期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2006
卷号：XXXVI-8/W48
页码：129-134
出版社：Copernicus Publications
摘要：Some years ago the MARS-FOOD group was established to support the Food Aid and Food Security policies of the EuropeanCommission. The activities are aimed at improving methods and information on crop yield prospects. Russia, Central Asia, and non-European Mediterranean countries (MECA region), Eastern Africa (IGAD sub-region) and the MERCOSUR region in SouthAmerica were selected as pilot areas. Crop growth indicators are produced based on low resolution remote sensing data, globalmeteorological modelling outputs (ECMWF model) and crop growth simulation models (CGMS and FAO-WSI). Crop yieldforecasting is done using predictors selected from the crop growth indicators. Dekadal SPOT-VEGETATION data are used as abasis for calculation of remote sensing indicators of crop growth. The Normalized Difference Vegetation Index (NDVI) and resultsof Dry Matter Production modelling (DMP) applying the Monteith approach (Monteith, 1972) are used as a main source of remotesensing indicators for the MECA region. The indicators are used in aggregated for sub-national administrative unit form applyingcrop mask. Some indicators are derived for a network of representative points. The current dekadal indicators are compared withprevious year dekadal values or with long-term average dekadal data. Additionally relative time mosaics of indicators are used as atool for crop growth monitoring (Savin, Nègre, 2002). We analyze additionally seasonal cumulative values of indicators bycomparing seasonal time profiles. As a result, near 10 remote sensing indicators can be derived for each crop for each dekad ofgrowing season in aggregated form and the same amount for representative points. Crop yield forecasting starts from an attempt tobuild simple regression equation between statistical crop yield and crop growth indicators. We found that regression with high R2can be built for many administrative units of MECA region. During the second phase of crop yield prediction the similarity analysisis applied. The aim of analysis is to define a year-analogue for indicator time profiles. This operation is conducted mainly for theadministrative units where regression analysis does not give acceptable results. The last phase is devoted to comparison ofindicator’s value with previous year or long-term average value. Final yield prediction is made by expert taking into considerationthe results of all phases of indicators analysis. The crop yield can be predicted quantitatively based only on remote sensingindicators for many administrative units of the region. For some units only a sign of crop yield changes can be predicted. In somecases it is impossible to predict crop yield based only on remote sensing indicators. The time when crop yield prediction can bemade differs from region to region. For the most part of administrative units of the region the best time for crop yield prediction isallocated near crop flowering. However, for some units the best time is shifted to earlier or to later period of crop growing season.The results of the crop growth monitoring and yield prediction are summarized in the form of agro-meteorological bulletins, issuedbimonthly for Russia and Central Asia, and for the Mediterranean countries
