Crop area and leaf area index simultaneous retrieval based on spatial scaling transformation

Accurate estimation of crop yields is crucial for ensuring food security. However, crops are distributed so fragmentally in China that mixed pixels account for a large proportion in moderate and coarse resolution remote sensing images. As a result, unmixing of mixed pixel becomes a major problem to estimate crop yield by means of remote sensing method. Aimed at mixed pixels, we developed a new method to introduce additional information contained in the spatial scaling transformation equation to the canopy reflectance model. The crop area and LAI can be retrieved simultaneously. On the basis of a precise and simple canopy reflectance model, directional second derivative method was chosen to retrieve LAI from optimal bands of hyper-spectral data; this method can reduce the impact of the canopy non-isotropic features and soil background. To evaluate the performance of the method, Yingke Oasis, Zhangye City, Gansu Province, was chosen as the validation area. This area was covered mainly by maize and wheat. A Hyperion/EO-1 image with the 30 m spatial resolution was acquired on July 15, 2008. Images of 180 m and 1080 m resolutions were generated by linearly interpolating the original Hyperion image to coarser resolutions. Then a multi-scale image serial was obtained. Using the proposed method, we calculated crop area and the average LAI of every 1080 m pixel. A SPOT-5 classification figure serves as the validation data of crop area proportion. Results show that the pattern of crop distribution accords with the classification figure. The errors are restrained mainly to −0.1–0.1, and approximate a Normal Distribution. Meanwhile, 85 LAI values obtained using LAI-2000 Plant Canopy Analyzer, equipped with GPS, were taken as the ground reference. Results show that the standard deviation of the errors is 0.340. The method proposed in the paper is reliable.