regular meeting every two weeks to discuss Mr. Shen's course work and research progress, and to provide him guidance on solving related problems during his research activities. Moreover, the PI frequently met with Mr. Shen to explain the theory and algorithm in processing satellite data for crop monitoring. As a result, Mr. Shen has improved greatly his knowledge and skills in analyzing seasonal time series of satellite data for crop monitoring. How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing Landsat-8 OLI and Sentinel-2 MSI with GOES-R ABI data We will continue to work on the establishment of climatological phenometrics and the development of algorithms and computer codes for calculating daily angularly-corrected EVI2 from GOES-R ABI. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic OLI-MSI-ABI EVI2 and climatological crop phenometrics every week We will develop algorithms and computer codes for simulating temporal trajectory of crop greenness development. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time We have no tasks planned for this objective. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. We will conduct field experimental observations and continue to investigate the crop progress using PhenoCam observations.
Impacts
What was accomplished under these goals?
Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing Landsat-8 OLI and Sentinel-2 MSI with GOES-R ABI data. (60% Accomplished) We performed the following tasks to reach the goal of this objective. (1) We developed a new algorithm to fuse high spatial resolution data [Landsat-8 OLI and Sentinel-2 MSI at 30m pixels, called harmonized Landsat and Sentinel-2 (HLS) data] with high temporal resolution data [daily Moderate Resolution Imaging Spectroradiometer (MODIS) and the Visible Infrared Imaging Radiometer Suite (VIIRS) observations] to generate synthetic time series of daily 30m crop greenness. In particular, this algorithm hypothesizes that the temporal shape of EVI2 at a finer resolution pixel is able to match to that at a coarser resolution pixel within a local area, although their magnitude and phase could be mismatched. Thus, the shape of the selected VIIRS (or MODIS) EVI2 time series can be used to dynamically match to the given HLS EVI2 time series to generate a HLS-VIIRS time series. We developed computer code (C programming and Perl scripts) and started to process 30m 3-day HLS-VIIRS time series. This algorithm was used to establish climatological phenometrics of corn and soybean in North Dakota, South Dakota, Nebraska, Minnesota, and Iowa. (2) We obtained a large set of GEOS-R ABI data across the US and investigated the diurnal (every 15 minutes) variations in ABI observations. Diurnal ABI EVI2 time series (15 minute) were analyzed to derive the kernel weights in a model of bidirectional reflectance distribution function (BRDF). Our results indicate that the diurnal EVI2 calculated from land surface reflectance varies within a small range and the kernel-driven BRDF model could be used to adjust diurnal ABI EVI2 slightly. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic OLI-MSI-ABI EVI2 and climatological crop phenometrics every week. (10% Accomplished) We investigated the algorithm to simulate a set of potential EVI2 temporal trajectories using a few selected samples of VIIRS data (500m). The algorithm worked well for the selected samples. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time. (0% Accomplished) No work was conducted for this objective during the reporting period. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. (5% Accomplished) We started to collect USDA NASS crop progress data and PhenoCam network observations to evaluate the 30m crop phenology derived from historical HLS-VIIRS time series.
Publications
|
Progress 04/01/19 to 03/31/20
Outputs
Target Audience:Our research in the first year generated field scale (30m) crop progress data for the 2018 growing season across North Dakota, South Dakota, Nebraska, Minnesota, and Iowa. This research output will provide improved information for crop monitoring communities of NASS regarding satellite capability to monitor crop progress at the field scale. The results will be presented in a peer-reviewed journal article to deliver the science-based knowledge to broad audiences. Changes/Problems:We actually started this project in August 2019 after a PhD student joined our team, although the project was funded to start on 06/01/2019. A post-doctorate student will not be available until summer 2020. Moreover, the COVID-19 is likely to impact our planed field observations however growth on a set of selected. What opportunities for training and professional development has the project provided?This project provided an opportunity to train one PhD student. Specifically, a PhD student (Yu Shen) joined this research project in late August 2019. The PI held regular meeting every two weeks to discuss Mr. Shen's course work and research progress, and to provide him guidance on solving related problems during his research activities. Moreover, the PI frequently met with Mr. Shen to explain the theory and algorithm in processing satellite data for crop monitoring. As a result, Mr. Shen has improved greatly his knowledge and skills in analyzing seasonal time series of satellite data for crop monitoring. How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing Landsat-8 OLI and Sentinel-2 MSI with GOES-R ABI data We will continue to work on the establishment of climatological phenometrics and the development of algorithms and computer codes for calculating daily angularly-corrected EVI2 from GOES-R ABI. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic OLI-MSI-ABI EVI2 and climatological crop phenometrics every week We will develop algorithms and computer codes for simulating temporal trajectory of crop greenness development. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time We have no tasks planned for this objective. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. We will conduct field experimental observations and continue to investigate the crop progress using PhenoCam observations.
Impacts
What was accomplished under these goals?
Objective 1: Generate synthetic time series of daily crop greenness, quantified using a two band enhance vegetation index (EVI2), by fusing Landsat-8 OLI and Sentinel-2 MSI with GOES-R ABI data. (60% Accomplished) We performed the following tasks to reach the goal of this objective. (1) We developed a new algorithm to fuse high spatial resolution data [Landsat-8 OLI and Sentinel-2 MSI at 30m pixels, called harmonized Landsat and Sentinel-2 (HLS) data] with high temporal resolution data [daily Moderate Resolution Imaging Spectroradiometer (MODIS) and the Visible Infrared Imaging Radiometer Suite (VIIRS) observations] to generate synthetic time series of daily 30m crop greenness. In particular, this algorithm hypothesizes that the temporal shape of EVI2 at a finer resolution pixel is able to match to that at a coarser resolution pixel within a local area, although their magnitude and phase could be mismatched. Thus, the shape of the selected VIIRS (or MODIS) EVI2 time series can be used to dynamically match to the given HLS EVI2 time series to generate a HLS-VIIRS time series. We developed computer code (C programming and Perl scripts) and started to process 30m 3-day HLS-VIIRS time series. This algorithm was used to establish climatological phenometrics of corn and soybean in North Dakota, South Dakota, Nebraska, Minnesota, and Iowa. (2) We obtained a large set of GEOS-R ABI data across the US and investigated the diurnal (every 15 minutes) variations in ABI observations. Diurnal ABI EVI2 time series (15 minute) were analyzed to derive the kernel weights in a model of bidirectional reflectance distribution function (BRDF). Our results indicate that the diurnal EVI2 calculated from land surface reflectance varies within a small range and the kernel-driven BRDF model could be used to adjust diurnal ABI EVI2 slightly. Objective 2: Simulate a set of potential EVI2 temporal trajectories from the timely available synthetic OLI-MSI-ABI EVI2 and climatological crop phenometrics every week. (10% Accomplished) We investigated the algorithm to simulate a set of potential EVI2 temporal trajectories using a few selected samples of VIIRS data (500m). The algorithm worked well for the selected samples. Objective 3: Produce weekly crop progress and condition at a 30m field scale throughout the growing season in near real time. (0% Accomplished) No work was conducted for this objective during the reporting period. Objective 4: Validate and calibrate crop progress and condition products using in-situ observations and develop a geospatial tool for USDA NASS to integrate into existing operational system of VegScape. (5% Accomplished) We started to collect USDA NASS crop progress data and PhenoCam network observations to evaluate the 30m crop phenology derived from historical HLS-VIIRS time series.
Publications
|
|