Evaluation and Exploitation of Retrieval Algorithms for Estimating Biophysical Crop Variables Using Sentinel-2, Venus, and PRISMA Satellite Data

doi:10.11947/j.JGGS.2020.0408

Abstract

Abstract:

This paper is devoted to the development and testing of the optimal procedures for retrieving biophysical crop variables by exploiting the spectral information of current multispectral optical satellite Sentinel-2 and Venus and in view of the advent of the new Sino-EU hyperspectral satellite (e.g., PRISMA, EnMAP, and GF-5). Two different methodologies devoted to the estimation of biophysical crop variables Leaf area index (LAI) and Leaf chlorophyll content (Cab) were evaluated: non-kernel-based and kernel-based Machine Learning Regression Algorithms (MLRA); Sentinel-2 and Venus data comparison for the analysis of the durum wheat-growing season. Results show that for Sentinel-2 data, Gaussian Process Regression (GPR) was the best performing algorithm for both LAI (R²=0.89 and RMSE=0.59) and Cab (R²=0.70 and RMSE=8.31). Whereas, for PRISMA simulated data the Kernel Ridge Regression (KRR) was the best performing algorithm among all the other MLRA (R²=0.91 and RMSE=0.51) for LAI and (R²=0.83 and RMSE=6.09) for Cab, respectively. Results of Sentinel-2 and Venus data for durum wheat-growing season were consistent with ground truth data and confirm also that SWIR bands, which are used as tie-points in the PROSAIL inversion, are extremely useful for an accurate retrieving of crop biophysical parameters.

Key words: biophysical crop parameters; PRISMA; GF-5; Sentinel 2; Venus

Raffaele CASA,Deepak UPRETI,Angelo PALOMBO,Simone PASCUCCI,Hao YANG,Guijun YANG,Wenjiang HUANG,Stefano PIGNATTI. Evaluation and Exploitation of Retrieval Algorithms for Estimating Biophysical Crop Variables Using Sentinel-2, Venus, and PRISMA Satellite Data[J]. Journal of Geodesy and Geoinformation Science, 2020, 3(4): 79-88.

