Mapping spatial variability of foliar nitrogen in coffee (Coffea arabica L.) plantations with multispectral Sentinel-2 MSI data

Abstract Nitrogen (N) is the most limiting factor to coffee development and productivity. Therefore, development of rapid, spatially explicit and temporal remote sensing-based approaches to determine spatial variability of coffee foliar N are imperative for increasing yields, reducing production costs and mitigating environmental impacts associated with excessive N applications. This study sought to assess the value of Sentinel-2 MSI spectral bands and vegetation indices in empirical estimation of coffee foliar N content at landscape level. Results showed that coffee foliar N is related to Sentinel-2 MSI B4 (R2 = 0.32), B6 (R2 = 0.49), B7 (R2 = 0.42), B8 (R2 = 0.57) and B12 (R2 = 0.24) bands. Vegetation indices were more related to coffee foliar N as shown by the Inverted Red-Edge Chlorophyll Index – IRECI (R2 = 0.66), Relative Normalized Difference Index – RNDVI (R2 = 0.48), CIRE1 (R2 = 0.28), and Normalized Difference Infrared Index – NDII (R2 = 0.37). These variables were also identified by the random forest variable optimisation as the most valuable in coffee foliar N prediction. Modelling coffee foliar N using vegetation indices produced better accuracy (R2 = 0.71 with RMSE = 0.27 for all and R2 = 0.73 with RMSE = 0.25 for optimized variables), compared to using spectral bands (R2 = 0.57 with RMSE = 0.32 for all and R2 = 0.58 with RMSE = 0.32 for optimized variables). Combining optimized bands and vegetation indices produced the best results in coffee foliar N modelling (R2 = 0.78, RMSE = 0.23). All the three best performing models (all vegetation indices, optimized vegetation indices and combining optimal bands and optimal vegetation indices) established that 15.2 ha (4.7%) of the total area under investigation had low foliar N levels (

[1]  S. Frolking,et al.  Canopy nitrogen, carbon assimilation, and albedo in temperate and boreal forests: Functional relations and potential climate feedbacks , 2008, Proceedings of the National Academy of Sciences.

[2]  M. Cho,et al.  Non-linear partial least square regression increases the estimation accuracy of grass nitrogen and phosphorus using in situ hyperspectral and environmental data , 2013 .

[3]  P. Curran.  Remote sensing of foliar chemistry , 1989 .

[4]  Gary R. Watmough,et al.  Evaluating the capabilities of Sentinel-2 for quantitative estimation of biophysical variables in vegetation , 2013 .

[5]  Kelly J. Watson,et al.  Context and contingency: the coffee crisis for conventional small‐scale coffee farmers in Brazil , 2008 .

[6]  F. Baret,et al.  Quantification of plant stress using remote sensing observations and crop models: the case of nitrogen management. , 2006, Journal of experimental botany.

[7]  Ademir de Oliveira Ferreira,et al.  Nitrogen fertilizer (15N) leaching in a central pivot fertigated coffee crop , 2012 .

[8]  Onisimo Mutanga,et al.  Capability of models to predict leaf N and P across four seasons for six sub-tropical forest evergreen trees , 2015 .

[9]  Andrew K. Skidmore,et al.  Potential of Sentinel-2 spectral configuration to assess rangeland quality , 2015 .

[10]  René Coste,et al.  Coffee: The Plant and the Product , 1992 .

[11]  J. Dash,et al.  The MERIS terrestrial chlorophyll index , 2004 .

[12]  Ian J. Yule,et al.  Mapping of macro and micro nutrients of mixed pastures using airborne AisaFENIX hyperspectral imagery , 2016 .

[13]  A. Ramoelo,et al.  Water-removed spectra increase the retrieval accuracy when estimating savanna grass nitrogen and phosphorus concentrations , 2011 .

[14]  A. Chemura.  The growth response of coffee (Coffea arabica L) plants to organic manure, inorganic fertilizers and integrated soil fertility management under different irrigation water supply levels , 2014, International Journal of Recycling of Organic Waste in Agriculture.

[15]  Steve Frolking,et al.  Nitrogen cycling, forest canopy reflectance, and emergent properties of ecosystems , 2013, Proceedings of the National Academy of Sciences.

[16]  P. Mecocci,et al.  Random Forest ensembles for detection and prediction of Alzheimer's disease with a good between-cohort robustness , 2014, NeuroImage: Clinical.

