Evapotranspiration and soil moisture estimation using different remote sensing platforms

Carregando...
Imagem de Miniatura

Data

2022-04-19

Autores

Safre, Anderson Luiz dos Santos [UNESP]

TĆ­tulo da Revista

ISSN da Revista

TĆ­tulo de Volume

Editor

Universidade Estadual Paulista (Unesp)

Resumo

Evapotranspiration (ET) and soil moisture (SM) are important parameters for agricultural water management. Sensors mounted on remote sensing platforms such as satellites and Unamed Aerial Vehicles (UAVs) can provide reliable information on crop reflectance at different spatial, temporal and radiometric resolution. Machine learning nonlinear approach have shown the potential of estimating SM using optical images. The Simple Algorithm for Evapotranspiration Retrieval (SAFER) uses data from remote sensing and meteorological stations for the energy balance estimations and then actual evapotranspiration. We evaluated the performance of SAFER to estimate ET. First, we evaluated the results of SAFER using Landsat 8 imagery (30 m pixel size for optical bands and 100 m for thermal within a 18 days revisit) from 2013-2017 and standard coefficients. Then the regression coefficients were calibrated using data from Eddy Covariance (EC) towers and the results from field and remote sensing were compared. The next step was to assess SAFER performance and calibrate the algorithm with Sentinel-2 (10 m pixel size and 5 days revisit) imagery. SAFER results using the thermal band and using only optical bands were compared with six EC flux stations, located at two different sites. We applied Support Vector Regression (SVR), Random Forest (RF) and Artificial Neural Network (ANN) machine learning algorithms to estimate the soil moisture from UAV high-resolution images (2.4 cm/pix). Three bands (G, R, NIR) and NDVI was used as input. After the model calibration SAFER showed good agreement with EC data using Landsat-8 and Sentinel-2. In the Landsat-8 dataset the RMSE was 0.70 mm d-1 using the data from 5 years. The lowest RMSE (0.53 mm d-1) was in 2015 and the highest RMSE (0.89 mm d-1) in 2013. Seasonal ET was estimated and compared with the EC flux towers, showing an R2 that ranged between 0.29 to 0.97. Regarding the SAFER using Sentinel-2 images results, the model RMSE was between 0.62 to 0.84 mm d-1. The model tends to underestimate ET values when there is less water available in the root zone. The seasonal ET estimated using Sentinel-2 images showed a R2 of 0.64, when compared to that from EC measurements. Results show that all three machine learning algorithms had a great performance on the estimation of soil moisture with RMSE < 1%. SVR was the best model with a RMSE of 0.45 % and R2 = 0.71. We conclude that UAVS data and machine learning can be a great tool for soil moisture spatial variability modeling in heterogeneous terrains.
EvapotranspiraĆ§Ć£o (ET) e umidade do solo (US) sĆ£o importantes parĆ¢metros para a gestĆ£o da Ć”gua na agricultura. Sensores equipados em plataformas de sensoriamento remoto, como satĆ©lites e Veiculos Ɓreos NĆ£o Tripulados (VANTā€™s), podem fornecer informaƧƵes confiĆ”veis sobre a reflectĆ¢ncia das culturas em diferentes resoluƧƵes espaciais, temporais e radiomĆ©tricas. A abordagem nĆ£o linear de aprendizado de mĆ”quina mostrou o potencial de estimar a US usando imagens Ć³pticas. O Simple Algorithm for Evapotranspiration Retrieval (SAFER) usa dados de sensoriamento remoto e dados de estaƧƵes meteorolĆ³gicas para estimar a ET. Avaliamos o desempenho do SAFER na estimativa da ET atual. Primeiro, avaliamos os resultados do SAFER usando imagens do Landsat-8 (tamanho do pixel de 30 m e revisita de 18 dias) de 2013-2017 e coeficientes padrƵes. Em seguida, os coeficientes do modelo foram calibrados usando dados de torres Eddy Covariance (EC) e os resultados foram avaliados. A prĆ³xima etapa foi avaliar o desempenho do SAFER e calibrar o modelo nas imagens do Sentinel-2B (tamanho de pixel de 10 m e revisita de 5 dias). Aplicamos algoritmos de aprendizado de mĆ”quina de Support Vector Regression (SVR), Random Forest (RF) e Artificial Neural Network (ANN) para estimar a umidade do solo a partir de imagens de alta resoluĆ§Ć£o de VANT. TrĆŖs bandas (G, R, NIR) e NDVI foram usados como entrada. ApĆ³s a calibraĆ§Ć£o do modelo, o SAFER mostrou boa concordĆ¢ncia com os dados do EC usando Landsat-8 e Sentinel-2. No conjunto de dados Landsat-8, o RMSE foi 0,70 mm d-1 usando os dados de 5 anos. O menor RMSE (0,53 mm d-1) foi em 2015 e o maior RMSE (0,89 mm d-1) em 2013. A ET sazonal foi estimada e comparada com as torres de fluxo EC, apresentado um R2 que variou entre 0,29 a 0,97. Em relaĆ§Ć£o aos resultados do SAFER usando imagens Sentinel-2, o RMSE do modelo ficou entre 0,62 a 0,84 mm d-1. O modelo tende a subestimar os valores de ET quando hĆ” menos Ć”gua disponĆ­vel na zona radicular. A ET sazonal estimada usando imagens Sentinel-2 apresentou R2 de 0,64 quando comparado ao EC. Os resultados mostram que todos os trĆŖs algoritmos de aprendizado de mĆ”quina tiveram um Ć³timo desempenho na estimativa da umidade do solo com RMSE < 1%. O SVR foi o melhor modelo com RMSE de 0,45 % e R2 = 0,71. ConcluĆ­mos que os dados de VANTs e o aprendizado de mĆ”quina podem ser uma Ć³tima ferramenta para modelagem da variabilidade espacial da umidade do solo em terrenos heterogĆŖneos.

DescriĆ§Ć£o

Palavras-chave

Remote sensing, Evapotranspiration, Soil moisture, Eddy covariance, Machine learning, Sensoriamento remoto, EvapotranspiraĆ§Ć£o, Umidade do solo

Como citar