Engineering topics
Michaelis, Andrew
Publications and source records attributed to Michaelis, Andrew.
GEONEX: Progressive Conditional Generative Adversarial Training Using Transfer learning
Obtaining accurate segmentation on large scale images is an open problem in deep learning. The main problem is the amount of labeled data that exists for large scale images. Traditionally, the common solution to this problem is to crop the large images into smaller images to increase the amount of available data and train a Conditional Generative Adversarial Network (CGAN). CGANs are currently the state of the art in image to image translation and provide better accuracy than the traditional method of training a encoder based conv-net architecture to minimize the loss at each pixel. This method can produce noisy and discontinuous images with inaccurate results. We seek to solve this problem by utilizing the concepts of transfer learning and progressive training to create a CGAN that can segment large scale images with a limited amount of labeled data. In transfer learning we recognize that many learned features are applicable to many classes from multiple domains. This introduces the concept of feature reusability, which is the basis for finetuning. Progressive training got its start in training models on the same images at different resolutions. In this work we instead train a GAN on increasing image scales by transferring the weights from the smaller scales to the larger scales. The learned features at the smaller scales are continually reused and applied to larger scales to create a CGAN that can perform accurate segmentation on large scale images. We apply this method to detect building footprints on very high-resolution overhead imagery (e.g Digital Globe and high resolution airborne platforms).
Linear Subpixel Learning Algorithm for Land Cover Classification from WELD using High Performance Computing
In this work, we use a Fully Constrained Least Squares Subpixel Learning Algorithm to unmix global WELD (Web Enabled Landsat Data) to obtain fractions or abundances of substrate (S), vegetation (V) and dark objects (D) classes. Because of the sheer nature of data and compute needs, we leveraged the NASA Earth Exchange (NEX) high performance computing architecture to optimize and scale our algorithm for large-scale processing. Subsequently, the S-V-D abundance maps were characterized into 4 classes namely, forest, farmland, water and urban areas (with NPP-VIIRS-national polar orbiting partnership visible infrared imaging radiometer suite nighttime lights data) over California, USA using Random Forest classifier. Validation of these land cover maps with NLCD (National Land Cover Database) 2011 products and NAFD (North American Forest Dynamics) static forest cover maps showed that an overall classification accuracy of over 91 percent was achieved, which is a 6 percent improvement in unmixing based classification relative to per-pixel-based classification. As such, abundance maps continue to offer an useful alternative to high-spatial resolution data derived classification maps for forest inventory analysis, multi-class mapping for eco-climatic models and applications, fast multi-temporal trend analysis and for societal and policy-relevant applications needed at the watershed scale.
GEONEX: Land Monitoring From a New Generation of Geostationary Satellite Sensors
The latest generation of geostationary satellites carry sensors such as ABI (Advanced Baseline Imager on GOES-16) and the AHI (Advanced Himawari Imager on Himawari) that closely mimic the spatial and spectral characteristics of Earth Observing System flagship MODIS for monitoring land surface conditions. More importantly they provide observations at 5-15 minute intervals. Such high frequency data offer exciting possibilities for producing robust estimates of land surface conditions by overcoming cloud cover, enabling studies of diurnally varying local-to-regional biosphere-atmosphere interactions, and operational decision-making in agriculture, forestry and disaster management. But the data come with challenges that need special attention. For instance, geostationary data feature changing sun angle at constant view for each pixel, which is reciprocal to sun-synchronous observations, and thus require careful adaptation of EOS algorithms. Our goal is to produce a set of land surface products from geostationary sensors by leveraging NASA's investments in EOS algorithms and in the data/compute facility NEX. The land surface variables of interest include atmospherically corrected surface reflectances, snow cover, vegetation indices and leaf area index (LAI)/fraction of photosynthetically absorbed radiation (FPAR), as well as land surface temperature and fires. In order to get ready to produce operational products over the US from GOES-16 starting 2018, we have utilized 18 months of data from Himawari AHI over Australia to test the production pipeline and the performance of various algorithms for our initial tests. The end-to-end processing pipeline consists of a suite of modules to (a) perform calibration and automatic georeference correction of the AHI L1b data, (b) adopt the Multi-Angle Implementation of Atmospheric Correction (MAIAC) algorithm to produce surface spectral reflectances along with compositing schemes and QA, and (c) modify relevant EOS retrieval algorithms (e.g., LAI and FPAR, GPP, etc.) for subsequent science product generation. Initial evaluation of Himawari AHI products against standard MODIS products indicate general agreement, suggesting that data from geostationary sensors can augment low earth orbit (LEO) satellite observations.