[17]  Philip A. Townsend,et al.  Disentangling the contribution of biological and physical properties of leaves and canopies in imaging spectroscopy data , 2013, Proceedings of the National Academy of Sciences.

[18]  Margaret Kalacska,et al.  Estimation of foliar chlorophyll and nitrogen content in an ombrotrophic bog from hyperspectral data: Scaling from leaf to image , 2015 .

[19]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[20]  Onisimo Mutanga,et al.  Empirical Modeling of Leaf Chlorophyll Content in Coffee (Coffea Arabica) Plantations With Sentinel-2 MSI Data: Effects of Spectral Settings, Spatial Resolution, and Crop Canopy Cover , 2017, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[21]  Onisimo Mutanga,et al.  Developing detailed age-specific thematic maps for coffee (Coffea arabica L.) in heterogeneous agricultural landscapes using random forests applied on Landsat 8 multispectral sensor , 2017 .

[22]  O. Mutanga,et al.  Remote sensing leaf water stress in coffee (Coffea arabica) using secondary effects of water absorption and random forests , 2017 .

[23]  Onisimo Mutanga,et al.  Exploiting machine learning algorithms for tree species classification in a semiarid woodland using RapidEye image , 2013 .

[24]  Andrew K. Skidmore,et al.  Regional estimation of savanna grass nitrogen using the red-edge band of the spaceborne RapidEye sensor , 2012, Int. J. Appl. Earth Obs. Geoinformation.

[25]  F. Damatta,et al.  Photosynthesis and photoprotection in coffee leaves is affected by nitrogen and light availabilities in winter conditions. , 2010, Journal of plant physiology.

[26]  Jan G. P. W. Clevers,et al.  Using Hyperspectral Remote Sensing Data for Retrieving Canopy Chlorophyll and Nitrogen Content , 2012, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[27]  Subashisa Dutta,et al.  Spatial variability of chlorophyll and nitrogen content of rice from hyperspectral imagery , 2016 .

[28]  O. Mutanga,et al.  Mapping leaf nitrogen and carbon concentrations of intact and fragmented indigenous forest ecosystems using empirical modeling techniques and WorldView-2 data , 2017 .

[29]  Mario Chica-Olmo,et al.  An assessment of the effectiveness of a random forest classifier for land-cover classification , 2012 .

[30]  Onisimo Mutanga,et al.  Separability of coffee leaf rust infection levels with machine learning methods at Sentinel-2 MSI spectral resolutions , 2017, Precision Agriculture.

[31]  S. Ollinger,et al.  Examining spectral reflectance features related to foliar nitrogen in forests: Implications for broad-scale nitrogen mapping , 2016 .

[32]  Philip Lewis,et al.  Hyperspectral remote sensing of foliar nitrogen content , 2012, Proceedings of the National Academy of Sciences.

[33]  K. Tully,et al.  More trees less loss: Nitrogen leaching losses decrease with increasing biomass in coffee agroforests , 2012 .

[34]  O. Mutanga,et al.  Examining the potential of Sentinel-2 MSI spectral resolution in quantifying above ground biomass across different fertilizer treatments , 2015 .

[35]  Marco Heurich,et al.  Vegetation Indices for Mapping Canopy Foliar Nitrogen in a Mixed Temperate Forest , 2016, Remote. Sens..

[36]  Anatoly A. Gitelson,et al.  Remote estimation of nitrogen and chlorophyll contents in maize at leaf and canopy levels , 2013, Int. J. Appl. Earth Obs. Geoinformation.

[37]  A. Viña,et al.  Remote estimation of canopy chlorophyll content in crops , 2005 .

[38]  D. Kutywayo,et al.  Bioclimatic modelling of current and projected climatic suitability of coffee (Coffea arabica) production in Zimbabwe , 2016, Regional Environmental Change.

[39]  G. Daily,et al.  The Nature and Value of Ecosystem Services: An Overview Highlighting Hydrologic Services , 2007 .

[40]  Jan G. P. W. Clevers,et al.  Remote estimation of crop and grass chlorophyll and nitrogen content using red-edge bands on Sentinel-2 and -3 , 2013, Int. J. Appl. Earth Obs. Geoinformation.

[41]  B. Jong,et al.  Carbon sequestration through agroforestry in indigenous communities of Chiapas, Mexico , 2009, Agroforestry Systems.