Earth Observation (EO) data are widely used in agricultural statistics production. However, the accuracy of EO-based land use classification is limited because of the limitations of using in situ census or survey data as training sets for EO applications. In this work, we provide recommendations for National Statistical Offices (NSO) to design in situ data collection campaigns that benefit both conventional statistics and EO-based assessments. Our recommendations are supported by a case study done by Chile's NSO (Instituto Nacional de Estadistica). The case study describes two methods for quality control of training samples used in EO image classification for agricultural statistics. The study describes methods for improving training samples for EO-based machine learning classification, which combines self-organised maps (SOM) with the sample balancing method SMOTE. Using the proposed method, the accuracy of the classification model increased from 95.2% to 98.5%, while the F1-scores for most classes also saw substantial improvements. The results suggest that combining multisensor EO data and advanced machine learning should be the focus of future research to enhance complex agricultural landscape classification.
PicoliMCamaraGSanchesI, et al.Big earth observation time series analysis for monitoring Brazilian agriculture. ISPRS J Photogramm Remote Sens2018; 145: 328–339.
2.
DefournyPBontempsSBellemansN, et al.Near real-time agriculture monitoring at national scale at parcel resolution: performance assessment of the Sen2-Agri automated system in various cropping systems around the world. Remote Sens Environ2019; 221: 551–568.
3.
WhitcraftAKBecker-ReshefIJusticeCO. A framework for defining spatially explicit earth observation requirements for a global agricultural monitoring initiative (GEOGLAM). Remote Sens (Basel)2015; 7: 1461–1481.
4.
MaxwellAEWarnerTAFangF. Implementation of machine-learning classification in remote sensing: an applied review. Int J Remote Sens2018; 39: 2784–2817.
5.
WaldnerFDe AbelleyraDVerónSR, et al.Towards a set of agrosystem-specific cropland mapping methods to address the global cropland diversity. Int J Remote Sens2016; 37: 3196–3231.
6.
FrenayBVerleysenM. Classification in the presence of label noise: a survey. IEEE Trans Neural Netw Learn Syst2014; 25: 845–869.
7.
HastieTTibshiraniR. Generalized additive models. Stat Sci1986; 1: 297–310.
8.
BontempsSAriasMCaraC, et al.Building a data set over 12 globally distributed sites to support the development of agriculture monitoring applications with Sentinel-2. Remote Sens (Basel)2015; 7: 16062–16090.
SantosLAFerreiraKRCamaraG, et al.Quality control and class noise reduction of satellite image time series. ISPRS J Photogramm Remote Sens2021; 177: 75–88.
DelincéJLemoineGDefournyP, et al.Handbook on remote sensing for agricultural statistics. Rome: FAO, 2017.
13.
ZhaoYFengDYuL, et al.Detailed dynamic land cover mapping of Chile: accuracy improvement by integrating multi-temporal data. Remote Sens Environ2016; 183: 170–185.
14.
GalleguillosMCeballos-ComissoAGimenoF, et al.CLDynamicLandCover. 2024.
AaldersIHAitkenheadMJ. Agricultural census data and land use modelling. Comput Environ Urban Syst2006; 30: 799–814.
18.
KerrJTCihlarJ. Land use and cover with intensity of agriculture for Canada from satellite and census data. Glob Ecol Biogeogr2003; 12: 161–172.
19.
HastingsFFuentesIPerez-BidegainM, et al.Land-cover mapping of agricultural areas using machine learning in google earth engine. In: GervasiOMurganteBMisraS, et al. (eds) Computational science and its applications – ICCSA 2020. Cham: Springer International Publishing, pp.721–736. ISBN 978-3-030-58811-3.
20.
HueteADidanKMiuraT, et al.Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sens Environ2002; 83: 195–213.
21.
Azócar de la CruzGAlfaroGAlonsoC, et al.Modeling the ignition risk: analysis before and after megafire on Maule region, Chile. Appl Sci2022; 12: 9353.
22.
AdamEMutangaORugegeD. Multispectral and hyperspectral remote sensing for identificacion and mapping wetland vegetation: a review. Wetl Ecol Manag2010; 18: 281–296.
23.
DelegidoJVerrelstJMezaC, et al.A red-edge spectral index for remote sensing estimation of green LAI over agroecosystems. Eur J Agron2013; 46: 42–52.
24.
SchulzJJCayuelaLRey-BenayasJM, et al.Factors influencing vegetation cover change in Mediterranean Central Chile (1975–2008). Appl Veg Sci2011; 14: 571–582.
25.
EslingPAgonC. Time-series data mining. ACM Comput Surv2012; 45: 1–34.
26.
MausVCamaraGCartaxoR, et al.A time-weighted dynamic time warping method for land-use and land-cover mapping. IEEE J Sel Top Appl Earth Obs Remote Sens2016; 9: 3729–3739.
27.
De SimoneLGennariP. Earth observations for official crop statistics in the context of scarcity of in-situ data. Stat J IAOS2022; 38: 1009–1019.
28.
JohnsonJMKhoshgoftaarTM. Survey on deep learning with class imbalance. J Big Data2019; 6: 27.
29.
HeHBaiYGarciaEA, et al.ADASYN: adaptive synthetic sampling approach for imbalanced learning. In 2008 IEEE international joint conference on neural networks (IEEE world congress on computational intelligence), pp.1322–1328.
30.
VesantoJAlhoniemiE. Clustering of the self-organizing map. IEEE Trans Neural Netw2000; 11: 586–600.
31.
PelletierCValeroSIngladaJ, et al.Assessing the robustness of random forests to map land cover with high resolution satellite image time series over large areas. Remote Sens Environ2016; 187: 156–168.
32.
PelletierCWebbGIPetitjeanF. Temporal convolutional neural network for the classification of satellite image time series. Remote Sens (Basel)2019; 11: 523.
33.
GarnotVLandrieuLGiordanoS, et al.Satellite image time series classification with pixel-set encoders and temporal self-attention. In: 2020 IEEE/CVF conference on computer vision and pattern recognition (CVPR), Seattle, WA, USA, pp.12322–12331. IEEE. ISBN 978-1-72817-168-5.
34.
SimoesRCamaraGQueirozG, et al.Satellite image time series analysis for big earth observation data. Remote Sens (Basel)2021; 13: 2428.
