|Rizzi, R - NAT. INST. SPACE RSCH.|
|Rudorff, B - NAT. INST. SPACE RSCH.|
|Shimabukuro, Y - NAT. INST. SPACE RSCH.|
Submitted to: International Journal of Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: December 20, 2006
Publication Date: December 20, 2006
Citation: Rizzi, R., Rudorff, B.F.T., Shimabukuro, Y.E., Doraiswamy, P.C. 2006. Assessment of MODIS LAI retrievals over soybean crop in Southern Brazil. International Journal of Remote Sensing. 27(19):4091-4100. Interpretive Summary: Data acquired onboard the Terra satellite by the MODIS sensor is processed by the MODIS team and products are developed and distributed to the user community for various applications. This paper evaluates the magnitude of the Collection 4 MODIS leaf area index (LAI) product for the soybean crop in southern Brazil. The LAI data for two crop seasons, 2001/2002 to 2004/2005 was acquired for this research. The data indicated that there was a large proportion of the soybean crop LAI rapidly reaching a maximum level suggesting that the LAI product was saturated not usable. Results indicated that the Collection 4 LAI product saturated for the soybean areas during peak growing season and occurred in every crop year we studied. The MODIS Land Cover Type 3 product was also analyzed for classification accuracy by comparing with the occurrence of the soybean LAI retrievals. The conclusion was that the MODIS LAI algorithm used in Collection 4 was not properly calibrated for monitoring seasonal changes over soybean area in southern Brazil.
Technical Abstract: This paper examines the magnitude of the Collection 4 MODIS leaf area index (LAI) product for the soybean crop for two crop seasons, 2001/2002 to 2004/2005, in Southern Brazil. The data indicated that there were a large proportion of digical counts were at a maximum level suggesting that the LAI was saturated and that no leaf area index (LAI) retrieval was possible for the considered pixel. Time series of eight-day composite LAI images (MOD15A2) and 16-day composite NDVI images (MOD13Q1) were both compared with a soybean reference map derived from multitemporal Landsat images. The Land Cover Type 3 product (MOD12Q1) was also analyzed to verify if the occurrence of those retrievals was related to misclassification of the broadleaf crops biome. Results indicated that the 255 standard fill value retrievals in Collection 4 LAI product were mainly related to soybean areas during peak growing season and occurred in every crop year we have studied. Eventual misclassification in the biome map was not the cause of those retrievals in the Collection 4 MODIS LAI product and concluded that the LAI algorithm was not properly calibrated for monitoring seasonal changes over soybean area in southern Brazil.