Very High Resolution Tree Cover Mapping for Continental United States using Deep Convolutional Neural Networks
Uncertainties in input land cover estimates contribute to a significant bias in modeled above ground biomass (AGB) and carbon estimates from satellite-derived data. The resolution of most currently used passive remote sensing products is not sufficient to capture tree canopy cover of less than ca. 10-20 percent, limiting their utility to estimate canopy cover and AGB for trees outside of forest land. In our study, we created a first of its kind Continental United States (CONUS) tree cover map at a spatial resolution of 1-m for the 2010-2012 epoch using the USDA NAIP imagery to address the present uncertainties in AGB estimates. The process involves different tasks including data acquisition ingestion to pre-processing and running a state-of-art encoder-decoder based deep convolutional neural network (CNN) algorithm for automatically generating a tree non-tree map for almost a quarter million scenes. The entire processing chain including generation of the largest open source existing aerial satellite image training database was performed at the NEX supercomputing and storage facility. We believe the resulting forest cover product will substantially contribute to filling the gaps in ongoing carbon and ecological monitoring research and help quantifying the errors and uncertainties in derived products.
Integrating Satellite and Surface Sensor Networks for Irrigation Management Applications in California
Satellite mapping of evapotranspiration (ET) from irrigated agricultural lands can provide agricultural producers and water managers with information that can be used to optimize agricultural water use, especially in regions with limited water supplies. The timely delivery of information on agricultural crop water requirements has the potential to make irrigation scheduling more practical, convenient, and accurate. We present a system for irrigation scheduling and management support in California and describe lessons learned from the development and implementation of the system. The Satellite Irrigation Management Support (SIMS) framework integrates satellite data with information from agricultural weather networks to map crop canopy development, basal crop coefficients (Kcb), and basal crop evapotranspiration (ETcb) at the scale of individual fields. Information is distributed to agricultural producers and water managers via a web-based irrigation management decision support system and web data services. SIMS also provides an application programming interface (API) that facilitates integration with other irrigation decision support tools, estimation of total crop evapotranspiration (ETc) and calculation of on-farm water use efficiency metrics. Accuracy assessments conducted in commercial fields for more than a dozen crop types to date have shown that SIMS seasonal ETcb estimates are within 10 mean absolute error (MAE) for well-watered crops and within 15 across all crop types studied, and closely track daily ETc and running totals of ETc measured in each field. Use of a soil water balance model to correct for soil evaporation and crop water stress reduces this error to less than 8 MAE across all crop types studied to date relative to field measurements of ETc. Results from irrigation trials conducted by the project for four vegetable crops have also demonstrated the potential for use of ET-based irrigation management strategies to reduce total applied water by 20-40 relative to grower standard practices while maintaining crop yields and quality.
