NASA-ISRO Synthetic Aperture Radar (NISAR) mission with a near-global coverage of land and cryosphere regions with 12 days repeat will operationally produce level-1 and level-2 science products from L-band radar by the NASA data system at Jet Propulsion Laboratory. The products will be available to the users through the NASA’s Distributed Active Archive Center (DAAC) at Alaska Satellite Facility (ASF). In this paper we present the algorithms to process, calibrate and assess the quality of the NISAR products after launch. We simulate NISAR raw data from different science modes and with NISAR L-band configurations (such as left looking, squinted beam and dithered data and the same transmit chirp and receive configuration as the L-band instrument) and process them through the NISAR standard processor to from level-1 and level-2 products. We quantify the quality of formed images and their derived level-2 products such as geocoded SLC, covariance and interferometry products over point targets. We further evaluate the performance of the NISAR processor and algorithms using real L-band data acquired by ALOS-1 and ALOS-2 and reformatted to NISAR format.

For NASA-ISRO Synthetic Aperture Radar (NISAR) mission processing, Quality Assurance (QA) plots, metrics, and summary statistics will be produced by the NISAR Science Data System (SDS) in human- and machine-readable formats for each level-1 (L1) and level-2 (L2) science data product and made freely and openly available for users. It is expected that NISAR will generate millions of individual science data products, with each product ranging in size from a few hundred MB to over 40 GB, for approximately 100 PB added to the NASA data archives over the nominal 3 year mission. This volume will be the largest of any NASA mission to date; it creates a need for automated checks and for each product to be digested into smaller, summary statistics and plots for quick review. This need will be addressed by the QA software, which will be run at scale as part of nominal NISAR mission processing on each L1/L2 data product. QA will generate small, standalone, human- and/or machine-parsable files that will be freely available from the Alaska Satellite Facility Distributed Active Archive Center (ASF DAAC) alongside the primary L1/L2 NISAR products. This paper introduces the QA output files, their content and formats, and the usage of QA in the context of NISAR mission processing. In the era of big data, these smaller files could be a significant firststep towards analyzing broad trends across the large, primary L1/L2 NISAR datasets.

This paper presents the Radar Echo Emulator (REE), a SAR simulation tool with high fidelity and flexibility that is being used to assess and quantify the radar instrument functionality, performance, and overall impulse response for various modes of the NASA-ISRO synthetic aperture radar (NISAR) mission. NISAR is a complex multi-channel multi-polarization wide-band wide-swath high-resolution SAR instrument based on SweepSAR architecture [1]. Its radar instruments support several configurable radar modes to fulfill variety of science applications [2]. The paper provides a brief overview of the REE tool as a general SAR simulator while primarily focusing on its SweepSAR and digital beamforming (DBF) application via an example of repeat-pass four-channel L-band split-spectrum NISAR simulation over an extended Amazon rainforest-like scene [3], [4]. An example of the ionosphere effect with split-spectrum analysis and an example with radio frequency interference (RFI) contamination are also presented.

Now, more than ever, there is a need for higher-quality products for remote-sensing end users. Following recent advancements in synthetic aperture radar (SAR) processing algorithms, we provide a full assessment of the radiometric dependence of geocoded C-band SAR backscatter processed with radiometric terrain correction (RTC) over differing target types. In particular, we compare the flatness of RTC-normalized backscatter coefficient gamma-naught with respect to local incidence angle over 46 Sentinel-1 (S1) datasets representing about 20 land classes in the Copernicus Global Land Service (CGLS) Land Cover 100m classification using the Observational Products for End-Users from Remote Sensing (OPERA) RTC-S1 product workflow and the ISCE3 framework. We also calculate the mean and median radar backscatter over areas of foreslope and backslope. Our results suggest that the dependence of gamma-naught on the local topography depends strongly on the land type and only exhibits near-constant behavior in certain forest land types. Evidence also suggests that as we move further away from densely tree-covered areas, the dependence of gamma-naught on the local incidence angle becomes stronger.

The Observational Products for End-Users from Remote Sensing Analysis (OPERA) project at the Jet Propulsion Laboratory (JPL) will provide a near-global Radiometric Terrain Corrected synthetic aperture radar (SAR) backscatter from Sentinel-1 (RTC-S1) product. The OPERA RTC-S1 product will deliver map-projected burst-based radar images with a geographic scope that includes all land masses excluding Antarctica, and with temporal sampling coincident with the availability of Sentinel-1 interferometric wide (IW) single-look complex (SLC) data. This paper presents the OPERA RTC-S1 product, providing details about its layers, static layers, and metadata; describing the product’s processing workflow, based on the ISCE3 framework and using the same algorithms developed for the NASA-ISRO Synthetic Aperture Radar (NISAR) mission; and outlining the algorithm verification and the product validation plan. We also present a global mosaic of preliminary OPERA RTC-S1 products generated from a global end-to-end test run from a Sentinel-1A orbit cycle. The OPERA RTC-S1 product will be publicly distributed through the Alaska Satellite Facility (ASF) Distributed Active Archive Center (DAAC) free of charge, with a release date scheduled for September 2023 with forward stream production.