Figures/Tables 7

Tab.1

Tab.2

Tab.3

Tab.4

Tab.5

Fig.1

Fig.2

References 40

[1]	GOFFART J P, OLIVIER M, FRANKINET M. Potato crop nitrogen status assessment to improve N fertilization management and efficiency: past-present-future[J]. Potato Research, 2008,51(3-4):355-383.
[2]	ZHANG C, KOVACS J M. The application of small unmanned aerial systems for precision agriculture: a review[J]. Precision Agriculture, 2012,13(6):693-712.
[3]	CILIA C, PANIGADA C, ROSSINI M, et al. Nitrogen status assessment for variable rate fertilization in maize through hyperspectral imagery[J]. Remote Sensing, 2014,6(7):6549-6565.
[4]	JAY S, MAUPAS F, BENDOULA R, et al. Retrieving LAI, chlorophyll and nitrogen contents in sugar beet crops from multi-angular optical remote sensing: comparison of vegetation indices and PROSAIL inversion for field phenotyping[J]. Field Crops Research, 2017,210:33-46.
[5]	CASA R, UPRETI D, PELOSI F. Measurement and estimation of leaf area index (LAI) using commercial instruments and smartphone-based systems[C]//IOP Conference Series: Earth and Environmental Science. IOP Publishing, 2019,275(1):012006.
[6]	ROUSE JR J W. Monitoring the vernal advancement and retrogradation (green wave effect) of natural vegetation[J]. 1973.
[7]	RONDEAUX G, STEVEN M, BARET F. Optimization of soil-adjusted vegetation indices[J]. Remote Sensing of Environment, 1996,55(2):95-107.
[8]	PASOLLI E, MELGANI F, ALAJLAN N, et al. Active learning methods for biophysical parameter estimation[J]. IEEE Transactions on Geoscience and Remote Sensing, 2012,50(10):4071-4084.
[9]	VERRELST J, RIVERA J P, VEROUSTRAETE F, et al. Experimental Sentinel-2 LAI estimation using parametric, non-parametric and physical retrieval methods-A comparison[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2015,108:260-272.
[10]	DURBHA S S, KING R L, YOUNAN N H. Support vector machines regression for retrieval of leaf area index from multiangle imaging spectroradiometer[J]. Remote Sensing of Environment, 2007,107(1-2):348-361.
[11]	LÁZARO-GREDILLA M, TITSIAS M K, VERRELST J, et al. Retrieval of biophysical parameters with heteroscedastic Gaussian processes[J]. IEEE Geoscience and Remote Sensing Letters, 2013,11(4):838-842.
[12]	CHAI L, QU Y, ZHANG L, et al. Lai retrieval from cyclopes and modis products using artificial neural networks[C]//IGARSS 2008-2008 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2008, 3: III-1064-III-1067.
[13]	WEISS M, BARET F. S2ToolBox Level 2 products: LAI, FAPAR, FCOVER, Version 1.1[J]. ESA Contract Nr. 4000110612/14/1-BG, 2016: 52.
[14]	VERRELST J, CAMPS-VALLS G, MUÑOZ-MARÍ J, et al. Optical remote sensing and the retrieval of terrestrial vegetation bio-geophysical properties-A review[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2015,108:273-290.
[15]	WILLIAMS C K I, RASMUSSEN C E. Gaussian processes for machine learning[M]. Cambridge, MA: MIT Press, 2006.
[16]	VERRELST J, DETHIER S, RIVERA J P, et al. Active learning methods for efficient hybrid biophysical variable retrieval[J]. IEEE Geoscience and Remote Sensing Letters, 2016,13(7):1012-1016.
[17]	VERRELST J, MALENOVSKY Z, VAN DER TOL C, et al. Quantifying vegetation biophysical variables from imaging spectroscopy data: a review on retrieval methods[J]. Surveys in Geophysics, 2019,40(3):589-629. doi: 10.1007/s10712-018-9478-y
[18]	PIGNATTI S, PALOMBO A, PASCUCCI S, et al. The PRISMA hyperspectral mission: Science activities and opportunities for agriculture and land monitoring [C]//2013 IEEE International Geoscience and Remote Sensing Symposium-IGARSS. IEEE, 2013: 4558-4561.
[19]	GUANTER L, KAUFMANN H, SEGL K, et al. The EnMAPspaceborne imaging spectroscopy mission for earth observation[J]. Remote Sensing, 2015,7(7):8830-8857. doi: 10.3390/rs70708830
[20]	JACQUEMOUD S, VERHOEF W, BARET F, et al. PROSPECT+ SAIL models: A review of use for vegetation characterization[J]. Remote Sensing of Environment, 2009,113:S56-S66.
[21]	BACOUR C, JACQUEMOUD S, TOURBIER Y, et al. Design and analysis of numerical experiments to compare four canopy reflectance models[J]. Remote Sensing of Environment, 2002,79(1):72-83.
[22]	BARET F, LEROY M, ROUJEAN J L, et al. CYCLOPES User Requirement Document[J]. INRA-CSE, Avignon, 2003.
[23]	HAGAN M T, MENHAJ M B. Training feedforward networks with the Marquardt algorithm[J]. IEEE transactions on Neural Networks, 1994,5(6):989-993. pmid: 18267874
[24]	BREIMAN L. Bagging predictors[J]. Machine Learning, 1996,24(2):123-140.
[25]	WOLD S, ESBENSEN K, GELADI P. Principal component analysis[J]. Chemometrics and intelligent laboratory systems, 1987,2(1-3):37-52.
[26]	BREIMAN L. Random forests[J]. Machine Learning, 2001,45(1):5-32.
[27]	DRAPER N R, SMITH H. Applied regression analysis[M]. [S.l.]: John Wiley & Sons, 1998.
[28]	GELADI P, KOWALSKI B R. Partial least-squares regression: a tutorial[J]. Analyticachimicaacta, 1986,185:1-17.
[29]	FRIEDMAN J, HASTIE T, TIBSHIRANI R. Additive logistic regression: a statistical view of boosting (with discussion and a rejoinder by the authors)[J]. The Annals of Statistics, 2000,28(2):337-407.
[30]	BREIMAN L, FRIEDMAN J, STONE C J, et al. Classification and regression trees[M]. CRC Press, 1984.
[31]	SAARELA M, ELOMAA T, RUOHONEN K. An analysis of relevance vector machine regression[M] //Advances in Machine Learning I. Heidelberg: Springer, 2010: 227-246.
[32]	VAPNIK V, GOLOWICH S E, SMOLA A J. Support vector method for function approximation, regression estimation and signal processing [C]//Advances in neural information processing systems. 1997: 281-287.
[33]	SUYKENS J A K, VANDEWALLE J. Least squares support vector machine classifiers[J]. Neural Processing Letters, 1999,9(3):293-300.
[34]	HUANG G, HUANG G B, SONG S, et al. Trends in extreme learning machines: a review[J]. Neural Networks, 2015,61:32-48. pmid: 25462632
[35]	VERRELST J, RIVERA J P, ALONSO L, et al. ARTMO: an automated radiative transfer models operator toolbox for automated retrieval of biophysical parameters through model inversion [C]//Proceedings of EARSeL 7th SIG-Imag. Spectrosc. Workshop. 2011: 11-13.
[36]	UPRETI D, HUANG W, KONG W, et al. A comparison of hybrid machine learning algorithms for the retrieval of wheat biophysical variables from sentinel-2[J]. Remote Sensing, 2019,11(5):481.
[37]	DOUAK F, MELGANI F, BENOUDJIT N. Kernel ridge regression with active learning for wind speed prediction[J]. Applied Energy, 2013,103:328-340.
[38]	VERRELSTJ, MUÑOZ J, ALONSO L, et al. Machine learning regression algorithms for biophysical parameter retrieval: Opportunities for Sentinel-2 and-3[J]. Remote Sensing of Environment, 2012,118:127-139.
[39]	VERRELST J, ALONSO L, CAMPS-VALLS G, et al. Retrieval of vegetation biophysical parameters using Gaussian process techniques[J]. IEEE Transactions on Geoence & Remote Sensing, 2012,50(5):1832-1843.
[40]	RIVERA-CAICEDO J P, VERRELST J, MUÑOZ-MARÍ J, et al. Hyperspectral dimensionality reduction for biophysical variable statistical retrieval[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2017,132:88-101.