IN13B-1660: Analytics and Visualization Pipelines for Big Data on the NASA Earth Exchange (NEX) and OpenNEX
We are developing capabilities for an integrated petabyte-scale Earth science collaborative analysis and visualization environment. The ultimate goal is to deploy this environment within the NASA Earth Exchange (NEX) and OpenNEX in order to enhance existing science data production pipelines in both high-performance computing (HPC) and cloud environments. Bridging of HPC and cloud is a fairly new concept under active research and this system significantly enhances the ability of the scientific community to accelerate analysis and visualization of Earth science data from NASA missions, model outputs and other sources. We have developed a web-based system that seamlessly interfaces with both high-performance computing (HPC) and cloud environments, providing tools that enable science teams to develop and deploy large-scale analysis, visualization and QA pipelines of both the production process and the data products, and enable sharing results with the community. Our project is developed in several stages each addressing separate challenge - workflow integration, parallel execution in either cloud or HPC environments and big-data analytics or visualization. This work benefits a number of existing and upcoming projects supported by NEX, such as the Web Enabled Landsat Data (WELD), where we are developing a new QA pipeline for the 25PB system.
Uncertainty Assessment of the NASA Earth Exchange Global Daily Downscaled Climate Projections (NEX-GDDP) Dataset
The NASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP) dataset is comprised of downscaled climate projections that are derived from 21 General Circulation Model (GCM) runs conducted under the Coupled Model Intercomparison Project Phase 5 (CMIP5) and across two of the four greenhouse gas emissions scenarios (RCP4.5 and RCP8.5). Each of the climate projections includes daily maximum temperature, minimum temperature, and precipitation for the periods from 1950 through 2100 and the spatial resolution is 0.25 degrees (approximately 25 km x 25 km). The GDDP dataset has received warm welcome from the science community in conducting studies of climate change impacts at local to regional scales, but a comprehensive evaluation of its uncertainties is still missing. In this study, we apply the Perfect Model Experiment framework (Dixon et al. 2016) to quantify the key sources of uncertainties from the observational baseline dataset, the downscaling algorithm, and some intrinsic assumptions (e.g., the stationary assumption) inherent to the statistical downscaling techniques. We developed a set of metrics to evaluate downscaling errors resulted from bias-correction ("quantile-mapping"), spatial disaggregation, as well as the temporal-spatial non-stationarity of climate variability. Our results highlight the spatial disaggregation (or interpolation) errors, which dominate the overall uncertainties of the GDDP dataset, especially over heterogeneous and complex terrains (e.g., mountains and coastal area). In comparison, the temporal errors in the GDDP dataset tend to be more constrained. Our results also indicate that the downscaled daily precipitation also has relatively larger uncertainties than the temperature fields, reflecting the rather stochastic nature of precipitation in space. Therefore, our results provide insights in improving statistical downscaling algorithms and products in the future.
DeepSAT: A Deep Learning Approach to Tree-Cover Delineation in 1-m NAIP Imagery for the Continental United States
High resolution tree cover classification maps are needed to increase the accuracy of current land ecosystem and climate model outputs. Limited studies are in place that demonstrates the state-of-the-art in deriving very high resolution (VHR) tree cover products. In addition, most methods heavily rely on commercial softwares that are difficult to scale given the region of study (e.g. continents to globe). Complexities in present approaches relate to (a) scalability of the algorithm, (b) large image data processing (compute and memory intensive), (c) computational cost, (d) massively parallel architecture, and (e) machine learning automation. In addition, VHR satellite datasets are of the order of terabytes and features extracted from these datasets are of the order of petabytes. In our present study, we have acquired the National Agriculture Imagery Program (NAIP) dataset for the Continental United States at a spatial resolution of 1-m. This data comes as image tiles (a total of quarter million image scenes with ~60 million pixels) and has a total size of ~65 terabytes for a single acquisition. Features extracted from the entire dataset would amount to ~8-10 petabytes. In our proposed approach, we have implemented a novel semi-automated machine learning algorithm rooted on the principles of "deep learning" to delineate the percentage of tree cover. Using the NASA Earth Exchange (NEX) initiative, we have developed an end-to-end architecture by integrating a segmentation module based on Statistical Region Merging, a classification algorithm using Deep Belief Network and a structured prediction algorithm using Conditional Random Fields to integrate the results from the segmentation and classification modules to create per-pixel class labels. The training process is scaled up using the power of GPUs and the prediction is scaled to quarter million NAIP tiles spanning the whole of Continental United States using the NEX HPC supercomputing cluster. An initial pilot over the state of California spanning a total of 11,095 NAIP tiles covering a total geographical area of 163,696 sq. miles has produced true positive rates of around 88 percent for fragmented forests and 74 percent for urban tree cover areas, with false positive rates lower than 2 percent for both landscapes.
