Seasonal Changes of the Relationships Between Photochemical Reflectance Index and Light Use Efficiency in Broadleaf Forest

摘要

In current net primary production (NPP) modelsbased on light use efficiency (LUE), the realized LUE is oftenestimated by considering changes of environment conditions suchas temperature and precipitation to compute stress relatedrestrictions to the potential LUE. The accuracy of this approachhowever, is limited by a lack of detailed knowledge of stressors atlandscape and global scales. As an alternative to determine therealized LUE from environmental stresses, the photochemicalreflectance index (PRI) introduced by Gamon shows strongcorrelation with LUE at leaf, canopy and region levels. In thispaper, the relationships between ecosystem-level LUE obtainedfrom an eddy covariance flux tower and MODIS-derivedphotochemical reflectance index (PRI) were investigated inMorgan-Monroe State Forest. We tried to detect seasonalchanges of their relationships throughout the growing seasonfrom June to October in 2004. Only the backscatter reflectancedata which was demonstrated to have strong correlation withPRI in cloud-free days were picked for our research. Allstatistical analysis was conducted in MATLAB (version 7.1)software to establish linear regression model of the PRI-LUErelationship. The monthly correlation showed notable variationsin the four months except July in which there was not enoughvalid data for cloud coverage. In June the correlation wasnegative. For the following two months the values of LUE andPRI became positively correlated and the slope of the regressionline increased obviously. Then in late October the slope of thepositive relation decreased to below that in August. Potentialenvironmental factors were analyzed to explain the seasonalvariations of PRI-LUE relationships. To validate the effectivenessof the line regression model, ground micrometeorological data ofthe same tower site in 2005 were used to calculate groundmeasured LUE. We compared the predicted LUE from MODISdata using PRI-LUE relationship in August of 2004 with groundmeasured LUE and found the line model successfully predictedthe true realized LUE.