This article describes a projection algorithm between radar and map coordinates based on the representation of radar samples as area elements (AEs) rather than point elements. Each AE on the map grid (geographic grid) is associated with a number of radar grid samples that intersect completely or partially the AE. The association enables the geocoding (i.e., the map projection of radar imagery) with adaptive multilooking, accurately accounting for all radar samples contributing to the geocoded elements according to topography and radar geometry. By using averaging rather than interpolation, the proposed projection does not suffer from interpolation overfitting. The area-based geocoding also enables the generation of the geocoded polarimetric covariance matrix (GCOV) and geocoded synthetic aperture radar (SAR) interferograms with adaptive multilooking. Analogously, the slant-range projection of geocoded data is improved by projecting geographic grid pixels onto the radar grid according to their corresponding location based on the radar geometry without leaving gaps. This approach is used to reduce the computation time of previously published radiometric terrain correction (RTC) algorithms, performing 3.6–6.5 times faster over multilooked data and up to 26.3 times faster over single-look data. We demonstrate the strengths of the proposed area projection (AP) algorithm for RTC and geocoding using Uninhabited Aerial Vehicle SAR (UAVSAR), Sentinel-1B, and ALOS-2/PALSAR-2 data, and evaluate the results in the context of the upcoming NASA-ISRO SAR (NISAR) mission.

The forthcoming launch of the NASA-ISRO Synthetic Aperture Radar (NISAR) mission will open the path to a new type of L-band measurements constituted by dense time-series of polarimetric backscatter and interferometric coherence with unprecedented spatial and temporal sampling. Here, we start the development of a theoretical framework that links L-band backscatter time-series with interferometric coherence measurements. The water-cloud-model (WCM) and an extended version of the random-motion-over-ground (RMoG) model are adopted to express radar measurements in terms of forest above-ground biomass and tree height. Time-series data collected during the 2019 UAVSAR AM-PM campaign in Southeastern United States are used to evaluate the correlation of various PolSAR- and InSAR-derived parameters with field-measured above-ground biomass.

This article presents a projection algorithm based on the representation of radar samples as area elements, rather than point elements as traditionally done in previous works. Each area element in the geographic grid (geogrid) is associated with a set of samples in the radar grid that intersect completely or partially the area element according to the topography and the radar geometry. Accurate geocoding with adaptive multi-looking is achieved by successively assigning the weighted average of the radar samples to the corresponding geogrid elements. Analogously, the slant-range projection of geocoded data is improved by projecting the geogrid pixels onto the radar grid according to their projected area. When our slant-range projection approach is used within previously-published radiometric terrain correction (RTC) algorithms, the processing time is significantly reduced, performing 4.2 to 6.5 times faster over multi-looked data and up to 16.7 over single-look data. We demonstrate the strength of the area projection algorithm for RTC and geocoding using UAVSAR and Sentinel-1 data, and evaluate the results in the context of the upcoming NISAR mission.

This article demonstrates how 3-D vegetation structure can be approximated by interferometric synthetic aperture radar (InSAR) backscatter-height histograms. Single-look backscatter measurements are plotted against the InSAR phase height and are aggregated spatially over a forest patch to form a 3-D histogram, referred to as InSAR backscatter-height histogram or simply InSAR histogram. InSAR histograms resemble LiDAR waveforms, suggesting that existing algorithms used to retrieve canopy height and ground topography from radar tomograms or LiDAR waveforms can be applied to InSAR histograms. Three algorithms are evaluated to generate maps of digital terrain, surface, and canopy height models: Gaussian decomposition, quantile, and backscatter threshold. Full-polarimetric L-band uninhabited aerial vehicle synthetic aperture radar (UAVSAR) data collected over the Gabonese Lopé National Park during the 2016 AfriSAR campaign are used to illustrate and compare the performance of the algorithms for the HH, HV, VV, HH+VV, and HH-VV polarimetric channels. Results show that radar-derived maps using the InSAR histograms differ by 4 m (top-canopy), 5 m (terrain), and 6 m (forest height) in terms of average root-mean-square errors (RMSEs) from standard maps derived from full-waveform laser, vegetation, and ice sensor (LVIS) LiDAR measurements.

