Monitoring how farmers till their fields can provide important information in estimating the contributions of the agriculture sector towards soil carbon sequestration and reductions in greenhouse gas emissions. Agriculture and Agri-Food Canada (AAFC) is testing the use of Coherent Change Detection, applied to Sentinel-1 Synthetic Aperture Radar (SAR) data to identify when fields are tilled. Data have been collected in sites in eastern Canada and results to date have been positive. By monitoring the temporal change in coherence, fields that were tilled were successfully flagged using this approach. A more comprehensive data set is currently being collected in order to extend validation of this method and to test if type of tillage can also be identified.

Agriculture and Agri-Food Canada (AAFC) delivers annual maps of crops grown across Canada, operationally, using Synthetic Aperture Radar (SAR) and optical satellite data. This study applies the AAFC methodology to sites in Latin America to test performance and adaptability to these cropping systems, using TerraSAR-X and RADARSAT SAR data. Overall classification results are promising (79.4% to 86.0%), but improvements will occur with better matching of SAR collection dates to local growing seasons, and by acquiring more robust field observations. These improvements will be the subject of additional research by AAFC and partner organizations, in this region.

The Water Cloud Model (WCM) has been widely used for estimation of Leaf Area Index (LAI) from Synthetic Aperture Radar (SAR). In different studies, it was demonstrated that this model performs well if it is calibrated well. However, calibration of this model requires access to both LAI and soil moisture for the calibration points. An alternative, if the soil moisture data are not available, is Machine Learning (ML) algorithms. However, ML methods are highly dependent on the number of calibration points. In this study, 6 different ML algorithms including Neural Network (NN), Support Vector Machine (SVM), Ensemble of Trees (ET), Regression Tree (RT), Radial Basis Function (RBF) and Gaussian Process Model (GPM) are used and compared with the WCM model for estimation of LAI over corn fields. This comparison was done using different numbers of calibration points. The results demonstrated that when a lower number of calibration points are used, WCM outperformed some of the ML algorithms including NN, SVM and ET algorithms. But with more calibration points, all machine learning algorithms outperformed the WCM. The highest accuracies were from the GPM model with a correlation coefficient (R) of 0.93, Root Mean Square (RMSE) of 0.56 m2m-2and Mean Absolute Error (MAE) of 0.38 m2m-2. Theses results were derived using the data collected during the SMAP Validation Experiment 2012 (SMAPVEX12) that was conducted in Manitoba, Canada. Further testing and comparison of the ML algorithms and WCM model using data from other Joint Experiment for Crop Assessment and Monitoring (JECAM) sites are ongoing.

In preparation for Canada's launch of the RADARSAT-Constellation, this study examines the use of VV-VH intensities to estimate the Leaf Area Index (LAI) of corn. LAI is indicative of crop productivity. Two implementations of the Water Cloud Model performed equally well in estimating corn LAI over sites in Poland and Canada with correlation coefficients over 0.8 and Root Mean Square Errors and Mean Average Errors of 0.72-0.73 m2m-2and 0.47-0.54 m2m-2, respectively. This research will continue to pull in data from other international sites. If results remain robust, a strong case can be made to use an integration of Sentinel-1 and RCM for operational crop condition monitoring.

Annual and within-season crop type monitoring and mapping is an important ongoing consideration for governments, global agricultural monitoring organizations and private interests worldwide. Successful country-wide operational remote sensing-based inventories are well-established utilizing optical-only and optical/single frequency Synthetic Aperture Radar (SAR) combinations of data. However, the drawbacks of these data combinations are the requirement of multiple sources of imagery throughout the entire growing season, which impedes within-season analysis, and cloud cover effects on the optical data. Currently, C-band SAR data are available with continuous global coverage from Sentinel-1A & B, RADARSAT-2 and from the expected launch of the RADARSAT Constellation Mission (RCM). With current and expected launches of several other frequency (L-, P-, etc.) SAR missions over the next few years (SAOCOM, NISAR, etc.) the opportunity for continuous, multi-frequency SAR coverage edges toward reality. The JECAM SAR Inter-Comparison Experiment is a multi-year, multi-partner project that aims to compare global methods for SAR-based crop monitoring and inventory. The third component of this experiment is the assessment of multi-frequency SAR data for crop classification and mapping. Earth observation (EO) data acquisitions of Sentinel-1A & B, RADARSAT-2/RCM, ALOS-2, SAOCOM1 and TerraSAR-X/TanDEM-X have been planned and requested for the 2019 growing season to complement within field surveys being conducted across the globe. In preparation for these data, this research analyzed ALOS-2, TerraSAR-X and RADARSAT-2 data for crop mapping at the JECAM Canada-Carman site using two dates of multi-frequency SAR data, in comparison to a traditional full season optical/SAR dataset. The multi-frequency data had similar overall accuracies as the optical/SAR dataset, and improved on several individual agricultural class accuracies.

