Bioenergy and ecosystem services trade-offs and synergies in marginal agricultural lands: A remote-sensing-based assessment method

Bioenergy and ecosystem services trade-offs and synergies in marginal agricultural lands: A remote-sensing-based assessment method

Journal of Cleaner Production 237 (2019) 117672 Contents lists available at ScienceDirect Journal of Cleaner Production journal homepage: www.elsevi...

4MB Sizes 0 Downloads 30 Views

Journal of Cleaner Production 237 (2019) 117672

Contents lists available at ScienceDirect

Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro

Bioenergy and ecosystem services trade-offs and synergies in marginal agricultural lands: A remote-sensing-based assessment method Davide Longato a, b, Mattias Gaglio c, *, Mirco Boschetti d, Elena Gissi a a

University IUAV of Venice, Santa Croce 191, 30135 Venezia, Italy Department of Civil, Environmental and Mechanical Engineering, University of Trento, Via Mesiano, 77, 38123 Trento, Italy Department of Life Sciences and Biotechnology, University of Ferrara, Via L.Borsari 46, 44122 Ferrara, Italy d Institute for Electromagnetic Sensing of Environment, National Research Council, Via Bassini 15, 20133 Milano, Italy b c

a r t i c l e i n f o

a b s t r a c t

Article history: Received 27 July 2018 Received in revised form 10 April 2019 Accepted 15 July 2019 Available online 22 July 2019

Growing non-food crops in marginal lands has been proposed as a solution to avoid land competition with food production. Mapping marginal agricultural lands is therefore fundamental for the sustainable development of rural landscapes. This study proposes a method based on remote sensing data to identify marginal agricultural lands for the production of wood biomass, and analyse potential trade-offs and synergies between the new wood crops, food production, and Ecosystem Services (ES) provided by vegetation. The province of Rovigo (northern Italy) was chosen as a representative case study. Three classes of marginal agricultural lands were mapped through the use of the Soil Adjusted Vegetation Index (SAVI): i) abandoned or unused agricultural lands, ii) potentially poorly or non-managed croplands, and iii) potentially low productivity croplands. Results showed that marginal agricultural lands cover 1.7% of the agricultural areas of the province, and approximately 13,642 MWh yr1 of Second-Generation (2G) bioenergy can be produced in marginal agricultural areas while enhancing ES provided by vegetation, and avoiding any trade-off with food production. Since this energy potential covers just 8.4% of the total potential authorized in the province, the enhancement of ES could provide a suitable argument to support the conversion of marginal agricultural lands and increase the multifunctionality of the agricultural landscape. © 2019 Elsevier Ltd. All rights reserved.

Handling Editor: Yutao Wang Keywords: Ecosystem services trade-off Soil adjusted vegetation index Second generation bioenergy Agricultural landscape Renewable energies Environmental planning

1. Introduction Over the last decade, there has been a growing interest in bioenergy as renewable energy sources to reach the target of 20% of total energy production, according to the EU Directive 2009/28/EC (EC, 2009a). On the other hand, bioenergy production induces changes in landscape patterns, and trade-offs with other Ecosystem Services (ES) (Gissi et al., 2016, 2018), especially with regulating and other provisioning ES, such as food provision (Power, 2010). According to the Common International Classification of ES (CICES, Haines-Young and Potschin, 2011), Biomass-Based Energy Sources (BBES) are considered as a provisioning ES. Their positive and negative effects on ecological functions have to be considered when

* Corresponding author. E-mail address: [email protected] (M. Gaglio). https://doi.org/10.1016/j.jclepro.2019.117672 0959-6526/© 2019 Elsevier Ltd. All rights reserved.

assessing trade-offs and synergies between ES (Gissi and Garramone, 2018). Bioenergy production can compete with the food sector either directly (food crops used as energy sources) or indirectly (bioenergy crops cultivated on land otherwise used for food production), impacting on food prices and security if the demand for agricultural products or land is significantly high (Popp et al., 2014). To deal with direct conflicts with food production, the European Union proposed to encourage the development of the SecondGeneration (2G) bioenergy and biofuel feedstocks (EC, 2009a), within a long-term roadmap for energy production that points to increase 2G feedstocks in the medium term (Holland et al., 2015). 2G feedstocks refer to ligno-cellulosic materials coming from nonfood biomass, including by-products (cereal straw, sugar cane bagasse, forest residues), wastes (organic components of municipal solid wastes), and dedicated feedstocks (purpose-grown vegetative grasses, short rotation forests and other energy crops) (Sims et al.,

2

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

2010). 2G-dedicated feedstocks are defined as perennial, lignocellulosic, non-food crops (Valentine et al., 2012; Milner et al., 2016). In temperate climates, they are likely represented by Miscanthus and fast-growing trees, such as poplar and willow as short-rotation coppice (SRC), or poplar as short-rotation forestry (SRF) (Hastings et al., 2014; Milner et al., 2016). Besides not directly competing with food production, 2G-dedicated feedstocks from wood crops (2G wood crops from now on) have the potential to complement intensive agriculture and provide positive environmental externalities (Bartle and Abadi, 2010; Gaglio et al., 2019). The impacts of agricultural land use transition to 2G wood crops on ES resulted in positive variations on several ES e i.e., hazard regulation, disease and pest control, pollination, soil quality, water quality e and biodiversity, but negative ones on food provision (Holland et al., 2015; Milner et al., 2016). Most importantly, 2G wood crops can be grown on poorer quality land and more marginal areas (i.e., land with low agricultural or biodiversity value, or abandoned land no longer suitable for quality food production) than those required for food production (Hastings et al., 2009a,b; Tilman et al., 2009; Milner et al., 2016). Anyway, energy crops, including 2G wood crops, will still be grown on lands in competition with food and fiber production (Sims et al., 2010). Growing energy crops on marginal agricultural lands is therefore considered as an effective strategy to reduce (direct and indirect) conflicts between bioenergy and food provision. In the case when a transition from annual crops to perennial vegetation occurs, as the case of 2G wood crops, energy crops can enhance multiple ES, as, for example, improved wildlife habitat and diversity, soil carbon sequestration, soil and water quality, and erosion control (Blanco-Canqui, 2016). Many authors focused on the use of marginal, abandoned lands, or areas not currently used for food cultivations for the exploitation of bioenergy (e.g., Fiorese and Guariso, 2010; Blanco-Canqui, 2016; Edrisi and Abhilash, 2016). In this sense, marginal lands can play a crucial role in bioenergy production from 2G wood crops (Kang et al., 2013). However, the identification of marginal lands is still under debate. Firstly, marginal lands are not univocally defined. Marginal lands have been generally described as soils that suffer for limitations, such as physical or chemical problems, uncultivated, affected by adverse climatic conditions, or characterized by low productivity croplands (Blanco-Canqui, 2016). These partially degraded areas are often classified as marginally productive croplands. Together with abandoned croplands, they can be suitable for growing 2G wood crops under proper management (Blanco-Canqui, 2016). Secondly, agricultural land marginality is a complex phenomenon depending on different elements (biophysical, agronomic and economic factors, the plants being considered, the technological level at different time periods); due to such complexity, it is very difficult to objectively measure the degree of marginality with some yardstick (Soldatos, 2015). Sallustio et al. (2018) mapped economic marginality of agricultural lands at the national scale in Italy, on the basis of the economic values of agricultural lands. Besides not being aimed at identifying marginal lands for bioenergy, the proposed classification considered only economic factors. It excludes, for example, factors that would have depicted the complex conditions for marginality, as, for example, biophysical marginality with a spatial geographical specificity. Fiorese and Guariso (2010) mapped marginal agricultural lands, i.e., abandoned agricultural land and set-aside land, at regional scale (Emilia Romagna Region, Northern Italy), in order to evaluate the biomass potential from energy crops on the basis of land use and statistical data. Like Sallustio et al. (2018), also Fiorese and Guariso (2010) did not take into account any biophysical parameter. Moreover, the method is based on land use and statistical data that are not regularly updated, besides the fact that set-aside practices were