GC31G-1182: Opennex, a Private-Public Partnership in Support of the National Climate Assessment
The NASA Earth Exchange (NEX) is a collaborative computing platform that has been developed with the objective of bringing scientists together with the software tools, massive global datasets, and supercomputing resources necessary to accelerate research in Earth systems science and global change. NEX is funded as an enabling tool for sustaining the national climate assessment. Over the past five years, researchers have used the NEX platform and produced a number of data sets highly relevant to the National Climate Assessment. These include high-resolution climate projections using different downscaling techniques and trends in historical climate from satellite data. To enable a broader community in exploiting the above datasets, the NEX team partnered with public cloud providers to create the OpenNEX platform. OpenNEX provides ready access to NEX data holdings on a number of public cloud platforms along with pertinent analysis tools and workflows in the form of Machine Images and Docker Containers, lectures and tutorials by experts. We will showcase some of the applications of OpenNEX data and tools by the community on Amazon Web Services, Google Cloud and the NEX Sandbox.
Using Docker Containers to Extend Reproducibility Architecture for the NASA Earth Exchange (NEX)
NASA Earth Exchange (NEX) is a data, supercomputing and knowledge collaboratory that houses NASA satellite, climate and ancillary data where a focused community can come together to address large-scale challenges in Earth sciences. As NEX has been growing into a petabyte-size platform for analysis, experiments and data production, it has been increasingly important to enable users to easily retrace their steps, identify what datasets were produced by which process chains, and give them ability to readily reproduce their results. This can be a tedious and difficult task even for a small project, but is almost impossible on large processing pipelines. We have developed an initial reproducibility and knowledge capture solution for the NEX, however, if users want to move the code to another system, whether it is their home institution cluster, laptop or the cloud, they have to find, build and install all the required dependencies that would run their code. This can be a very tedious and tricky process and is a big impediment to moving code to data and reproducibility outside the original system. The NEX team has tried to assist users who wanted to move their code into OpenNEX on Amazon cloud by creating custom virtual machines with all the software and dependencies installed, but this, while solving some of the issues, creates a new bottleneck that requires the NEX team to be involved with any new request, updates to virtual machines and general maintenance support. In this presentation, we will describe a solution that integrates NEX and Docker to bridge the gap in code-to-data migration. The core of the solution is saemi-automatic conversion of science codes, tools and services that are already tracked and described in the NEX provenance system, to Docker - an open-source Linux container software. Docker is available on most computer platforms, easy to install and capable of seamlessly creating and/or executing any application packaged in the appropriate format. We believe this is an important step towards seamless process deployment in heterogeneous environments that will enhance community access to NASA data and tools in a scalable way, promote software reuse, and improve reproducibility of scientific results.
Using Cloud-Based Storage Technologies for Earth Science Data
Cloud based infrastructure may offer several key benefits of scalability, built in redundancy and reduced total cost of ownership as compared with a traditional data center approach. However, most of the tools and software systems developed for NASA data repositories were not developed with a cloud based infrastructure in mind and do not fully take advantage of commonly available cloud-based technologies. Object storage services are provided through all the leading public (Amazon Web Service, Microsoft Azure, Google Cloud, etc.) and private (Open Stack) clouds, and may provide a more cost-effective means of storing large data collections online. We describe a system that utilizes object storage rather than traditional file system based storage to vend earth science data. The system described is not only cost effective, but shows superior performance for running many different analytics tasks in the cloud. To enable compatibility with existing tools and applications, we outline client libraries that are API compatible with existing libraries for HDF5 and NetCDF4. Performance of the system is demonstrated using clouds services running on Amazon Web Services.