Agriculture and Agri-Food Canada (AAFC) has combined RADARSAT-2 C-band dual polarization data with optical imagery to map crop types across the agricultural extent of Canada yearly since 2009. In preparation for the launch of the RADARSAT Constellation Mission (RCM) primary research has been focused on incorporating similar-mode dual polarization RCM data in the operational system. The availability of the compact polarimetry (CP) mode with continuous coverage has important implications for crop mapping and inventory. CP mode on RCM has a circular transmit and two orthogonal linear receive structure and maintains phase information. The addition of CP data to AAFC's operational crop type mapping will expand the information that the current dual polarization Synthetic Aperture Radar (SAR) component provides. This will increase the SAR contribution from the simple intensity of backscatter to capturing the scattering characteristics of the target. There are many parameters and decompositions that can be derived from CP data. Many of these parameters are highly correlated or may not provide information that are important for crop type identification. The goal of this research was to derive twenty-four CP parameters and decompositions for a growing season of SAR imagery and to assess the importance and contribution of these features to an overall classification of crops in southern Manitoba, Canada.

Cloudy conditions impede and reduce the utility of optical imagery. With the launch of Sentinel-1A and B, the ongoing availability of RADARSAT-2 imagery, and the expected launch of the RADARSAT Constellation Mission (RCM), dense time series of C-band Synthetic Aperture Radar (SAR) data will now be readily available. For crop classification and mapping, SAR imagery has yet to be used to its full potential and has generally been combined with optical imagery. The JECAM SAR Inter-Comparison Experiment is a multi-year, multi-partner project that aims to compare global methods for SAR-based crop monitoring and inventory. Sets of dense time-series SAR imagery which include RADARSAT-2 and Sentinel-1 data were prepared for this experiment. AAFC's operational Decision Tree (DT) and newly implemented Random Forest (RF) classification methodologies were applied to these SAR only data-stacks, and to optimized, traditional data-stacks of optical/SAR combinations. This paper outlines the results of these dense time-series classifications and how these results were affected by changing numbers of agriculture classes, numbers of available SAR imagery and numbers of training and validation data points for individual crop types. In general, for the dense time-series SAR stacks, overall accuracies of greater than 85%, a typical operational goal, were obtained for 6 of 12 sites. These results have important operational implications for particularly cloudy regions where the availability of optical imagery is limited.

In this study, a cross-calibration approach was applied to combine RADARSAT-2 and RapidEye sensors for biomass monitoring over corn fields. First, RapidEye and RADARSAT-2 sensors were compared in terms of biomass estimation. Then the estimated biomass from RADARSAT-2 was cross-calibrated with respect to the biomass estimated from RapidEye. Combination of the optical and cross-calibrated Synthetic Aperture Radar (SAR) derived biomass was proposed to have higher temporal resolution biomass maps. Vegetation indices including normalized difference vegetation index (NDVI), red-edge triangular vegetation index (RTVI), simple ratio (SR) and red-edge simple ratio (SRre) were used for modeling of biomass estimation from RapidEye. Water Cloud Model (WCM) was also used for biomass estimation from RADARSAT-2. Data collected during SMAP Validation Experiment 2012 (SMAPVEX12) field campaign was used for validation. The results demonstrate that the accuracies of biomass estimations from RapidEye and RADARSAT-2 are close. For RapidEye, the highest accuracies derived from RTVI index with correlation coefficient (R) of 0.92 and Root Mean Square of (RMSE) of 118.18 gr/m2. The R values derived from RADARSAT-2 is 0.83 and its RMSE is 171.93 gr/m2. After cross-calibration of the biomass derived from RADARSAT-2 versus those derived from RapidEye, the RMSE of estimates dropped by 18.86 gr/m2.