abolished within the Common Agricultural Policy through the EU Council Regulation No 73/2009 (EC, 2009b). In addition, if marginal lands would be cultivated with 2G wood crops for bioenergy production, these areas have to be geographically and spatially identified and analyzed. In fact, in order to design an effective supply chain, areas producing feedstocks should be assessed for their contribution in terms of feedstock potential, but also for their spatial distribution, in order to assess transportation costs as well as the sustainability of the chain, e.g., in terms of greenhouse gasses (GHG) emissions. A growing number of studies are focusing on the optimization techniques for the design of effective supply networks as a fundamental challenge for sustainable goals (Fathollahi-Fard et al., 2018a). These theories find applications in several domains, such as circular economies (Sahebjamnia et al., 2018), health care (Fathollahi-Fard et al., 2018b), species conservation (Fathollahi-Fard and HajiaghaeiKeshteli, 2016), and renewable energies (Sarker et al., 2019), among others. However, assessing the spatial distribution of supply areas is at the basis of the design of effective supply chains to ensure economic feasibility and improve environmental performances, within the framework of sustainable management strategies. When applied to biological resources, as in the case of 2G bioenergy feedstocks, remote sensing techniques are valuable instruments for mapping potential supply areas according to crop phenological characteristics. Remote sensing can support both the identification of marginally productive and abandoned/unused lands together with the assessment and mapping of ES and their trade-offs and synergies. Firstly, to map land marginality, processing multi-seasonal images allows to accurately capture the different phenological stages during the growing season, hence to map crop presence and rotation (Low et al., 2013; 2015; Conrad et al., 2014). Moreover, multiannual images can distinguish long-term from short-term changes, and are needed to detect changes due to management intensities and related perennial developments (Stefanski et al., 2014; Low et al., 2015). Alterations in the surface conditions (i.e., type and density of the vegetation cover) is often a symptom of cropland abandonment, which can be mapped by the support of satellite images (e.g., Weissteiner et al., 2011; Alcantara et al., 2012; Low et al., 2015). Secondly, remote sensing can provide quantitative, spatially-explicit and (in some case) physically-based estimates of a number of biophysical parameters underpinning ecological functions (Andrew et al., 2014). For example, narrowband and broadband Vegetation Indices (VIs) can be used as indicators of productivity in a crop growing season, since time integral of the VIs is a recognized proxy of seasonal productivities (Verma et al., 2013; Zhang et al., 2018; Semeraro et al., 2019). In fact, they are able to characterize variations in phenology and photosynthetic potential of crops, useful to identify the cropping cycle and growth (e.g., De Araujo Barbosa et al., 2015; Boschetti et al., 2017, Malladi and Sowlati, 2017). The analysis of vegetation behavior and trend through VIs derived from satellite data was already applied for mapping abandonment in agricultural land in different studies. Field observation data from ground campaigns or medium resolution Land Use/Land Cover (LULC) data were used to perform image classification and analysis for the identification of marginal lands on the dinaru et al., basis of vegetation pattern (e.g., Low et al., 2015; Gra 2019). Some efforts in mapping marginal areas that produced land abandonment have been done adopting the syndrome approach (Downing et al., 2002), and exploiting geospatial information and remote sensing time series synthetized in a fuzzy logic approach (Weissteiner et al., 2011). However, none of these studies focused on the use of satellite data to analyze spatial and temporal vegetation dynamics in single parcels of agricultural land.

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

This paper proposes a methodology based on the analysis of satellite data to identify marginal agricultural lands at the parcelscale to be potentially destined to 2G wood crops. We also analyze and discuss trade-offs and synergies with food provision and ES provided by vegetation resulting from the potential land use change to 2G wood crops. The method is applied in the area of the province of Rovigo, an intensively cultivated rural landscape in northern Italy. In order to answer to the need of avoiding direct and indirect conflicts between bioenergy and food production, marginal agricultural lands are defined as croplands that are i) potentially not currently or poorly involved in food production, and ii) currently involved in food production but affected by potentially low productivity. In the following sections, after introducing the study site, we firstly describe the dataset of remote sensing data exploited for the analysis (sec. 2.2.1 and 2.2.2). Then, we describe the methodology for the identification of marginal agricultural lands through satellite data (sec. 2.2.3, 2.2.4, and 2.2.5). Finally, we perform the mapping and analysis of trade-offs and synergies between 2G wood crops (called “wood BBES” by reference to the classification of ES) and other ES in the identified marginal agricultural lands (sec. 2.3). 2. Materials and methods 2.1. Study site The province of Rovigo is located in the Veneto Region (northern Italy) (Fig. 1). The area has been subjected to a rapid increase of permit demand for bioenergy plants, as a result of national incentives granting (Gissi et al., 2016). Its territory is characterized by a flat plain, bounded to the north by the Adige River and to the south by the Po River. With an area of 1,789 km2, it stretches for about 100 km till the Po River Delta in the eastern part. The province has a semi-continental climate with precipitations mainly in spring and autumn. It is an intensely cultivated area mostly destined to agricultural use (74%), while forested and semi-natural areas are very limited.

