Conference PaperPDF Available

The Free Brazilian Repository for Open Soil Data

April 2018

Conference: Latin America and the Caribbean Scientific Data Management Workshop
At: Rio de Janeiro, Brazil

Project: Data Repository of Brazilian Soils (FEBR)

Authors:

Alessandro Samuel-Rosa

Federal University of Technology - Paraná/Brazil (UTFPR)

Brazilian soil science has produced a great deal of data. Most of the information is published as a single paper, and the primary data is unavailable to other researchers. Lately, soil scientists have increased their concerns with data discoverability and reusability, and reproducible research. To address this issue, Brazilian soil scientists have recently created a data repository using community-built standards and following open data policies. The Free Brazilian Repository for Open Soil Data-febr, www.ufsm.br/febr-is a centralized repository targeted at storing open soil data and serving it in a standardized and harmonized format. The repository infrastructure was built using open source and/or free (of cost) software, and was primarily designed for the individual management of datasets. This is accomplished by storing each dataset using a collection of Google spreadsheets accessible online.

Content uploaded by Alessandro Samuel-Rosa

Content may be subject to copyright.

A preview of the PDF is not available

ResearchGate has not been able to resolve any citations for this publication.

ResearchGate has not been able to resolve any references for this publication.

Project

Digital soil mapping in the Brazilian Amazon -- using the knowledge from a reference area

The Brazilian Amazon region is a huge territory encompassing different ecosystems. Most of the existing soil information was produced during the 1970s and 1980s by the RADAMBRASIL Project, the soil maps being published at a coarse scale (1:250 000). Localized more detailed soil information exists but is very sparse. The lack of detailed soil information for the Brazilian Amazon is mainly due the following reasons: a) the access to most areas is only possible by boat or airplane, b) the Tropical Forest offers a great variety of potential diseases such as leishmaniasis, malaria and yellow fever, and c) there is a lack of more detailed base maps. These aspects call for the use of alternative, less field intensive methods to produce detailed soil information for the Brazilian Amazon region. In this study, we compare the use of linear and nonlinear statistical tools to learn the rules of soil-landscape relationships constructed in a reference mapped area to make soil spatial predictions in poorly accessible areas of the Brazilian Amazon. Preliminary results indicate that the choice of the statistical learner depends on the target category being modelled, non-linear methods having a slightly superior performance than linear ones. ... [more]

Project

Spatial modelling of elemental stocks in Amazonian Dark Earths

We aim at modelling elemental (carbon, calcium and phosphorus) soil stocks in Amazonian Dark Earths using geostatistical methods taking into account the deterministic effect of environmental covari ates as well as the horizontal and vertical soil spatial (auto)correlation. ... [more]

Project

Optimization of sample configurations for spatial trend estimation for soil mapping

The spatial trend is the part of a soil spatial model that deterministically explains the soil spatial variation using covariates. The sampling method most commonly used to identify and estimate th e spatial trend is conditioned Latin hypercube sampling (CLHS). In this study we propose conceptual and algorithmic improvements on CLHS, which are evaluated using synthetic data derived from a real-world case study in southern Brazil. The improvements include: 1) using the Pearson's r only when all covariates are numeric, and the Cramér's V when some or all covariates are factors, 2) defining marginal sampling strata using only the unique values of the sample quantiles estimated with a discontinuous function, and 3) scaling the objective functions to the same approximate range of values using the upper-lower bound approach with the Pareto maximum and minimum before aggregating them into a single utility function using a weighted sum. Compared to the original CLHS, our proposed modifications resulted in a sampling algorithm, which we named ACDC, with an improved numerical behaviour that translated into improved calibration and prediction performance – almost 5% more variance explained – for samples of size n = 100, 200, and 400. Sample size had a larger influence on prediction accuracy than the sampling algorithm, but the benefits of using ACDC instead of CLHS were remarkably greater for smaller sample sizes. This suggests that the less resources we have, the more strict about the tools we use we should be. ... [more]

Project

Mapeamento Digital de Solos por Regressão Logística Múltipla em Ambiente de Mar de Morros Pinheiral, RJ

