A. Castrignanò

Geostatistical modelling of within-field soil and yield variability for management zones delineation: a case study in a durum wheat field

The paper proposes a geostatistical approach for delineating management zones (MZs) based on multivariate geostatistics, showing the use of polygon kriging to compare durum wheat yield among the different MZs (polygons). The study site was a durum wheat field in southern Italy and yield was measured over three crop seasons. The first regionalized factor, calculated with factorial cokriging, was used to partition the field into three iso-frequency classes (MZs). For each MZ, the expected value and standard deviation of yield were estimated with polygon kriging over the three crop seasons. The yield variation was only in part related to soil properties but most of it might be ascribable to different patterns of meteorological conditions. Both components of variation (plant and soil) in a cropping system should then be taken into account for an effective management of rainfed durum wheat in precision agriculture. The proposed approach proved multivariate Geostatistics to be effective for MZ delineation even if further testing is required under different cropping systems and management.

Assessment of Groundwater Salinisation Risk Using Multivariate Geostatistics

The risk assessment at regional scale requires modelling spatial variability of environmental variables. Traditional approach, based on estimating point environmental indicators, cannot be considered satisfactory for this purpose, because it does not take into account spatial dependence between variables. We propose the application of an approach to the problem of groundwater salinisation, in which multivariate geostatistics and GIS are combined to integrate primary information with exhaustive secondary information. The dataset consisted of 454 private wells used for irrigation and located in Apulia region (south Italy). Three variables were processed: concentration of chlorides and nitrates, as primary variables, and the distance from the coast, as auxiliary variable. The approach highlighted the widespread degradation of water resources in the Apulian groundwater. The maps of the global indicator allowed us to delineate the zones at high risk of groundwater contamination and also to identify those parameters most responsible for water degradation, so that a wiser management of water resources could be planned. This approach can be used as operational support to a wide range of activities and in decision making among several remediation alternatives.

A data fusion and spatial data analysis approach for the estimation of wheat grain nitrogen uptake from satellite data

ABSTRACT The selection of the optimal band combination for the estimation of specific crop variables is a key aspect in order to obtain reliable estimation of in-field variability from multi- and hyperspectral remote-sensing data. The selection of the bands is strongly influenced by the phenological stage of the crop at the acquisition time. In this work, the influence of the growing stage on the combination of spectral bands related to grain nitrogen (N) uptake in wheat was evaluated using multispectral (Satellite Pour l’Observation de la Terre – SPOT) and hyperspectral (Compact High Resolution Imaging Spectrometer – CHRIS-PROBA) satellite images at different growth stages over two wheat growth seasons in central Italy. In order to identify the more appropriate covariates (spectral bands) for each phenological stage, stepwise regression with backward selection was combined with stepwise variance inflation factors (VIFs) analysis and linear mixed effect model (LMEM). The results obtained in this study suggest that the spectral region most related to N uptake varies over the growing season of the wheat crop. For SPOT data, near-infrared (NIR) region was selected at all the phenological stages in both growing seasons, except for the latest stage, with low chlorophyll content due to the onset of senescence, in which the red band was selected. At stem elongation, the shortwave infrared (SWIR) band of SPOT data was also selected. At this stage, the best N estimation accuracy was obtained using an LMEM (root mean square error, RMSE = 0.012 t ha−1). The inclusion of a spatial component in the estimation model by means of LMEMs provided a more accurate estimation than ordinary least square (OLS) models at all growth stages. The test carried out with CHRIS-PROBA data at the fourth stage confirmed the importance of NIR and in particular of the red-edge region for N uptake prediction. A novel methodology is proposed, which involves two crucial aspects in the context of the use of remote-sensing data in precision agriculture: i) the standardization of the spatial resolution for in-field and satellite data by a geostatistical data technique (data fusion); and ii) the selection of the most appropriate spectral bands for each phenological stage, taking into account both correlation with the target variable and collinearity.