3

The main crop seasons that can be identified in northern Italy are: i) from autumn to late spring/early summer for winter crops (e.g., durum/soft wheat and/or barley), and ii) from spring to early autumn for summer crops (corn, rice, soybean, etc.) (Azar et al., 2016). In the province of Rovigo, agricultural land is mostly occupied by arable lands (93% of total agricultural areas) and large part of it is cultivated with the three most common annual crops: wheat, corn, and soybean (Istituto Nazionale di Statistica, Agricoltura e Zootecnia: http://agri.istat.it/, accessed 15 June 2018). The typical calendar of the main crops is shown in Fig. 2. As regards the bioenergy production, in the province of Rovigo the production of biomass for energy purposes has already produced conflicts with food provision (Gissi et al., 2011). At the same time, the regional Rural Development Programme 2014e2020 (Regione del Veneto, 2015) aims to promote the diversification of rural activities, especially those that may generate positive impacts on the environment. 2.2. Identification of marginal agricultural lands: methodological approach Given the need of avoiding direct and indirect conflicts between bioenergy and food production, the methodological approach for the identification of marginal agricultural lands is based on the assumption that such areas include: i) agricultural lands officially classified as abandoned or unused, ii) undermanaged or nonmanaged croplands (class: “potentially poorly or non-managed areas”) and iii) managed croplands showing low productivity (class: “potentially low productivity areas”). Fig. 3 provides the methodological workflow of the different steps involved in the marginal agricultural land identification. The process can be divided in five steps: data acquisition (Step 1), data pre-processing and extraction of “abandoned or unused agricultural lands” (type 1) from the LULC map (Step 2), detection of “potentially poorly or non-managed croplands” (type 2) by means of crop presence analysis (Step 3), assessment of “potentially low productivity croplands” (type 3) by means of productivity proxy

Fig. 1. Province of Rovigo: localization (a), land use/land cover map (b), and study area (c).

4

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

Fig. 2. Typical crop calendar and phenology of major crops in the province of Rovigo (adapted from Azar et al., 2016 and CREA - Consiglio per la Ricerca in Agricoltura e l’Analisi dell'economia agraria: http://www.crea.gov.it) and dates of Landsat 8 OLI scenes acquired.

analysis (Step 4), and merging the previously generated information to create the final set of marginal agricultural lands (Step 5). 2.2.1. Data acquisition (step 1) In order to identify marginal agricultural lands, the methodology makes use of a dataset that includes both a thematic cartography (LULC map) and satellite data. A LULC map was acquired from the Land Parcel Information System (LPIS) of the Veneto Regional Agency for Payment in Agriculture (AVEPA). LPIS is a geo-spatial database, which is part of the identification system for all agricultural cadastral parcels in EU member states. The LULC map (updated in 2016) provides information on the crop types cultivated within each parcel and already identifies some marginal agricultural areas (i.e., the LULC classes attributable to “abandoned or unused agricultural lands”). In our method, these areas were considered as marginal agricultural lands without further analysis. However, this dataset largely underestimates the real amount of marginal lands in the study area, as it represents a static picture based on official farmers’ declaration of what is already abandoned or unused. The dataset cannot provide any information about areas that are undermanaged or less productive for socio-economic reasons or environmental limitations that can reduce their potential crop production. These conditions can be identified only through a direct field survey, the acquisition of yield data at the field level, or by analyzing agronomic and biophysical crop parameters. Due to the lack of such direct data, a proxy of crop productivity was derived from a multi-temporal series of satellite data. Landsat 8 OLI (Operational Land Imager) Surface Reflectance data at 30 m spatial resolution, together with VIs and associated cloud cover mask data, were collected for the years 2014, 2015, and 2016 from the U.S. Geological Survey (USGS) Earth Explorer database (https://earthexplorer.usgs.gov/, accessed 20 June 2018) for the study area (path 192, row 29). The dates of Landsat images were chosen in order to capture the phenology of both winter and summer crop seasons, especially the vegetation growth period between the sown and the harvest (Fig. 2). After screening for data

availability and cloud coverage, a total of 21 scenes (i.e., 6, 7, and 8 scenes for 2014, 2105, and 2016 respectively) were acquired. The Soil Adjusted Vegetation Index (SAVI) was chosen for the analysis of crop seasonal dynamics (i.e., vegetation phenology), since it was developed to minimize the effect of soil background on canopy spectral (Huete, 1988), unlike, for example, the most commonly used Normalized Difference Vegetation Index (NDVI). 2.2.2. Data pre-processing (step 2) Because of their predominant use for conservation targets, areas located within the protected areas of Natura 2000 network were excluded. The analysis of marginal agricultural lands was performed only on agricultural areas covered by annual arable crops, extracted from the LULC map. Perennial crops and other types of cultivations (e.g., horticultural crops, nursery plants, orchards) were therefore masked out. Due to the fact that the minimum area extension considered suitable for cultivating 2G wood crops for commercial purposes is 1 ha, we also excluded from the analysis all the LULC polygons smaller than that size. Finally, polygons containing one or more cloudy pixels (information provided by the cloud cover mask available for each satellite scene) were considered to have not sufficient information to be analyzed and, consequently, were excluded from the analysis. 2.2.3. Crop management analysis (step 3) The crop management analysis was performed to distinguish between “potentially poorly or non-managed croplands” and “managed croplands”. A parcel is considered as “managed cropland” if it is characterized by a clear crop phenological dynamic, hence showing vegetation cover signal during the considered winter or summer season. This analysis consists in a multitemporal interpretation of the seasonal phenology occurring within the parcels of the study area. The hypothesis is that winter and summer crops present bare soil during the summer and winter season, respectively. In our case, in order to identify the vegetation pattern and the seasonal phenology, an average seasonal SAVI value

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

5

Fig. 3. Methodological workflow used for the identification of marginal agricultural lands.

(SAVI_AVG) was calculated for each year from the available images in both winter (January to May) and summer (June to September) seasons, producing a total of six images for the analyzed period (2 seasons for 3 years). The next step involved the identification of the seasonal presence/absence of vegetation cover by classifying each parcel according to the specific SAVI_AVG value. In order to define the appropriate threshold value for the identification of the two classes (“vegetated” and “bare soil”), SAVI statistics were extracted from some visually inspected parcels. A stratified random sampling

was performed on the available polygons to select more than 1400 fields per year (see Table 1). A minimum of two-third of the polygons labelled as “vegetated” were used as training areas to extract SAVI statistics, while the remaining ones (together with some other polygons labelled as “bare soil”) were used for the validation of the classification results. Table 1 provides the details of the selected polygons and their separation in training and validation for each year (2014, 2015, and 2016) and season (summer and winter) analyzed.

6

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

Table 1 Training and validation set of polygons according to each year and crop season. Year

Crop season

Vegetation cover

Training polygons

Validation polygons

2014

Winter

yes no yes no yes no yes no yes no yes no

314

156 143 218 130 221 151 180 168 149 151 198 148

Summer 2015

Winter Summer

2016

Winter Summer

477 480 401 324 450

The SI threshold was defined on the basis of the mean (m) and standard deviation (s) values extracted from training data (Barbosa et al., 1999; Smith et al., 2007; Boschetti et al., 2010). The thresholds used for the classification of vegetated parcels were calculated through the formula:

thi ¼ mi  2si

(1)

