Avocado is a high value crop ranking fourth among the planted fruit species in Chile with more than 32,000 ha.

Yield estimation is an important challenge in avocado due to its phenology, the size of the tree, and to the large variability usually observed within the orchards.

Due to the practical difficulties to sample the trees we use the following approach: 1) establish a systematic, non-aligned grid with > 20 sampling points (trees)/field, 2) previous to harvest, and once the fruit has reached the commercial size, all the fruits are removed from the tree, sorted and weighted, 3) maps are produced using kriging interpolation, 4) yield estimation is performed by field, 4) cross validation is performed by comparing real versus estimated yield. Extracted fruit weight are added to the total weight to calculate the “true” yield.

Results showed that the approach produced very close estimations of the actual yields with errors < 5%. The maps made it possible to identify the areas of lower yield,  manily at the border of the fields, caused in part “non agronomic” losses. With this information it is possible to focus management strategies to try to reduce fruit losses. Using high resolution satellite imagery to establish homogeneous vigor zones, “non agronomic” fruit losses were estimated to be as high as 30%.