Impact of uncertainties on assessment and validation of MODIS leaf area index products.

摘要

Green leaf area of vegetation governs the exchanges of energy, mass (e.g., water and CO2) and momentum between the Earth's surface and the atmosphere. Therefore, it is being operationally derived from measurements of the MODerate resolution Imaging Spectroradiometer (MODIS) onboard NASA's Terra and Aqua spacecrafts. The validation of such remote sensing products is significant because it provides accuracy information to users and serves as a reference for further product refinement. The objectives of this research are to validate and refine the MODIS leaf area index (LAI) product, with emphasis on the effects of uncertainties on algorithm performance and validation. The uncertainties are related to MODIS data and products as well as the field and reference data used in validation.; A mismatch between reflectances modeled by the algorithm and MODIS measurements led to anomalies in the Collection 3 MODIS LAI products. Following revision for Collection 4, the main algorithm generates over 80% of the retrievals and LAI estimates agree well with several sets of field measurements. Validation of the Collection 4 LAI product over grasses and cereal crops using data from Alpilles, France suggests that the leaf area product has an accuracy of 0.3 LAI with a precision and uncertainty of 0.23 LAI and 0.25 LAI, respectively.; The existence of geolocation offsets between MODIS observations and predefined MODIS storage bins complicates product validation. The value in a bin may only partially be derived from the location of the bin. To minimize the impact of geolocation offsets, validation of the MODIS LAI product should occur at a resolution at least 8 times the native resolution. Validating multidate composite products can be more complex because bias results from the use of spectral information as the compositing criterion.; In conclusion, validation of the MODIS LAI product yields: (a) valuable insights that lead to algorithm improvements; (b) measures of the overall accuracy of the product; and (c) insights into geolocation issues pertinent to the validation of many MODIS products.