where th is the SI threshold of the i satellite image, m is the mean and s is the standard deviation of the SAVI_AVG values extracted from the training polygons of the i satellite image. All the parcels with SAVI_AVG values higher than the respective threshold were considered to be vegetated, hence classified as “cultivated croplands” with summer, winter, double, or mixed crops, depending on the occurrence of the identified vegetation coverage. Parcel classification is explained in detail in Appendix A. Parcels classified as “cultivated croplands” in at least one of the three-year period were considered to be “managed croplands”. Parcels that were not classified as “cultivated croplands” because of lack of vegetation coverage in all the three years were re-classified as “potentially poorly or non-cultivated croplands”. The fields falling in this category were considered to be marginal agricultural lands since potentially not involved or poorly involved in food production during the whole 3-years period of analysis. An accuracy assessment of the maps produced with the proposed approach was performed for each year exploiting the data reported in Table 1 (see the section 3.1). 2.2.4. Cropland productivity analysis (step 4) In this step, parcels classified as “managed croplands” were analyzed to highlight conditions of potentially low productivity. We considered the SAVI (i.e. SAVI_AVG) as a proxy of crop productivity. In fact, when satellite VIs are acquired on specific phenological periods in correspondence with the vegetative peak (single scene) or the seasonal growing season (multi-temporal series), they are strongly correlated to the herbaceous biomass or crop yield (Boschetti et al., 2007; Benedetti and Rossini, 1993). Archives of VIs data from decametric instruments such as Landsat can be an effective tool to identify the crop yield variability within fields, even if characterized by a low fertility condition (Pascucci et al., 2018). Under this assumption, parcels showing a statistically lower SAVI_AVG value, with respect to the corresponding population of all the analyzed fields for a specific season, were assumed to have a potentially low productivity condition. For a detailed description of the productivity analysis, see Appendix B. 2.2.5. Final set of marginal agricultural lands (step 5) As resulting from the step-by-step methodology, the final set of marginal agricultural lands is composed by: i) abandoned or

unused agricultural lands defined by the LULC map (step 1), ii) potentially poorly or non-managed croplands (step 3), and iii) potentially low productivity croplands (step 4). Their spatial distribution was described using a kernel density analysis, computed in ArcMap 10.4 (ESRI) using a spatial variant of Silverman's Rule of Thumb that is robust to outliers. The energy potential from wood biomass was then calculated under the hypothesis of converting the entire surface of identified marginal agricultural lands to 2G wood crops. The calculation was computed using the data of the lower heating value obtained from a power plant located near the study area, and hypothesizing the cultivation of the most common two-years cycle wood energy crops present in the study area. These includes two natives, poplar (Populus spp.) and willow (Salix spp.), and one non-native species, black locust (Robinia pseudoacacia), having an overall average yield of 12.2 t ha1 (dry matter) (ARPAV, 2010). 2.3. Trade-offs and synergies between wood BBES and other ES in marginal agricultural lands In order to map trade-offs and synergies between wood BBES and other ES, in line with Gissi et al. (2016, 2018), we defined the trade-off as a potential conflict between the exploitation of wood BBES and the provision of another ES (specifically, food provision), and the synergy as a potential co-benefit between the exploitation of wood BBES and the provision of another ES, meaning that the two are simultaneously provided. In our case, the potential cobenefits concern some ES provided by vegetation. Their provision, which relies on ecological functions that in turn depend on some biophysical processes occurring in plant biomass, is correlated to the primary productivity of vegetation (Richmond et al., 2007), and thus can be mapped through VIs. Table 2 reports the list of such ES on the basis of the previous literature. The SAVI was therefore used as a proxy to map the ES provided by vegetation. Trade-off and synergy analysis was performed using the method described by Gissi et al. (2016, 2018). 2.3.1. Trade-offs between wood BBES and food provision The bioenergy feedstock provision competes with food provision by leading to land diversion from food production (Popp et al., 2014; Gissi et al., 2018). The destination of marginal agricultural lands not involved in food production to bioenergy feedstock provision avoids such trade-off. However, since marginal agricultural lands classified as “potentially low productivity croplands” (see Section 2.2.4) supply food products, even if in a relative limited amount, the attribution of a trade-off is needed to highlight the potential impact of their conversion to 2G wood crops. For this reason, a “low” level of trade-off with wood BBES (i.e., mitigated trade-off) were attributed to the “potentially low

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

7

Table 2 List of ES provided by vegetation (and related biophysical processes) that can be mapped using vegetation indices as a proxy. Ecosystem Service Biophysical process

References

Climate regulation Carbon sequestration/storage by De Araujo Barbosa et al., (2015); Feng et al., (2010); Atzberger, (2013); Rembold et al., (2013); Pettorelli, (2013); vegetation Pettorelli et al., (2014); Zurlini et al., (2014); Egoh et al., (2007) Soil erosion Cover of vegetation De Araujo Barbosa et al., (2015); Andrew et al., (2014); Ayanu et al., (2012) regulation Natural hazard Cover of vegetation for mass De Araujo Barbosa et al. (2015) regulation stabilisation Water cycling and Structural and functional Zurlini et al. (2014) regulation properties of vegetation Maintenance of soil Structural and functional Ayanu et al., (2012); Zurlini et al., (2014) fertility properties of vegetation

productivity croplands”, while a “null” level (i.e., trade-off completely avoided) to the “potentially poorly or non-managed croplands” and “abandoned or unused agricultural lands”.

2.3.2. Trade-offs and synergies between wood BBES and ES provided by vegetation Growing new 2G wood crops on marginal agricultural lands can lead to a positive or negative variation of vegetation cover and related primary productivity, depending on the previous condition of the interested parcel. Assuming that ES provided by vegetation (see Table 2) can be mapped and assessed using the SAVI as a proxy (see Section 2.3), after calculating the three-year SAVI_AVG value for the whole province of Rovigo, we firstly estimated the three-year SAVI_AVG statistics (namely, mean (m) and standard deviation (s)) from the existing wood energy crops identified in the LULC map. Secondly, in order to detect potential gain or loss of vegetation cover and density and related ES, we compared those values with the three-year SAVI_AVG values of each parcel classified as marginal agricultural lands. In doing so, we assumed that the new 2G wood crops will have similar conditions of vegetation cover and density and, consequently, of SAVI_AVG values to those of the current wood energy crops. The comparison between the current SAVI_AVG values of marginal agricultural lands and the predicted ones (due to the potential land use change) resulted in the attribution of synergies and tradeoffs, depending on the type of SAVI_AVG variation. The detailed explanation of the analysis is reported in Appendix C.

2.3.3. Combinations of impact of wood BBES on ES Once trade-offs and synergies with food provision and ES provided by vegetation were assessed and mapped in marginal agricultural lands, the two maps were overlapped in order to obtain different combinations of impact, following the method used in Gissi et al. (2016, 2018). Out of the possible six combinations obtainable by the maps overlapping, five combinations emerged (Table 3). For a detailed description of the five combinations see Appendix D.