Selecting optimal hyperspectral bands to discriminate nitrogen status in durum wheat: a comparison of statistical approaches

Hyperspectral data can provide prediction of physical and chemical vegetation properties, but data handling, analysis, and interpretation still limit their use. In this study, different methods for selecting variables were compared for the analysis of on-the-ground hyperspectral signatures of wheat grown under a wide range of nitrogen supplies. Spectral signatures were recorded at the end of stem elongation, booting, and heading stages in 100 georeferenced locations, using a 512-channel portable spectroradiometer operating in the 325–1075-nm range. The following procedures were compared: (i) a heuristic combined approach including lambda-lambda R2 (LL R2) model, principal component analysis (PCA), and stepwise discriminant analysis (SDA); (ii) variable importance for projection (VIP) statistics derived from partial least square (PLS) regression (PLS-VIP); and (iii) multiple linear regression (MLR) analysis through maximum R-square improvement (MAXR) and stepwise algorithms. The discriminating capability of selected wavelengths was evaluated by canonical discriminant analysis. Leaf-nitrogen concentration was quantified on samples collected at the same locations and dates and used as response variable in regressive methods. The different methods resulted in differences in the number and position of the selected wavebands. Bands extracted through regressive methods were mostly related to response variable, as shown by the importance of the visible region for PLS and stepwise. Band selection techniques can be extremely useful not only to improve the power of predictive models but also for data interpretation or sensor design.

A geostatistical approach for modelling and combining spatial data with different support

The paper proposes a geostatistical framework to solve the issues of heterogeneous support for spatial estimation. Apparent soil electrical conductivity (ECₐ) was measured in a field cropped with San Marzano tomato using a multiple frequency electromagnetic profiler with 6 operating frequencies. Mixed support kriging was used to estimate ECₐ taking into account the change of support. The method includes punctual kriging with the error being the dispersion variance associated with each frequency. The individual ECₐ maps were weighted by the dispersion variance to obtain a map which was used for field partition in management zones.

On-the-go Acquisition of Hyperspectral Data on a Durum Wheat Field - A Methodological Approach

Hyperspectral proximal sensors, operating in the Vis-NIR-SWIR ranges, are usually employed for static recording. The availability of data at a fine spatial resolution through on-the-go spectra collection would open new frontiers to this field of study, allowing in real time the acquisition of a huge amount of information related to plant response. In this paper we describe a methodological approach for analysing on-the-go hyperspectral data in order to delineate homogeneous zones in an agricultural field cropped with durum wheat. HS data were acquired in southern Italy at shooting stage of durum wheat. Spectral readings were recorded using a high resolution spectroradiometer, FieldSpec 4 (350-2500 nm). The sensor was mounted on a plot seeder. Collected data were subjected to pre-processing and then analysed through principal component analysis. Afterwards, retained factors were analyzed through block co-kriging to produce continuous maps. The method was very effective to disclose differences in the spectral response of the plants; however, the interpretation of the results in terms of agronomical behaviour of the wheat needs more survey and investigation.

Field Partitioning by Proximal Hyperspectral and Fluorescence Sensor Data and Multivariate Geostatistics

Hyperspectral and fluorescence devices can provide relevant information on physiological plant status related to canopy cover, plant nutrition, water status, pigments concentration and functionality. The aim of this study was to combine data from hyperspectral and fluorescence sensors with plant variables, to delineate homogeneous sub-field areas, using multivariate geostatistics. Proximal sensor and biometric data were collected in a 5-ha durum wheat field at anthesis stage, at 104 georeferenced positions. Fluorescence and hyperspectral data were analysed by principal component analysis to reduce the dimensions of the datasets; the retained components together with plant variables were analysed by means of a multivariate geostatistics approach, factorial co-kriging analysis. A linear model of coregionalization, fitted to the direct and cross experimental variograms of the Gaussian transformed variables, included a nugget effect and a spherical model with a range of 125 m. The first regionalised factor at 125m-scale, explaining 66.9% of the corresponding variance, was able to discriminate areas characterised by better overall plant status and photosynthetic performance from more stressed areas. The approach was sensitive to split the field into two main areas. However, repeated measurements over the crop season are needed to confirm the previous results.