Utilizing Synthetic Aperture Radar (SAR) sensors for crop inventory and condition monitoring offers many advantages, particularly the ability to collect data under cloudy conditions. The JECAM SAR Inter-Comparison Experiment is a multi-year, multi-partner project that aims to compare global methods for SAR crop monitoring and inventory, and Leaf Area Index (LAI) and biomass retrieval. Data sets of SAR imagery including RADARSAT-2 and Sentinel-1 are being prepared for this experiment and it is important to develop best practices to ensure consistency across the data sets. This paper outlines the speckle filter testing results based upon changing filter types and window sizes in comparison with changing field size. In general, it was found that the adaptive Touzi filter resulted in the highest overall classification accuracies for all field sizes. It was also found that there may be importance to speckle filter window size in relation to agriculture field size for other filter types.

Multisplit is a broadly useful parallel primitive that permutes its input data into contiguous buckets or bins, where the function that categorizes an element into a bucket is provided by the programmer. Due to the lack of an efficient multisplit on GPUs, programmers often choose to implement multisplit with a sort. However, sort does more work than necessary to implement multisplit, and is thus inefficient. In this work, we provide a parallel model and multiple implementations for the multisplit problem. Our principal focus is multisplit for a small number of buckets. In our implementations, we exploit the computational hierarchy of the GPU to perform most of the work locally, with minimal usage of global operations. We also use warp-synchronous programming models to avoid branch divergence and reduce memory usage, as well as hierarchical reordering of input elements to achieve better coalescing of global memory accesses. On an NVIDIA K40c GPU, for key-only (key-value) multisplit, we demonstrate a 3.0-6.7x (4.4-8.0x) speedup over radix sort, and achieve a peak throughput of 10.0 G keys/s.

For large-scale graph analytics on the GPU, the irregularity of data access/control flow and the complexity of programming GPUs have been two significant challenges for developing a programmable high-performance graph library. "Gunrock," our high-level bulk-synchronous graph-processing system targeting the GPU, takes a new approach to abstracting GPU graph analytics: rather than designing an abstraction around computation, Gunrock instead implements a novel data-centric abstraction centered on operations on a vertex or edge frontier. Gunrock achieves a balance between performance and expressiveness by coupling high-performance GPU computing primitives and optimization strategies with a high-level programming model that allows programmers to quickly develop new graph primitives with small code size and minimal GPU programming knowledge. We evaluate Gunrock on five graph primitives (BFS, BC, SSSP, CC, and PageRank) and show that Gunrock has on average at least an order of magnitude speedup over Boost and PowerGraph, comparable performance to the fastest GPU hardwired primitives, and better performance than any other GPU high-level graph library.

For large-scale graph analytics on the GPU, the irregularity of data access and control flow and the complexity of programming GPUs have been two significant challenges for developing a programmable high-performance graph library. "Gunrock", our graph-processing system, uses a high-level bulk-synchronous abstraction with traversal and computation steps, designed specifically for the GPU. Gunrock couples high performance with a high-level programming model that allows programmers to quickly develop new graph primitives with less than 300 lines of code. We evaluate Gunrock on five graph primitives and show that Gunrock has at least an order of magnitude speedup over Boost and PowerGraph, comparable performance to the fastest GPU hardwired primitives, and better performance than any other GPU high-level graph library.

Finding the shortest paths from a single source to all other vertices is a fundamental method used in a variety of higher-level graph algorithms. We present three parallel friendly and work-efficient methods to solve this Single-Source Shortest Paths (SSSP) problem: Work front Sweep, Near-Far and Bucketing. These methods choose different approaches to balance the trade off between saving work and organizational overhead. In practice, all of these methods do much less work than traditional Bellman-Ford methods, while adding only a modest amount of extra work over serial methods. These methods are designed to have a sufficient parallel workload to fill modern massively-parallel machines, and select reorganizational schemes that map well to these architectures. We show that in general our Near-Far method has the highest performance on modern GPUs, outperforming other parallel methods. We also explore a variety of parallel load-balanced graph traversal strategies and apply them towards our SSSP solver. Our work-saving methods always outperform a traditional GPU Bellman-Ford implementation, achieving rates up to 14x higher on low-degree graphs and 340x higher on scale free graphs. We also see significant speedups (20-60x) when compared against a serial implementation on graphs with adequately high degree.