3. Results 3.1. Accuracy assessment The results show that both the user's and producer's accuracy is over 85% in all the seasons considered (Table 4), meaning that our method, based on statistical calibration of SI thresholds, is robust in mapping the seasonal vegetation cover. With respect to each single seasonal map performance, the lower accuracy value was observed for the map of winter crops in 2014 (Overall Accuracy ¼ 95.99%, k ¼ 0.92), while the higher value was found for the map of summer crops in 2015 (OA ¼ 99.71%, k ¼ 0.99). In almost all the six seasons considered, the user's and producer's accuracy values related to the class “no vegetation cover” (i.e., bare soil) reached 100%. Validation data show that misclassifications can occur when the vigor and/or density of the vegetation is very low in early season images and when there is an early crop harvesting. 3.2. Identification and spatial distribution of marginal agricultural lands A total of 2,257 ha of marginal agricultural lands was identified Table 4 Accuracy assessment for each year and crop season. Year

Crop season

2014

winter Producer's summer

2015

Producer's winter Producer's summer

2016

Producer's winter Producer's summer Producer's

Vegetation cover

yes no accuracy yes no accuracy yes no accuracy yes no accuracy yes no accuracy yes no accuracy

(%)

(%)

(%)

(%)

(%)

(%)

yes

no

User's accuracy (%)

144 0 100.00 212 0 100.00 211 0 100.00 179 0 100.00 148 0 100.00 194 1 99.49

12 143 92.26 6 130 95.59 10 151 93.79 1 168 99.41 1 151 99.34 4 147 97.35

92.31 100.00 OA ¼ 95.99% 97.25 100.00 OA ¼ 98.28% 95.48 100.00 OA ¼ 97.31% 99.44 100.00 OA ¼ 99.71% 99.33 100.00 OA ¼ 99.67% 97.98 99.32 OA ¼ 98.55%

k ¼ 0.92

k ¼ 0.96

k ¼ 0.94

k ¼ 0.99

k ¼ 0.99

k ¼ 0.97

Table 3 Trade-off and synergy combinations of impact on ES occurring under the hypothesis of cultivating 2G wood crops in marginal agricultural lands.

Comb. Comb. Comb. Comb. Comb.

1 2 3 4 5

Trade-off with food production

Trade-off with ES provided by vegetation

Synergy with ES provided by vegetation

Null Null Null Low Low

Null Null Potential Null Null

Potential Null Null Potential Null

8

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

to be suitable for the cultivation of 2G wood crops (Table 5). First, 18 ha of “abandoned or unused agricultural lands” (type 1) were identified from the LULC map (step 2). Second, the crop management analysis (step 3) resulted in the identification of 443 ha of “potentially poorly or non-managed croplands” (type 2) (Fig. 4). Another 1,796 ha of “potentially low productivity croplands” (type 3) were identified in the subsequent productivity analysis (step 4) (Fig. 5). Overall, marginal agricultural lands cover 2.3% of the study area e i.e., agricultural areas covered by annual arable crops e (96,882 ha) and 1.7% of the total agricultural areas of the province of Rovigo (135,785 ha). The kernel density analysis of marginal agricultural lands (Fig. 6) highlights that these areas are concentrated around the municipality of Rovigo, and along the provincial road that connects the two major urban poles outside the province in the north/south direction. This area is characterized by the presence of the fossil dunes of the Po river delta (Gissi et al., 2016).

with food production nor with ES provided by vegetation. These areas could potentially generate 13,642 MWh per year. Notably, the 95.8% of the area covered by marginal agricultural lands could be destined to 2G wood crops without harming ES provided by vegetation (comb. 1, 2, 4, and 5), for a potential generation of 67,762 MWh per year. Furthermore, most of the marginal agricultural lands area (74.9%, classified with potential synergies) could contribute to enhance ES provided by vegetation, while generating a bioenergy potential of 52,952 MWh per year. 19.6% of the area occupied by marginal agricultural lands could be exploited without harming food provision (comb. 1, 2, and 3), generating a bioenergy potential of 13,843 MWh per year. However, for the fact of using marginal agricultural lands for bioenergy generation, the level of trade-offs with all the ES analyzed is at most low. Finally, it has to be noted that it was not possible to assign any combination to 3.9% of marginal agricultural lands area because of the lack of data due to cloud coverage.

3.3. Distribution of the energy potential from wood biomass and relation with other ES

4. Discussion 4.1. Management implications

Considering the average yield of the most common wood energy crops, the conversion parameters, and a yearly functioning of 7,500 h for a typical energy plant, a total energy of 71,903 MWh per year can potentially be obtained from the annual wood production of 27,081 t of dry matter in the marginal agricultural lands identified. Trade-offs between wood BBES and food provision and tradeoffs and synergies between wood BBES and ES provided by vegetation are mapped in Fig. 7. Tables 6 and 7 show the results of the trade-off and synergy analysis in marginal agricultural lands according to the different mapped categories. There is no conflict (null trade-off) and a low trade-off with food provision in 20.4% and 79.6% of the area covered by marginal agricultural lands, respectively. Regarding the tradeoffs and synergies with ES provided by vegetation, a synergy in 74.9% of the area covered by marginal agricultural lands was found. The remaining area is occupied by a null trade-off/synergy and a low trade-off in 20.9% and 0.3% of such area, respectively. 3.4. Trade-off and synergy combinations between wood BBES and other ES Fig. 8 shows the different combinations of impact on ES (i.e. food provision and ES provided by vegetation), obtained by overlapping the trade-off and synergy maps from the section 3.3. Table 8 reports the area distribution and the energy potential that can be obtained from marginal agricultural lands, aggregated for the different combinations of impact on ES. The analysis of combinations of impact shows that 19.3% of the area occupied by marginal agricultural lands (comb. 1 and 2) can be exploited for 2G wood crops without causing trade-offs neither

Exploiting 2G wood crops instead of dedicated food crops avoids the competition for the final use of the product. On the other hand, 2G wood crops require dedicated areas to be cultivated, thus leading anyway to the competition with food/feed crops for land use. Land use transition from croplands to 2G-dedicated feedstocks has positive effects on many priority ES, but negative effects on crops and livestock production (Holland et al., 2015). The use of marginal agricultural lands for bioenergy production can prevent inappropriate land use change of areas suitable for food production. This would enhance food security in a food reservoir such as the province of Rovigo, which serves the densely inhabited northern Italian plain (Gissi et al., 2016). At the same time, marginal land restoration can limit agricultural land degradation, which affects landscape functionality (Gaglio et al., 2017; Sallustio et al., 2018). Our method based on remote sensing data allows a spatiallyexplicit identification and mapping of marginal agricultural lands. Spatial information is essential to delineate areas prone to specific conditions (e.g., flood-prone areas; Samela et al., 2017; 2018), including land abandonment, as well as for land use and energy planning, especially when planning energy supply chains (Sarker et al., 2019). These maps are particularly important because updated spatially-explicit information are not commonly available or can be generated only at coarse resolution by analyzing statistics on rural areas, which are usually collected at the administrative level (e.g., in the best case at the municipal level). Understanding the energy potential that can be obtained from the cultivation of 2G wood crops in marginal agricultural lands is a crucial step to design an effective energy supply chain. In our case study, an energy potential of 71,903 MWh yr1 may be obtained from a total of 2,257 ha of marginal agricultural lands. This potential

Table 5 Statistics of marginal agricultural lands according to their different typologies. Marginal agricultural lands

Number of parcels Total area (ha) Average area (ha) Min area (ha)a Max area (ha) a

Potentially poorly or non-managed

Potentially low productivity

Abandoned or unused (from LULC map)

Total

182 443 2.43 1 16.4

699 1796 2.57 1 35.63

9 18 1.99 1.02 3.93

890 2257 2.54 1 35.63

For the sake of economic efficiency in cultivating 2G dedicated feedstocks, a minimum area of 1 ha was set.

