Preliminary study on the potential use of RPA images to quantify the influence of the defoliation after coffee harvesting to its yield

L.M. dos Santos¹*, G.A.S. Ferraz¹, M.A.F. Carvalho², M.S. Vilela³ and P.H.O. Estima¹

¹Federal University of Lavras, Department of Agricultural Engineering, University Campus, BR37.200-000 Lavras-MG, Brazil
²Embrapa Café, Brasília 70770-901, Distrito Federal, Brazil
³Federal University of Lavras, Department of Agronomy/ Phytotechnics, University Campus, BR37.200-000 Lavras-MG, Brazil
*Correspondence: luanna_mendess@yahoo.com.br

Abstract:

Coffee is an agricultural commodity with global commercial importance capable of impacting the production chain. The quantification of defoliation at harvest is important for monitoring crop yield because defoliation is one of the main types of damage caused by this agricultural operation in coffee crops. Thus, the objective of this study was to evaluate the relationship between yield and defoliation obtained in the field and obtained through remotely piloted aircraft (RPA) images. The experiment was conducted in a coffee plantation belonging to the Federal University of Lavras (UFLA), Lavras, Minas Gerais state, Brazil. An RPA with a rotary wing containing a multispectral camera was used in autonomous flight mode with a height of 30 m, an image overlap of 80%, and a speed of 3 m s^-1. The images were collected before and after the 2020 and 2021 harvest, defoliation data obtained in the field were measured in 2020 and 2021, and the yield was measured from 2019 to 2021. Image processing was performed in the software PhotoScan, postimage processing was performed in QGIS, and statistical analyses were performed using the software R. With the processing of the images in 2020, the crop showed reductions of 17.3% and 18.4% in leaf area and volume, respectively, after harvest. In 2021, the crop showed reductions of 12.8% and 9.8% in leaf area and volume, respectively, after harvest. The leaf area and leaf volume of the coffee plantation after harvest could be quantified by means of images obtained by RPA, which allowed the observation of the loss of area and volume of the coffee plantation. Furthermore, it was possible to analyse the interactions between field data and the yield of the same harvest year, which were directly proportional, and the interaction of image data from one year with the previous yield, which were inversely proportional. In the year 2020, there was a reduction of 17.3% in leaf area after harvest, and a reduction of 18.4% in leaf volume after harvest in the plots under study.In the processing carried out in 2021, there was a 12.8% reduction in leaf area after harvest, and a 9.8% decrease in leaf volume after harvest in the plots under study.

Key words:

canopy volume, Coffea arabica L., digital image processing, harvest, systems of unmanned aircraft

Coffee crop coefficient prediction as a function of biophysical variables identified from RGB UAS images

L.M. Santos¹*, G.A.S. Ferraz¹, A.V. Diotto², B.D.S. Barbosa¹, D.T. Maciel², M.T. Andrade¹, P.F.P. Ferraz¹ and G. Rossi³

¹Federal University of Lavras, Department of Agricultural Engineering, University Campus, BR37.200-000 Lavras-MG, Brazil
²Federal University of Lavras, Department of Water Resources and sanitation, University Campus, BR37.200-000 Lavras, Brazil
³University of Florence, Department of Agricultural, Food, Environment and Forestry (DAGRI), Via San Bonaventura, 13, IT50145 Florence, Italy
*Correspondence: luanna_mendess@yahoo.com.br

Abstract:

Because of different Brazilian climatic conditions and the different plant conditions, such as the stage of development and even the variety, wide variation may exist in the crop coefficients (𝐾𝑐) values, both spatially and temporally. Thus, the objective of this study was to develop a methodology to determine the short-term 𝐾𝑐 using biophysical parameters of coffee plants detected images obtained by an Unmanned Aircraft System (UAS). The study was conducted in Travessia variety coffee plantation. A UAS equipped with a digital camera was used. The images were collected in the field and were processed in Agisoft PhotoScan software. The data extracted from the images were used to calculate the biophysical parameters: leaf area index (LAI), leaf area (LA) and 𝐾𝑐. GeoDA software was used for mapping and spatial analysis. The pseudo-significance test was applied with p < 0.05 to validate the statistic. Moran’s index (I) for June was 0.228 and for May was 0.286. Estimates of 𝐾𝑐 values in June varied between 0.963 and 1.005. In May, the 𝐾𝑐 values were 1.05 for 32 blocks. With this study, a methodology was developed that enables the estimation of 𝐾𝑐 using remotely generated biophysical crop data.

Key words: