Comparison between historical yield and soybean crop EVI values using correlation map
Abstract
Timely and accurate yield estimates using remote sensing represents an important advance towards objective crop forecasting in Brazil. Vegetation index values integrated over a period have been used to generate agronomic parameters such as crop yield. Several studies showed the strong relationship between accumulated vegetation index and historical yield, once it represents crop photosynthetic activity. The main goal of this study was to create correlation maps between Modis/TERRA EVI and historical yield during the soybean crop cycle in Paraná state, Brazil, from 2000 to 2010. The soybean cycle was separated in four variables corresponding to the crop stage: emergence to maturity, emergence to flowering, flowering to maturity, flowering to the grain filling. For each variable a correlation map was created between the accumulated EVI and soybean yield at pixel level. All variables showed a good correlation, but among all of them the best correlation was the period between flowering to maturity. This happened because of exclusion of months where EVI response was low, corresponding to period of crop emergence (October and November). A percentage map of soybean crop was confronted with the correlation map to check out whether the highest correlation was corresponding to soybean pixel. On the percentage map pixels showing above 70% of soybean were selected, then on the correlation map pixels with correlation coefficients above 0.7 were selected. Within the data set 43% of pixels from the correlation map had land cover greater than or equal to 70% of soybean crop.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2012
- Bibcode:
- 2012AGUFM.B43E0450F
- Keywords:
-
- 0402 BIOGEOSCIENCES / Agricultural systems;
- 1640 GLOBAL CHANGE / Remote sensing