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

9

Fig. 4. Map of “potentially poorly or non-managed croplands”.

Fig. 5. Map of “potentially low productivity croplands”.

may cover 43.8% of the total potential authorized in the province of Rovigo (tot ¼ 164,250 MWh yr1; Gissi et al., 2016). However, considering only marginal agricultural lands with no trade-offs with food provision, the potential (13,843 MWh yr1) may cover just 8.4% of the total potential authorized. When these potential values are compared with the projection for the year 2020 (provided by the Provincial Energy Program; Provincia di Rovigo, 2010) of the total energy demand for the agricultural sector, the results of our analysis highlight that cultivating 2G wood crops in agricultural marginal lands could significantly support the local agricultural sector toward sustainable goals. In fact, the bioenergy potential that could be obtained may exceed the provincial energy demand of the agricultural sector by 115.8%, 117.3%, or 119.1%, according to the high, medium, and low growth scenarios, respectively. However, when considering only marginal agricultural lands with no tradeoffs with food provision, the potential contribution falls to 22.3%, 22.6%, or 22.9%, respectively. Moreover, our results show that marginal agricultural lands are scattered around the province of Rovigo. Consequently, their fragmented spatial distribution further reduces the economic

feasibility of their exploitation for energy purposes because of the potentially high collecting costs. In fact, a significant portion of the total biomass cost is related to transportation costs (Noon and Daly, 1996; Malladi and Sowlati, 2017). This also affects the performances in terms of GHG savings, since negative impacts arise proportionally with the increase of transport route length between feedstock production and storing sites, especially if the transportation is ensured by trucks (Zhang et al., 2015). To optimally locate storage sites and energy plants, the kernel density map (Fig. 6) provides suitable spatial information that need to be analyzed in combination with environmental and legislative restrictions. We also found that density hotspots are located around the main urban area of the province (the city of Rovigo), and along the main road infrastructure running north/south. The former are peri-urban areas, which host urban population probably not devoted to agriculture. Their marginality is likely a consequence of conflicts between intense agricultural activities and urban population, as already witnessed in other peri-urban areas (e.g., Ahmed et al., 2011), or potentially due to farm abandonment for other jobs in the nearby city. The latter could be driven by the disturbance of traffic and

10

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

Fig. 6. Map of the kernel density analysis of marginal agricultural lands.

Fig. 7. Map of trade-offs between wood BBES and food provision (a) and map of trade-offs and synergies between wood BBES and ES provided by vegetation (b) in marginal agricultural lands.

industrialization. Understanding the drivers of marginality of agricultural parcels is a key challenge, which is of primary importance, especially to build effective management strategies for the 2G-dedicated feedstock production. Further investigations that

consider both ecological and socio-economic factors is needed to understand the phenomena driving the marginality status of agricultural lands and their spatial distribution. Anyway, the trade-off and synergy analysis demonstrated that

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

11

Table 6 Trade-off levels between wood BBES and food provision in marginal agricultural lands according to their different typologies. Marginal agricultural lands

Trade-offs with food provision

Potentially poorly or non-managed

Number of Area (ha) Number of Area (ha) Number of Area (ha) Number of Area (ha)

Potentially low productivity Abandoned or unused (from LULC map) Total

parcels parcels parcels parcels

Null

%

Low

%

182 443 0 0 9 18 191 461

20.5 19.6 0 0 1 0.8 21.5 20.4

0 0 699 1,796 0 0 699 1,796

0 0 78.5 79.6 0 0 78.5 79.6

Table 7 Trade-off and synergies between wood BBES and ES provided by vegetation in marginal agricultural lands according to SAVI_AVG thresholds. Marginal agricultural lands

Trade-offs and synergies with ES provided by vegetation

with SAVI_AVG value < 0.348 with SAVI_AVG value between 0.348 and 0.465 with SAVI_AVG value > 0.465 Total

Number of Area (ha) Number of Area (ha) Number of Area (ha) Number of Area (ha)

parcels parcels parcels parcels

Synergy

%

Null trade-off/synergy

%

Low trade-off

%

No Data

%

681 1,690 0 0 0 0 681 1,690

76.5 74.9 0 0 0 0 76.5 74.9

0 0 183 473 0 0 183 473

0 0 20.6 20.9 0 0 20.6 20.9

0 0 0 0 3 6 3 6

0

e e e e e e 23 88

e e e e e e 2.6 3.9

0 0.3 0.3 0.3 0.3

Fig. 8. Map of trade-off and synergy combinations (as described in Table 4) between wood BBES and other ES (i.e., food provision and ES provided by vegetation) in marginal agricultural lands.

Table 8 Trade-off and synergy combinations between wood BBES and other ES: area distribution and energy potential. Marginal agricultural lands

Trade-off and synergy combinations between wood BBES and other ES Combination 1

Combination 2

Combination 3

Combination 4

Combination 5

No data

Number of parcels Number of parcels (%) Area (ha) Area (%) Energy potential (MWh per year)

177 19.9 410 18.2 12,846.27

9 1.0 25 1.1 796

3 0.3 6 0.3 200.88

504 56.7 1280 56.7 40,106.04

174 19.5 447 19.8 14,013.63

23 2.6 88 3.9 2,760.62

using marginal agricultural lands to cultivate 2G wood crops not only involves low levels of trade-off with food production, but can also have a general positive effect on several other ES (Table 3). In

fact, significant ameliorations of ecological functions, and the related enhancement of ES provision, can be obtained by cultivating 2G wood crops in marginal areas characterized by previous lower

