GOOGLE earth engine para mapeamento de culturas agrícolas no Paraná
| Ano de defesa: | 2019 |
|---|---|
| Autor(a) principal: |
Paludo, Alex
 |
| Orientador(a): |
Johann, Jerry Adriani
|
| Banca de defesa: |
Johann, Jerry Adriani
,
Grzegozewski, Denise Maria
,
Mercante, Erivelto
|
| Tipo de documento: | Dissertação |
| Tipo de acesso: | Acesso aberto |
| Idioma: | por |
| Instituição de defesa: |
Universidade Estadual do Oeste do Paraná
Cascavel |
| Programa de Pós-Graduação: |
Programa de Pós-Graduação em Engenharia Agrícola
|
| Departamento: |
Centro de Ciências Exatas e Tecnológicas
|
| País: |
Brasil
|
| Palavras-chave em Português: | |
| Área do conhecimento CNPq: | |
| Link de acesso: | http://tede.unioeste.br/handle/tede/4470 |
Resumo: | Brazilian agricultural production directly influences the country's economy, with national grain production covering a large part of this sector. In this way, the information of area sown with the main agricultural crops has great value in the planning of logistic actions, in public or private policies of the commodities market. One method to obtain more reliable data from this sector of the economy is with the aid of remote sensing, obtaining data more quickly and efficiently. However, in order to work with remote sensing, a high computational capacity is required for data processing. To overcome this problem, the use of cloud data processing in the Google Earth Engine (GEE) platform was employed, which is available to users free of charge, has been used to perform various activities related to orbital remote sensing. Thus, the objective of this work was to map the main summer and winter crops in the state of Paraná, for the 2016/2017 and 2017/2018 harvest years, using the GEE platform. For this purpose, images of the OLI and MSI sensors, images of the digital elevation model (SRTM), false color mosaic segmentation of the sensor images and the Naive Bayes Continuous algorithm as classification methods were used. The mapping process was parameterized separately for each of the 39 microregions of the state. Finally, the crop areas (soybean, 1st and 2nd maize and winter crops) were quantified by municipality and compared with official data and field data. The accuracy of the summer mapping resulted in Global Accuracy ranging from 87.6% (Maize, 2017/2018) to 96.7% (Soybean, 2016/2017), with a kappa index (K) between 72% (Maize, 2017/2018 crop year) and 91% (Soybean, 2016/2017). The linear correlation (r) between the mapped area and the official area per municipality was 0.92 and the agreement index dr = 0.81 for the soybean crop and for the 1st crop corn yielded r = 0.59 and dr = 0.53. For the mapping of 2nd crop maize and winter crops, the Global Accuracy varied between 95% (winter crops, 2018 crop year) and 96.7% (Maize 2nd crop, 2017 crop year), with index kappa (K) between 90% (winter crops, 2018 crop year) and 92% (Maize 2nd crop, 2017 crop year). Between the mapped area and the official area per municipality, r = 0.95 and dr = 0.83 for maize 2ndcrop and r = 0.78 and dr = 0.76 for winter crops were obtained. The linear (r) correlation between the mapped area and the field data was 0.96 and the agreement index dr = 0.86 for the 1st crop corn and for the soybean yield r = 0.96 and p = 0.92 for the crop year 2017/2018, r = 0.79 and dr = 0.71 for maize 2nd crop r = 0.80 and dr = 0.72 for winter crops in the crop year 2017, r = 0.88 and dr = 0.86 for maize crop 2nd crop er = 0.71 and dr = 0.78 for winter crops in the crop year 2018. The mapping of areas with agricultural crops carried out with the GHG platform can be carried out quickly, accurately and efficiently. Through the mappings it is possible to have the spatial distribution of crops per crop plot, as well as the quantification of areas by area of coverage of a company, municipality, microregion, mesoregion and for every state. |