Imager	Spectral bands	Spectral range	FWHM	SNR
PRISMA	247	400÷2500	7÷11	600@650 >400@1550 >200@2100
EnMAP	242	420÷2450	10	>500@495 >150@2200
AHSI-GF5	—	388÷2518	4.5÷8.6	702@600 506@1300 251@2200

Variable		Minimum	Maximum	Mode	Std	Nb_Class	Law
Canopy	LAI	0.0	15.0	2.0	2.0	6	log_normal
	ALA/(°)	30	80	60	20	3	gauss
	Crown Cover	1.0	1.0	0.8	0.4	1	uni
	HsD	0.1	0.5	0.2	0.5	1	gauss
Leaf	N	1.20	1.80	1.50	0.30	3	gauss
	Cab (μg·cm^-2)	20	90	45	30	4	gauss
	Cdm (g·m^-2)	0.0030	0.0110	0.0050	0.0050	4	gauss
	Cw_Rel	0.60	0.85	0.75	0.08	4	uni
	Cbp	0.00	2.00	0.00	0.30	3	gauss
Soil	Bs	0.50	3.50	1.20	2.00	4	gauss

Sentinel-2		Venus
2018/01/29	2018/04/21	2018/01/28
2018/02/10	2018/05/19	2018/04/20
2018/02/13	2018/05/26	2018/04/26
2018/04/06	2018/05/31	2018/05/20