12

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

level of primary productivity and soil carbon storage (Gissi et al., 2014; Blanco-Canqui, 2016). Other studies argued that benefits can be even greater than those provided by renaturing such areas, as, for example, in GHG mitigation (Evans et al., 2015). This study therefore highlights that the improved provision of ES may be an important argument to sustain the management of marginal agricultural lands for bioenergy production, as part of a multifunctional landscape. This result is in line with the findings of Baumber (2017), who suggested the adoption of integrated policies to enhance ES while supporting the cultivation of perennial energy crops. By doing so, the implementation of existing mechanisms offers different opportunities to meet this goal. For instance, SRC has been recognized among the options to establish ecological focus areas as part of the ‘greening’ measures of the Common Agricultural Policy. The rising interest in market-based instruments, such as the Payment for Ecosystem Services (PES) schemes, can provide additional opportunities for enhancing bioenergy production in synergy with the provision of other ES. The put in place of PES schemes has a promising potential in moving feedstock producers towards more sustainable practices, mainly concerning perennial crops. For instance, Jager and Efroymson (2018) found that PES schemes could drive producers in choosing feedstocks according to their resistance or resilience to disturbance, in order to increase reliability in ES provision over time. However, PES schemes still suffer for a lack of clear and consistent normative references (Pan et al., 2017; Mauerhofer, 2018), as well as for other barriers such as the conflictual role of large-scale companies (Chinangwa et al., 2017). Nevertheless, the methodology applied in this study showed the potential of remote sensing and multi-source data analysis to support and inform decision making and planning processes based on an ES approach. Its application may vary among a broad set of territorial, environmental, and sectoral planning policies, such as for the planning of a sustainable bioenergy supply chain, which is our case. 4.2. Methodological advances and limits Our methodology is based on the integration of geo-spatial information such as field boundaries, LULC map, and multi-temporal satellite data from free and open source archives. In particular, the latter data source can be used to derive spatially-explicit information for seasonal crops at the parcel level. This input data is a prerequisite to export this approach in other contexts. Temporal analysis of agricultural land management and productivity at the farm/local level or e even better e at the management unit level (i.e., parcel-scale) is a novel contribution for the identification of marginal agricultural lands. In fact, marginal agricultural lands are usually identified on the basis of land physical conditions that can limit cropping activities (e.g., using the Land Capability Classification developed by Klingebiel and Montgomery, 1961). However, such approach does not take into account i) potential agro-techniques that may have been put in place to overcome specific physical and environmental constraints, ii) the farmers’ individual attitudes to cultivate the land, or iii) socio economic drivers that make (in)convenient the crop production (Pulighe et al., 2019). The method here proposed can contribute to identify and map ongoing processing related to land use (e.g., crop rotation, land management, and related productivity) and land use changes (e.g., hints of abandonment). Archives of satellite data allowed to perform retrospective analysis and generate thematic maps that are usually not available or not updated as required by specific studies, such as land use planning studies. The replicability of such analysis in other case studies is ensured by the intrinsic characteristics of satellite data: large spatial coverage, timely

availability, temporal continuity, and free access. For the historical reconstruction of past land management, satellite data, such as the ones from the Landsat missions, are available since late 80ies, although cloud contamination can lower down the theoretical revisiting cycle of 16 days in some cases. In our methodology, the mapping approach was supervised by an expert in order to i) select the appropriate satellite data, ii) identify SAVI thresholds to perform the classification, and iii) refine the classification results. However, it has to be noted that these actions based on a human operator, especially regarding the second and the third point, can be time consuming, hence reducing the repeatability of the whole approach in other contexts in case of time constraints. As regards the first point, we selected and analyzed satellite data for both summer and winter crop seasons, exploiting multi-seasonal images that are a fundamental source of information to produce accurate classification (Prishchepov et al., 2012). To implement the approach, a good knowledge on the crop calendar occurring in the study area is needed in order to select and collect the appropriate satellite dataset. Such information can be easily found through expert knowledge or in grey literature. Concerning the second and the third point, alternative approaches that reduce the need of human intervention and speed up the process can be used to directly map crop presence and rotation by exploiting more advanced machine learning techniques when ground truth data are available (Azar et al., 2016; Villa et al., 2015). A similar approach, based on supervised classification method, is operationally used by ARPA Emilia Romagna to produce an early season crop map to estimate irrigation needs of summer crops (https://www.arpae.it/ dettaglio_generale.asp?id¼2824&idlivello¼1599, accessed 26 March 2019). As regards satellite data available for crop mapping and monitoring, the new era of Sentinel missions (in the framework of the EU Copernicus program, https://www.copernicus.eu/en, accessed 26 March 2019) can provide multispectral images with a spatial resolution of 10 m and a revisiting cycle of 5 days since 2015. Such new data can strengthen the crop analysis at the farm level, even in condition of parcels with low size, thanks to the finer resolution that better track the crop spatial and temporal variability. In addition, they can be further exploited for urban vegetation studies, which in general need data with a finer spatial resolution (e.g., Maragno et al., 2018). Open source software platforms, such as the Sen2agri software, are now available for a full exploitation of the enormous amount of information available in S2 data (http://www. esa-sen2agri.org/, accessed 26 March 2019; Defourny et al., 2019). This platform can be used to generate both crop maps and a “biophysical indicator” which can provide quantitative information on the vegetation status useful to better define crop production. The method proposed in this paper can therefore further benefit of such new data and tools that can faster the operator's work. Moreover, our three-year period analysis could be further extended on a longer time series (e.g., 5 years) to better track potential lack of management and abandonment (Pointereau et al., 2008; Estel et al., 2015). As regards the trade-off and synergy analysis between wood BBES and other ES, SAVI data have been exploited to map ES. VIs summed over the growth season have positive relationship with the fraction of absorbed photosynthetically active radiation and, thus, are used as a proxy for net primary production (Fensholt et al., 2009; de Jong et al., 2011; Wessels et al., 2012), which represents the amount of energy used by plants for storage, growth, and reproduction (Richmond et al., 2007). Since net primary production can be considered as a flow that maintains the stock of natural capital that generates ES, it is positively correlated with the flow of many provisioning (e.g., biomass provision) and regulating services (e.g., carbon sequestration) (Richmond et al., 2007). Assuming that

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

to higher VIs values it corresponds a higher amount of net primary production, the provision of related ES can be assessed, hence allowing for comparison when using land use scenarios. The temporal dimension provided by satellite data is a fundamental factor to take into account when selecting and using such proxy indicators for ES provided by vegetation, especially when mapping seasonal vegetation of crops. However, ES provided by vegetation may also depend on other elements not directly detectable by satellite data, such as vegetation height and canopy structure. Moreover, ES studies can today also benefit from the availability of some estimated biophysical variables such as LAI (Leaf Area index) or fAPAR (fraction of Absorbed Photosynthetic Active Radiation) that can be directly exploited for yield estimation (Lobell et al., 2003; Pagani et al., 2018), thus providing a fine-tuned quantitative information on crop phenology. These aspects should be further investigated together with the full exploitation of temporal dimension of remote-sensing-based indicators for the mapping and assessment of ES. Finally, the LULC data refined on cadastral parcels used in this study, which allowed to perform the analysis at the management unit level, is another requirement that could limit the implementation and replicability of our approach in other case studies if this is not available. Alternatively, field boundaries can be provided by sub-national authorities or can be derived directly from remote sensing data. Eventually, if LULC data are missing, both pixel-based and object-based image analysis approaches can be used as an alternative to derive this fundamental information. 5. Conclusion This paper presented a method, based on the analysis of multitemporal remote sensing data, for the identification of marginal agricultural lands to be potentially destined to biomass production from 2G wood crops in the province of Rovigo (northern Italy). Understanding the energy potential that can be obtained from the cultivation of 2G wood crops in marginal agricultural lands, as well as the potential consequences of such land use change on other ES (especially food provision), is a crucial step to design an effective and sustainable energy supply chain. This case study provides useful spatially-explicit information to be applied for planning a sustainable regional bioenergy supply chain, by i) identifying marginal agricultural lands in order to avoid the conflict with food production, ii) quantifying the potential energy contribution from the cultivation of 2G wood crops in them, and iii) by using an ES approach in order to highlight potential synergies and tradeoffs. Such approach can support and inform decision-making in land use and energy planning at the landscape scale. In most cases, cultivating 2G wood crops in marginal agricultural lands led to an increase in the provision of several ES. However, in order to meet the projected energy demand of the agricultural sector, as well as to cover the energy potential authorized for the province of Rovigo, while avoiding any trade-off with food production, the wood bioenergy supply chain must be integrated with other types of biomass feedstocks. Our study demonstrated that the temporal analysis of agricultural land management and productivity at the management unit level (i.e., parcel scale) is fundamental for the identification of marginal agricultural lands, which marginality may depend on the farmers’ choices and individual attitudes to cultivate the land as a consequence of both environmental and socio-economic factors. Multi-annual and multi-seasonal satellite images demonstrated to be a very useful tool, providing relevant proxy data for both the identification of marginal agricultural lands, which are in connection with food provision service, and the mapping and assessment of different (provisioning and regulating) ES. Further analysis might include also the assessment of cultural