Uncertainty Assessment of the NASA Earth Exchange Global Daily Downscaled Climate Projections (NEX-GDDP) Dataset
The NASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP) dataset is comprised of downscaled climate projections that are derived from 21 General Circulation Model (GCM) runs conducted under the Coupled Model Intercomparison Project Phase 5 (CMIP5) and across two of the four greenhouse gas emissions scenarios (RCP4.5 and RCP8.5). Each of the climate projections includes daily maximum temperature, minimum temperature, and precipitation for the periods from 1950 through 2100 and the spatial resolution is 0.25 degrees (approximately 25 km by 25 km). The GDDP dataset has received warm welcome from the science community in conducting studies of climate change impacts at local to regional scales, but a comprehensive evaluation of its uncertainties is still missing. In this study, we apply the Perfect Model Experiment framework (Dixon et al. 2016) to quantify the key sources of uncertainties from the observational baseline dataset, the downscaling algorithm, and some intrinsic assumptions (e.g., the stationary assumption) inherent to the statistical downscaling techniques. We developed a set of metrics to evaluate downscaling errors resulted from bias-correction ("quantile-mapping"), spatial disaggregation, as well as the temporal-spatial non-stationarity of climate variability. Our results highlight the spatial disaggregation (or interpolation) errors, which dominate the overall uncertainties of the GDDP dataset, especially over heterogeneous and complex terrains (e.g., mountains and coastal area). In comparison, the temporal errors in the GDDP dataset tend to be more constrained. Our results also indicate that the downscaled daily precipitation also has relatively larger uncertainties than the temperature fields, reflecting the rather stochastic nature of precipitation in space. Therefore, our results provide insights in improving statistical downscaling algorithms and products in the future.
A High Performance Computing Approach to Tree Cover Delineation in 1-m NAIP Imagery Using a Probabilistic Learning Framework
Tree cover delineation is a useful instrument in deriving Above Ground Biomass (AGB) density estimates from Very High Resolution (VHR) airborne imagery data. Numerous algorithms have been designed to address this problem, but most of them do not scale to these datasets, which are of the order of terabytes. In this paper, we present a semi-automated probabilistic framework for the segmentation and classification of 1-m National Agriculture Imagery Program (NAIP) for tree-cover delineation for the whole of Continental United States, using a High Performance Computing Architecture. Classification is performed using a multi-layer Feedforward Backpropagation Neural Network and segmentation is performed using a Statistical Region Merging algorithm. The results from the classification and segmentation algorithms are then consolidated into a structured prediction framework using a discriminative undirected probabilistic graphical model based on Conditional Random Field, which helps in capturing the higher order contextual dependencies between neighboring pixels. Once the final probability maps are generated, the framework is updated and re-trained by relabeling misclassified image patches. This leads to a significant improvement in the true positive rates and reduction in false positive rates. The tree cover maps were generated for the whole state of California, spanning a total of 11,095 NAIP tiles covering a total geographical area of 163,696 sq. miles. The framework produced true positive rates of around 88% for fragmented forests and 74% for urban tree cover areas, with false positive rates lower than 2% for both landscapes. Comparative studies with the National Land Cover Data (NLCD) algorithm and the LiDAR canopy height model (CHM) showed the effectiveness of our framework for generating accurate high-resolution tree-cover maps.