We formalize sampling a function using k -d darts. A k -d Dart is a set of independent, mutually orthogonal, k -dimensional hyperplanes called k -d flats. A dart has d choose k flats, aligned with the coordinate axes for efficiency. We show k -d darts are useful for exploring a function's properties, such as estimating its integral, or finding an exemplar above a threshold. We describe a recipe for converting some algorithms from point sampling to k -d dart sampling, if the function can be evaluated along a k -d flat. We demonstrate that k -d darts are more efficient than point-wise samples in high dimensions, depending on the characteristics of the domain: for example, the subregion of interest has small volume and evaluating the function along a flat is not too expensive. We present three concrete applications using line darts (1-d darts): relaxed maximal Poisson-disk sampling, high-quality rasterization of depth-of-field blur, and estimation of the probability of failure from a response surface for uncertainty quantification. Line darts achieve the same output fidelity as point sampling in less time. For Poisson-disk sampling, we use less memory, enabling the generation of larger point distributions in higher dimensions. Higher-dimensional darts provide greater accuracy for a particular volume estimation problem.

Abstract We provide a simple algorithm and data structures for d‐dimensional unbiased maximal Poisson‐disk sampling. We use an order of magnitude less memory and time than the alternatives. Our results become more favorable as the dimension increases. This allows us to produce bigger samplings. Domains may be non‐convex with holes. The generated point cloud is maximal up to round‐off error. The serial algorithm is provably bias‐free. For an output sampling of size n in fixed dimension d, we use a linear memory budget and empirical θ(n) runtime. No known methods scale well with dimension, due to the “curse of dimensionality.” The serial algorithm is practical in dimensions up to 5, and has been demonstrated in 6d. We have efficient GPU implementations in 2d and 3d. The algorithm proceeds through a finite sequence of uniform grids. The grids guide the dart throwing and track the remaining disk‐free area. The top‐level grid provides an efficient way to test if a candidate dart is disk‐free. Our uniform grids are like quadtrees, except we delay splits and refine all leaves at once. Since the quadtree is flat it can be represented using very little memory: we just need the indices of the active leaves and a global level. Also it is very simple to sample from leaves with uniform probability.

We present a multi-stage method for solving large tridiagonal systems on the GPU. Previously large tridiagonal systems cannot be efficiently solved due to the limitation of on-chip shared memory size. We tackle this problem by splitting the systems into smaller ones and then solving them on-chip. The multi-stage characteristic of our method, together with various workloads and GPUs of different capabilities, obligates an auto-tuning strategy to carefully select the switch points between computation stages. In particular, we show two ways to effectively prune the tuning space and thus avoid an impractical exhaustive search: (1) apply algorithmic knowledge to decouple tuning parameters, and (2) estimate search starting points based on GPU architecture parameters. We demonstrate that auto-tuning is a powerful tool that improves the performance by up to 5x, saves 17% and 32% of execution time on average respectively over static and dynamic tuning, and enables our multi-stage solver to outperform the Intel MKL tridiagonal solver on many parallel tridiagonal systems by 6-11x.

We solve the problem of generating a uniform Poisson-disk sampling that is both maximal and unbiased over bounded non-convex domains. To our knowledge this is the first provably correct algorithm with time and space dependent only on the number of points produced. Our method has two phases, both based on classical dart-throwing. The first phase uses a background grid of square cells to rapidly create an unbiased, near-maximal covering of the domain. The second phase completes the maximal covering by calculating the connected components of the remaining uncovered voids, and by using their geometry to efficiently place unbiased samples that cover them. The second phase converges quickly, overcoming a common difficulty in dart-throwing methods. The deterministic memory is O ( n ) and the expected running time is O ( n log n ), where n is the output size, the number of points in the final sample. Our serial implementation verifies that the log n dependence is minor, and nearly O ( n ) performance for both time and memory is achieved in practice. We also present a parallel implementation on GPUs to demonstrate the parallel-friendly nature of our method, which achieves 2.4x the performance of our serial version.