Dados do Trabalho

Título

THE BRAZILIAN SATELLITE SOIL SPECTRAL LIBRARY: IMPACTS ON SOCIETY

Resumo

Understanding soils are the basis for any land use planning. Topics regarding agriculture, environmental monitoring, and soil health are some of today's most debated ones. The objective of this work was to describe the Brazilian soil spectral signatures acquired by Landsat and sentinel satellites. We acquired 22,000 sites all over Brazil and depicted the respective reflectance of all bands. We constructed the spectral signatures based on clay, carbon and Fe2O3 contents of samples from wet laboratory analysis. Soil textural classes presented different morphological behavior; as clay increases, reflectance intensity decreases. As more quartz, the reflectance in SWIR2 is higher. The iron causes a decrease in reflectance because it absorbs energy. Therefore, reflectance data can differentiate the soil iron content from hypoferric, low content of iron oxides (< 8%), to perferric, very high content of iron oxides (≥ 36%). Carbon also presents specific behavior where a concave shape behavior occurs from band 1 to 4 and consistently the higher the content the lower the reflectance. In general, quantification by cubist method presented models with R2 of 0.47, 0.81, 0.80 for both sensors. Quantitative data ratified qualitative ones. The SWIR2 band reached higher importance (40%) than others in prediction. Differently from Landsat, Sentinel-2’s sensor has the readedge band, which had indicated higher significance to carbon modeling. Therefore, understanding spectral behavior is fundamental to the modeling results interpretations. For instance, human ponderings are crucial to assist the machine learning approaches. The results can have a direct impact on soil mapping programmes such as PronaSolos and, by extension, on professional activities.  

Palavras-chave

soil health; soil sensing; pedology; soil mapping and survey; laboratory.

Instituição financiadora

The State of São Paulo Research Foundation (FAPESP) 2021/05129-8.

Agradecimentos

The authors thanks GeoCis Research Group for sampling and analyses.