Hyperspectral Estimates of Soil Moisture Content Incorporating Harmonic Indicators and Machine Learning

Soil is one of the most significant natural resources in world, and its health closely related to food security, ecological water security. It basic task soil environmental quality assessment monitor temporal spatial variation properties scientifically reasonably. moisture content (SMC) an important property, which plays role agricultural practice, hydrological process, balance. In this paper, a hyperspectral SMC estimation method for mixed types was proposed combining some spectral processing technologies principal component analysis (PCA). The original spectra were processed by wavelet packet transform (WPT), first-order differential (FOD), harmonic decomposition (HD) successively, then PCA dimensionality reduction used obtain two groups characteristic variables: WPT-FOD-PCA (WFP) WPT-FOD-HD-PCA (WFHP). On basis, three regression models (PCR), partial least squares (PLSR), back propagation (BP) neural network applied compare predictive ability different parameters. Meanwhile, we also compared results with estimates conventional indices. indicate that based on indices have errors. Moreover, BP (WFP-BP WFHP-BP) show more accurate when same variables are selected. For model, choice important. WFHP (WFHP-PCR, WFHP-PLSR, all high accuracy maintain good consistency prediction low values. optimal model determined be WFHP-BP R2 0.932 error below 2%. This study can provide information farm entropy before planting crops arable land as well technical reference estimating from images (satellite UAV, etc.).