	Sentinel-2	Venus
Spatial resolution	MS: 10, 20, 60m	MS: 5.3m
Spectral resolution (Center W.-Band- width)	B1: 0.44~0.20μm B2: 0.49~0.65μm B3: 0.56~0.35μm B4: 0.66~0.30μm B5: 0.71~0.15μm B6: 0.74~0.15μm B7: 0.78~0.20μm B8: 0.84~1.15μm B8a: 0.86~0.20μm B9: 0.95~0.20μm B10: 1.38~0.30μm B11: 1.61~0.90μm B12: 2.2~1.80μm	B1: 0.42~0.40μm B2: 0.44~0.40μm B3: 0.49~0.40μm B4: 0.55~0.40μm B5: 0.62~0.40μm B6: 0.62~0.40μm B7: 0.67~0.30μm B8: 0.70~0.24μm B9: 0.74~0.16μm B10: 0.78~0.16μm B11: 0.86~0.40μm B12: 0.91~0.20μm
Swath width	69km with 2 cameras	27.6km
Revisit capability	5 days	2 days

MLRA	SENTINEL-2 (8 Bands)				PRISMA (213 Bands)
	LAI		Cab		LAI		Cab
	R²(SD)	RMSE(SD)	R²(SD)	RMSE(SD)	R²(SD)	RMSE(SD)	R²(SD)	RMSE(SD)
GPR	0.89 (0.06)	0.59 (0.36)	0.70 (0.13)	8.31 (2.05)	0.89 (0.08)	0.53 (0.23)	0.79 (0.10)	6.82 (2.00)
VHGPR	0.86 (0.11)	0.70 (0.43)	0.71 (0.12)	8.31 (2.07)	0.89 (0.05)	0.55 (0.20)	0.75 (0.15)	7.08 (1.62)
RVM	0.87 (0.08)	0.63 (0.27)	0.66 (0.10)	9.21 (1.95)	0.88 (0.07)	0.60 (0.27)	0.75 (0.12)	7.57 (2.58)
SVR	0.88 (0.07)	0.65 (0.37)	0.65 (0.08)	9.02 (2.02)	0.86 (0.08)	0.70 (0.37)	0.78 (0.10)	8.60 (2.85)
KRR	0.89 (0.09)	0.58 (0.21)	0.68 (0.12)	8.51 (1.74)	0.91 (0.05)	0.51 (0.29)	0.83 (0.11)	6.09 (2.25)
ELM	0.86 (0.07)	0.69 (0.27)	0.64 (0.12)	8.85 (3.47)	0.81 (0.08)	0.81 (0.31)	0.61 (0.11)	9.16 (1.60)
NN	0.71 (0.04)	1.00 (0.15)	0.51 (0.17)	10.15 (3.58)	0.81 (0.03)	0.82 (0.12)	0.61 (0.04)	9.61 (0.54)
BagT	0.74 (0.05)	0.96 (0.13)	0.44 (0.18)	11.01 (3.90)	0.80 (0.03)	0.84 (0.11)	0.45 (0.05)	11.54 (0.62)
PCR	0.67 (0.07)	1.06 (0.22)	0.48 (0.17)	10.42 (3.70)	0.78 (0.03)	0.89 (0.10)	0.59 (0.02)	9.85 (0.46)
RFTB	0.72 (0.06)	0.98 (0.13)	0.45 (0.19)	10.79 (3.81)	0.79 (0.03)	0.86 (0.11)	0.45 (0.06)	11.46 (0.60)
LSLR	0.67 (0.07)	1.07 (0.20)	0.51 (0.17)	10.11 (3.60)	0.72 (0.04)	1.04 (0.11)	0.49 (0.02)	11.24 (0.64)
PLSR	0.67 (0.07)	1.07 (0.19)	0.51 (0.17)	10.11 (3.60)	0.72 (0.04)	1.04 (0.11)	0.48 (0.03)	11.39 (0.69)
BooT	0.63 (0.06)	1.13 (0.13)	0.42 (0.17)	11.75 (4.62)	0.62 (0.06)	1.24 (0.11)	2.87 (7.78)	12.28 (0.45)
RegT	0.57 (0.09)	1.28 (0.17)	0.27 (0.20)	13.92 (4.88)	0.60 (0.06)	1.26 (0.13)	0.16 (0.04)	16.34 (0.68)
RFFE	0.56 (0.09)	1.35 (0.21)	0.25 (0.20)	14.56 (5.47)	0.60 (0.09)	1.30 (0.14)	0.16 (0.05)	16.77 (0.84)