Continental Spatio-Temporal Data Analysis with Linear Spectral Mixture Model Using FOSS
This work demonstrates the development and implementation of a Fully Constrained Least Squares (FCLS) unmixing model developed in C++ programming language with OpenCV package and boost C++ libraries in the NASA Earth Exchange (NEX). Visualization of the results is supported by GRASS GIS and statistical analysis is carried in R in a Linux system environment. FCLS was first tested on computer simulated data with Gaussian noise of various signal-to-noise ratio, and Landsat data of an agricultural scenario and an urban environment using a set of global end members of substrate (soils, sediments, rocks, and non-photosynthetic vegetation), vegetation that includes green photosynthetic plants and dark objects which encompasses absorptive substrate materials, clear water, deep shadows, etc. For the agricultural scenario, a spectrally diverse collection of 11 scenes of Level 1 terrain corrected, cloud free Landsat-5 TM data of Fresno, California, USA were unmixed and the results were validated with the corresponding ground data. To study an urbanized landscape, a clear sky Landsat-5 TM data were unmixed and validated with coincident World View-2 abundance maps (of 2 m spatial resolution) for an area of San Francisco, California, USA. The results were evaluated using descriptive statistics, correlation coefficient, RMSE, probability of success, boxplot and bivariate distribution function. Finally, FCLS was used for sub-pixel land cover analysis of the monthly WELD (Wen-enabled Landsat data) repository from 2008 to 2011 of North America. The abundance maps in conjunction with DMSP-OLS nighttime lights data were used to extract the urban land cover features and analyze their spatial-temporal growth.
Mapping Urban Expansion Across North America Using Multi-Temporal Landsat and Nighttime Lights Data
Urban expansion and the associated changes in land cover have important climatic, hydrologic, biophysical and ecologic and socio-economic impacts on the environment. Yet, despite todays abundance of remote sensing data, an automated characterization of large-scale historical changes in urban spatial extent remains a challenge due to the inherent complexity and variability of the urban environment, the lack of a spectral signature unique to urban land cover, and the absence of an unambiguous definition of what is urban versus non-urban.Here we present a consistent, robust, scalable, physically- based methodology for characterization of urban expansion using Landsat observations. We use atmospherically corrected Landsat Global Land Survey time series, Web-enabled Landsat data time series, DMSP-OLS and NPP-VIIRS nighttime lights, for mapping the built-up and vegetated components of urban settlements at 30m resolution through multi- temporal standardized spectral mixture analysis. The methodology is tested and validated over the North American continent where it provides a first quantification of urban expansion and vegetation abundance changes from 1990 to 2010.
Generating Global Leaf Area Index from Landsat: Algorithm Formulation and Demonstration
This paper summarizes the implementation of a physically based algorithm for the retrieval of vegetation green Leaf Area Index (LAI) from Landsat surface reflectance data. The algorithm is based on the canopy spectral invariants theory and provides a computationally efficient way of parameterizing the Bidirectional Reflectance Factor (BRF) as a function of spatial resolution and wavelength. LAI retrievals from the application of this algorithm to aggregated Landsat surface reflectances are consistent with those of MODIS for homogeneous sites represented by different herbaceous and forest cover types. Example results illustrating the physics and performance of the algorithm suggest three key factors that influence the LAI retrieval process: 1) the atmospheric correction procedures to estimate surface reflectances; 2) the proximity of Landsatobserved surface reflectance and corresponding reflectances as characterized by the model simulation; and 3) the quality of the input land cover type in accurately delineating pure vegetated components as opposed to mixed pixels. Accounting for these factors, a pilot implementation of the LAI retrieval algorithm was demonstrated for the state of California utilizing the Global Land Survey (GLS) 2005 Landsat data archive. In a separate exercise, the performance of the LAI algorithm over California was evaluated by using the short-wave infrared band in addition to the red and near-infrared bands. Results show that the algorithm, while ingesting the short-wave infrared band, has the ability to delineate open canopies with understory effects and may provide useful information compared to a more traditional two-band retrieval. Future research will involve implementation of this algorithm at continental scales and a validation exercise will be performed in evaluating the accuracy of the 30-m LAI products at several field sites. ©