| This article has been：Browse 463Times Download 502Times |
scan it! |
|
|
| DOI：10.13522/j.cnki.ggps.2024186 |
|
| Estimating leaf water potential of potato using band-optimized spectral indices |
|
GAO Kai, YANG Haibo, YIN Hang, WANG Wei, SUN Yu, ZHAO Liang, LI Fei
|
|
1. College of Resources and Environmental Sciences, Inner Mongolia Agricultural University, Huhhot 010011, China
2. Inner Mongolia Key Laboratory of Soil Quality and Nutrient Resources/Key Laboratory of Agricultural Ecological Security and Green Development at Universities of Inner Mongolia Autonomous, Inner Mongolia Agricultural University,
Huhhot 010011, China; 3. Ulanqab Institute of Agriculture and Forestry Sciences, Ulanqab 012000, China
|
| Abstract: |
| 【Objective】Leaf water potential (LWP) is a key physiological trait reflecting plant responses to environmental conditions. However, large-scale field measurements remain challenging. This study explores the feasibility of using spectral imaging for indirect LWP estimation.【Method】A field experiment with a soil water gradient was conducted in a potato field in Wuchuan County and Chahar Right Front Banner, Inner Mongolia. Leaf spectral reflectance and water potential were measured at different growth stages. Moisture-sensitive spectral bands were analyzed, and the optimal combination of difference, ratio, and normalized difference spectral indices was identified within the 350-2 500 nm wavelength range. 【Result】The selection of optimal sensitive bands varied across growth stages and significantly impacted estimation accuracy. For individual growth stages, the most sensitive bands were primarily located in the red-edge and near-infrared regions, while for combined growth stages, the most effective bands were predominantly in the near-infrared region. A strong linear correlation was observed between the optimized spectral index and LWP. The Optimized Normalized Difference Spectral Index (NDSI) was the most effective index for estimating potato LWP at a single growth stage. As growth progressed, the correlation between spectral indices and LWP improved, with the highest estimation accuracy and robustness observed during the starch accumulation stage. At this stage, the coefficient of determination (R2) for training data reached 0.90, with an RMSE of 0.07 MPa and an MRE of 6.70%, while for validation data, R2 was 0.87, RMSE was 0.08 MPa, and MRE was 7.41%. When multiple growth stages were combined, the optimized spectral index enhanced LWP estimation accuracy during the tuber formation and expansion stage. For these stages, the R2 values for training and validation data were 0.64 and 0.53, respectively, with RMSE values of 0.14 MPa and 0.16 MPa, and MRE values of 10.39% and 12.40%. Compared to existing moisture-sensitive spectral indices, the optimized spectral index significantly improved accuracy and stability across combined growth stages.【Conclusion】The newly developed spectral index, based on band optimization, enhances sensitivity and reduces data dispersion compared to existing spectral indices. It significantly improves the stability of the potato LWP estimation model, offering valuable insights for optimizing spectral indices in potato water status monitoring and precision irrigation management. |
| Key words: potato; leaf water potential; water index; band optimization; hyperspectral; growth stages |
|
|