We report observations of the three-dimensional polarimetric covariance matrix derived from interferometric synthetic aperture radar (InSAR) histograms. Single-look covariance samples plotted versus InSAR phase are aggregated spatially over a forest patch to form a three-dimensional histogram. To first order, InSAR backscatter histograms resemble lidar waveforms, suggesting that this type of measurement can be used as a proxy for horizontal and vertical tree structure. In this paper, we extend the formulation of the InSAR backscatter histograms to the covariance matrix and discuss possible new applications. Full-polarimetric L-band and P-band UAVSAR (Uninhabited Aerial Vehicle Synthetic Aperture Radar) data collected over Delta Junction in Alaska during the Arctic-Boreal Vulnerability Experiment (ABoVE) are used to illustrate the results. Validation is conducted with the full-waveform LVIS (Laser, Vegetation and Ice Sensor) lidar instrument. Important parameters such as the alpha and the entropy will be extracted along the tree vertical direction and illustrated in the final paper.

We employ dual-polarimetric P-band interferometric synthetic aperture radar (InSAR) backscatter-phase histograms of the Amazon rainforest of Urucu to estimate the elevation of the forest terrain. High-resolution single-look phase and phase-height histograms weighted by backscatter intensity are projected in the elevation direction and are combined in the range and azimuth directions to form a three-dimensional backscatter-phase histogram. Profiles of the resulting histograms resemble forest reflectivity. Higher ground backscatter is observed in the P-HH compared to the P-HV profile, in agreement with scattering models. Results suggest that the peak of the difference between the normalized P-HH and P-HV profiles can be used as an estimator for the terrain elevation from a model-free approach. We qualitatively compare the results to traditional P-band interferometry and polarimet-ric InSAR (PolInSAR) random-volume over ground (RVoG) three-stage approach over five different baselines 20m, 65m, 80m, 100m, and 145m.

We study the temporal variability of soil and vegetation backscatter at L-band using a dense airborne time-series. Backscatter is assumed to change over time due to diurnal variations in soil and canopy water content as well as precipitations. A two-layer SAR backscattere model traditionally used for above-ground biomass retrieval is augmented here with the time dimension in order to guide the data analysis. The model is informed by examining a 5-year time-series of 32 L-band polarimetric UAVSAR images acquired over a vegetated area near the Sacramento Delta in California. Our initial results reported in this paper show that the temporal variability of soil and vegetation backscatter in absence of precipitation events fits well a lognormal probability distribution with mean and standard deviation related to each other. Characterizing the diurnal, seasonal and interannual variability of L-band backscatter may be critical for the successful estimation of ecosystem variables from dense time-series to be acquired globally by the NISAR mission.

PLANT (Polarimetric-interferometric Lab and Analysis Tools) is a new collection of software tools developed at the Jet Propulsion Laboratory to support processing and analysis of Synthetic Aperture Radar (SAR) data for ecosystem and land-cover/land-use change science and applications. PLANT inherits code components from the Interferometric Scientific Computing Environment (ISCE) to generate high-resolution, coregistered polarimetric-interferometric SLC stacks from Level-0/1 data for a variety of airborne and spaceborne sensors. The goal is to provide the ecosystem and land-cover/land-use change communities with rigorous and efficient tools to perform multi-temporal, polarimetric and tomographic analyses in order to generate calibrated, geocoded and mosaicked Level-2 and Level-3 products (e.g., maps of above-ground biomass and forest disturbance). In this paper we introduce the capabilities of PLANT and report first results obtained with the tools developed up to date.

The random volume over ground model has been successfully used to estimate forest parameters with the interferometric synthetic aperture radar. Besides its typical employment with polarimetric data at lower frequency bands, it has also been used with single-polarimetric X-band data to estimate forest height. Based on this approach, we propose estimating the forest interferometric height and subtract it from the X-band surface model, in order to generate a preliminary model of the terrain using X-band only. Since P-band interferometric phase center is closer to the ground, the use of the preliminary terrain model as reference reduces the interferometric phase modulation, facilitating the phase unwrapping. This is particularly important for PolInSAR terrain height estimation or long baseline interferometry, which is used in dual-band In-SAR to improve the terrain height estimation. Two datasets with three-baselines X-HH and dual-polarimetric P-HH and P-HV data, provided by Bradar, are used to demonstrate the method.

There are basically two methods for retrieving the ground and the canopy height from the interferometric synthetic aperture radar (InSAR) in forested areas. The first method is based on the different dual-band (DB) InSAR height estimation, typically done at X- and P-HH-bands. The second method is based on the modeling of the Polarimetric (Pol) InSAR response of the forest volumetric backscatter, typically at L- or P-band. This paper investigates the possibility of combining both methods, in the sense that the P-band derived ground height is enhanced by the use of the polarimetric data and the RVoG model. In this proposed method, only the ground phase is retrieved from the PolInSAR technique, reducing the numbers of unknowns to be inverted, while the X-band InSAR is used to estimate the canopy. Therefore, we avoid the more complex (and more failure susceptible) inversion of the PolInSAR approach. The data collected in the Amazon region with the OrbiSAR-1 sensor from Bradar (former Orbisat) is used to demonstrate the method. Better ground estimation over range is obtained with the proposed method in comparison with the single-polarization approach for different baselines.