igarss21

Find our colleagues Vassilis Sito and Stela Girtsou in the session TU2.MM-25: Machine Learning Methods in Hazard Assessment on Tuesday 13th of July (13:00 - 14:10 CEST)!

Stela Girtsou and Vassilis Sito will present the following papers:

TU2.MM-25.7 A MACHINE LEARNING METHODOLOGY FOR NEXT DAY WILDFIRE PREDICTION

(13:30-13:35 CEST)

TU2.MM-25.8 SEMI-SUPERVISED PHENOLOGY ESTIMATION IN COTTON PARCELS WITH SENTINEL-2 TIME-SERIES

(13:35-13:40 CEST)

 

Abstract

Vassilis Sitokonstantinou

SEMI-SUPERVISED PHENOLOGY ESTIMATION IN COTTON PARCELS WITHSENTINEL-2 TIME-SERIES; TU2.MM – 25.8 Machine Learning Methods in Hazard Assessment; Tue, 13 Jul, 11:35-11:40 (UTC)

https://we.tl/t-Yvbe4MHb5d

This study presents a dynamic phenology estimation methodology  for  cotton  towards  early  warning  and  mitigation  advice  against  natural  disasters.   First,  a  time-series  comparison algorithm that is based on Earth Observation (EO) data is used to assign pseudo-labels to approximately 1,000 parcels. The knowledge is extracted from only limited ground truth information.   The pseudo-labels are then used to train Random Forest (RF) regression models for phenology estimation. The  pseudo-labeling  process  is  used  to  augment  the  annotated dataset and allow for modelling the growth of cotton. The models are applied and evaluated on two different test sites in Greece;  for which field campaigns were carried out to collect the labels.  The results are satisfactory and show-case the successful generalization of the models to other areas.  The dynamic predictions for cotton growth and extreme weather  events,  from  numerical  weather  prediction  (NWP)models, are invaluable information for decision-making relevant to agricultural insurance schemes and farm management.

Index Terms—cotton phenology, agricultural insurance, semi-supervised learning, early warning, natural disasters