O levantamento de solos compreende a descrição das características morfológicas, físicas e químicas dos solos, a descrição da superfície e a determinação da distribuição dos solos na paisagem de um a determinada área, e pode ser utilizado para diversos fins. O conhecimento das características dos solos e sua distribuição na paisagem são fundamentais para o planejamento do uso e conservação das terras. O uso de técnicas de mapeamento digital de solos (MDS) pode otimizar a realização e aperfeiçoar os produtos dos levantamentos pedológicos através de ferramentas de sistemas de informações geográficas (SIG). O objetivo geral do projeto envolve a criação de banco de dados com dados legados e novos pontos de amostragem tendo como área de estudo a Fazenda São José do Pinheiro e adjacência, no município de Pinheiral (RJ), bem como a avaliação detalhada da representação da altimetria na área (avaliação do modelo digital de elevação - MDE) e sua relação com a acurácia do mapa de solo. Também é objetivo deste projeto proceder com a correção atmosférica na imagem de satélite (Rapid Eye) e avaliar o efeito da correção sobre o mapeamento de uso da terra além da avaliação do erro posicional da imagem. Por fim, proceder-se-á com o MDS em ambiente de Mar de Morros, aplicando a técnica de regressão logística múltipla, sendo avaliado se esse método permite boa acurácia dos mapas de classes de solos e unidades de mapeamento, que possam constituir material base para estudos ambientais e planejamento do uso das terras.. ... [more]

Article

Lunar soft landing trajectory optimization in a 6DoF dynamical model

January 2011

An algorithm for the optimization of a lunar soft-landing trajectory is presented. A 6DoF modeling of the dynamics is adopted together with an accurate description of the Moon gravity field. The problem is faced as a direct optimization problem with the goal of obtaining a vertical landing whilst minimizing the overall fuel consumption. The descent trajectory is supposed to start from the ... [Show full abstract] periselenium of a low Moon orbit. Four optimization phases are considered. Each phase is characterized by a different set of optimization variables, constraints, and increasing level of complexity. In the first phase the thrust direction is optimized considering the translational motion of the lander only. Furthermore, no throttle capability is considered. In the second phase the thrust direction is fixed in the spacecraft body reference frame. The proper thrust orientation is obtained by optimizing the control torques supplied to the lander by the attitude sub-system. In the third phase the thrust magnitude is optimized too, and the constraint of landing on specific site is added. Furthermore, more restrictive constraints on the final velocities (linear and angular) are set. Finally, in the fourth phase a more accurate gravitational model of Moon that includes the main harmonics is considered. The algorithm is tested on a lunar landing trajectory in an area close to the Moon's equator, with a candidate spacecraft for Google Lunar X-prize.

Article

An Overview of Open Access in the Fields of Business and Management

April 2011 · Journal of Business & Finance Librarianship

The extent to which scholarly literature can be openly accessed and freely downloaded on the Internet varies greatly depending on the discipline. This article assesses the current state of open access in the fields of business and management. Google was used to search for free versions of articles from a sample of articles published in early 2009 in a broad array of business and management ... [Show full abstract] journals. Free versions of 36% of the articles included in this study were identified, available from a wide variety of websites, including repositories, author profile pages, open access journals, and more.

Preprint

Resource-Efficient Neural Architect

June 2018

Neural Architecture Search (NAS) is a laborious process. Prior work on automated NAS targets mainly on improving accuracy, but lacks consideration of computational resource use. We propose the Resource-Efficient Neural Architect (RENA), an efficient resource-constrained NAS using reinforcement learning with network embedding. RENA uses a policy network to process the network embeddings to ... [Show full abstract] generate new configurations. We demonstrate RENA on image recognition and keyword spotting (KWS) problems. RENA can find novel architectures that achieve high performance even with tight resource constraints. For CIFAR10, it achieves 2.95% test error when compute intensity is greater than 100 FLOPs/byte, and 3.87% test error when model size is less than 3M parameters. For Google Speech Commands Dataset, RENA achieves the state-of-the-art accuracy without resource constraints, and it outperforms the optimized architectures with tight resource constraints.

Article

Building smartphone apps by using free cloud services from Facebook, Dropbox and Google

March 2016 · Frontiers in Artificial Intelligence and Applications

Businesses are interested in developing smartphone apps due to huge market potentials. However, it is laborious, time-consuming and costly to establish the needed information systems environment including the infrastructure, platform and software to support smartphone apps, beyond the reach of many companies especially for small and medium-sized enterprises (SMEs) with limited expertise and ... [Show full abstract] resource. Our study aims to alleviate this issue by utilizing free Internet services to build smartphone apps. In this article, we not only describe how to build smartphone apps using applicable free Internet services, but also discuss the advantages and values of our approach by comparing our app which is built by the proposed approach with two other similar and popular smartphone apps.