Clay content prediction using spectra data collected from the ground to space platforms in a smallholder tropical area

Autor(es):Mostrar menos -

Bellinaso, Henrique ^{[1, 2]} ; Silvero, Nelida E. Q. ^[1] ; Chimelo Ruiz, Luis Fernando ^[1] ; Accorsi Amorim, Merilyn Taynara ^[1] ; Rosin, Nicolas Augusto ^[1] ; Mendes, Wanderson de Sousa ^[1] ; Barbosa de Sousa, Gabriel Pimenta ^[1] ; Araujo Sepulveda, Leno Marcio ^[1] ; de Queiroz, Louise Gunter ^[1] ; Nanni, Marcos Rafael ^[3] ; Dematte, Jose A. M. ^[1]

Afiliação do(s) autor(es)

^[1] Univ Sao Paulo, Luiz de Queiroz Coll Agr, Dept Soil Sci, Padua Dias Av 11, Postal Box 09, BR-13416900 Sao Paulo - Brazil

^[2] EDR Piracicaba, Secretariat Agr & Supply CDRS SAA, Coordinat Sustainable Rural Dev, Campos Salles St 507, Piracicaba, SP - Brazil

^[3] Univ Estadual Maringa, Dept Agron, Colombo Av 5790, Maringa, Parana - Brazil

• Tipo de documento: Artigo Científico
• Fonte: Geoderma; v. 399, OCT 1 2021.
• Citações Web of Science: 0

Resumo

Proximal and remote sensors are emerging as powerful sources of soil spectral information at an array of temporal and spatial resolutions. This study investigated clay content prediction at three spectral acquisition levels: laboratory, airborne, and spaceborne. Two approaches were tested, the use of prediction models developed with local and regional spectral libraries (52 samples for local scale and 950, 200 e 224 samples for regional scale), termed internal and external models respectively. Local soil samples (52), were collected in a smallholder area, 83 ha, located in southeastern Brazil. Spectral data in the visible (Vis), near-infrared (NIR), and shortwave infrared (SWIR) regions were acquired in the laboratory using FieldSpec 3 sensor, and the clay content was determined by sedimentation technique. Afterward, bare soil images from AISA-FENIX, Planetscope, Sentinel-2 MultiSpectral Instrument (MSI) and Landsat-8 Operational Land Imager (OLI) were obtained. The clay content determined in the laboratory was related to the soil spectra acquired by each of the sensors and was predicted using the Cubist regression tree algorithm. The results obtained from local spectral libraries showed good predictions using FieldSpec 3 and AISA-FENIX sensors. Landsat-8 OLI and Sentinel-2 MSI provided satisfactory results, while PlanetScope gave poor results. For the prediction using regional spectral libraries, only lab-based FieldSpec 3 sensor provided a fair prediction, while other sensors gave poor results. This study demonstrated that soil sensing is possible at any level taking into account its advantages and limitations. This approach paves the way for acquiring soil spectra for smallholder farms. (AU)

• Processo FAPESP: 16/26124-6 - Pedologia de precisão: caracterização e mapeamento de solos em tempo real por geotecnologias
• Beneficiário: Wanderson de Sousa Mendes
• Modalidade de apoio: Bolsas no Brasil - Doutorado Direto
• Processo FAPESP: 14/22262-0 - Geotecnologias no mapeamento digital pedológico detalhado e biblioteca espectral de solos do Brasil: desenvolvimento e aplicações
• Beneficiário: José Alexandre Melo Demattê
• Modalidade de apoio: Auxílio à Pesquisa - Temático
• Processo FAPESP: 16/01597-9 - Pedotransferência por geotecnologias associada à fotopedologia com vistas ao mapeamento pedológico de áreas agrícolas do estado de São Paulo
• Beneficiário: José Lucas Safanelli
• Modalidade de apoio: Bolsas no Brasil - Doutorado Direto