13

services. However, we recommend to use more advanced machine learning techniques since the method proposed is pretty timeconsuming. This would reduce the need of human intervention for the exploitation of satellite data. Moreover, the use of new Sentinel data (when available) will allow for a more detailed analysis, since images have a finer resolution compared to the Landsat ones. As shown, the ES perspective adopted through the analysis and mapping of ES trade-offs and synergies is a powerful tool that allows a better understanding of sustainable requirements for the use of biomass sources for energy purposes. In particular, it can support the implementation and communication to stakeholders of sustainable policies and programs at local and regional levels and can be included in the framework of bioenergy certification schemes to better promote the environmental sustainability of this sector. Declarations of interest None. Funding This study was supported by the European Social Funds 2014e2020 for the Regione del Veneto, Project no. 2122-14-21212015, entitled “Innovative method for the analysis of biomass-based energy potential through remote sensing applied to Ecosystem Services assessment for business strategies”. We would like to thank the business partners involved in this study - San Marco Bioenergie S.p.A., Gemmlab s.r.l., SMA s.r.l. Green and Smart Solutions and Punto Confindustria SIVE Business Unit Education e Formazione for the kind support in the research activities. Abbreviations 2G BBES ES GHG LPIS LULC PES SAVI SAVI AVG SRC VIs

Second Generation Biomass-Based Energy Sources Ecosystem Services Greenhouse Gases Land Parcel Information Systems Land Use/Land Cover Payment for Ecosystem Services Soil-Adjusted Vegetation Index SAVI Average Short Rotation Coppice Vegetation Indices

Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.jclepro.2019.117672. References Ahmed, N., Englund, J.E., Åhman, I., Lieberg, M., Johansson, E., 2011. Perception of pesticide use by farmers and neighbors in two periurban areas. Sci. Total Environ. 412, 77e86. https://doi.org/10.1016/j.scitotenv.2011.10.022. Alcantara, C., Kuemmerle, T., Prishchepov, A.V., Radeloff, V.C., 2012. Mapping abandoned agriculture with multi-temporal MODIS satellite data. Remote Sens. Environ. 124, 334e347. https://doi.org/10.1016/j.rse.2012.05.019. Andrew, M.E., Wulder, M.A., Nelson, T.A., 2014. Potential contributions of remote sensing to ecosystem service assessments. Prog. Phys. Geogr. 38, 328e353. https://doi.org/10.1177/0309133314528942. de Araujo Barbosa, C.C., Atkinson, P.M., Dearing, J.A., 2015. Remote sensing of ecosystem services: a systematic review. Ecol. Indicat. 52, 430e443. https://doi. org/10.1016/j.ecolind.2015.01.007. ARPAV, 2010. Colture energetiche e protezione del suolo. http://www.arpa.veneto. it/temi-ambientali/suolo/file-e-allegati/documenti/carta-dei-suoli/Colture% 20energetiche%20e%20protezione%20del%20suolo_2010.pdf. (Accessed 4 April

14

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672

2019). Atzberger, C., 2013. Advances in remote sensing of agriculture: context description, existing operational monitoring systems and major information needs. Remote Sens. 5 (2), 949e981. https://doi.org/10.3390/rs5020949. Ayanu, Y.Z., Conrad, C., Nauss, T., Wegmann, M., Koellner, T., 2012. Quantifying and mapping ecosystem services supplies and demands: a review of remote sensing applications. Environ. Sci. Technol. 46 (16), 8529e8541. https://doi.org/10.1021/ es300157u. Azar, R., Villa, P., Stroppiana, D., Crema, A., Boschetti, M., Brivio, P.A., 2016. Assessing in-season crop classification performance using satellite data: a test case in Northern Italy. Eur. J. Remote Sens. 49, 361e380. https://doi.org/10.5721/ EuJRS20164920. Barbosa, P.M., Gregoire, J.M., Pereira, J.M.C., 1999. An algorithm for extracting burned areas from time series of AVHRR GAC data applied at a continental scale. Remote Sens. Environ. 69, 253e263. https://doi.org/10.1016/S0034-4257(99) 00026-7. Bartle, J.R., Abadi, A., 2010. Toward sustainable production of second generation bioenergy feedstocks. Energy Fuels 24, 2e9. https://doi.org/10.1021/ef9006438. Baumber, A., 2017. Enhancing ecosystem services through targeted bioenergy support policies. Ecosyst. Serv. 26, 98e110. https://doi.org/10.1016/j.ecoser.2017. 06.012. Benedetti, R., Rossini, P., 1993. On the use of NDVI profiles as a tool for agricultural statistics: the case study of wheat yield estimate and forecast in Emilia Romagna. Remote Sens. Environ. 45 (3), 311e326. https://doi.org/10.1016/00344257(93)90113-C. Blanco-Canqui, H., 2016. Growing dedicated energy crops on marginal lands and ecosystem services. Soil Sci. Soc. Am. J. 80, 846e858. https://doi.org/10.2136/ sssaj2016.03.0080. Boschetti, M., Bocchi, S., Brivio, P.A., 2007. Assessment of pasture production in the Italian Alps using spectrometric and remote sensing information. Agric. Ecosyst. Environ. 118 (1e4), 267e272. https://doi.org/10.1016/j.agee.2006.05.024. Boschetti, M., Stroppiana, D., Brivio, P.A., 2010. Mapping burned areas in a Mediterranean environment using soft integration of Spectral Indices from highresolution satellite images. Earth Interact. 14 (17), 1e20. https://doi.org/10. 1175/2010EI349.1. Boschetti, M., Busetto, L., Manfron, G., Laborte, A., Asilo, S., Pazhanivelan, S., Nelson, A., 2017. PhenoRice: a method for automatic extraction of spatiotemporal information on rice crops using satellite data time series. Remote Sens. Environ. 194, 347e365. https://doi.org/10.1016/j.rse.2017.03.029. Chinangwa, L., Gasparatos, A., Saito, O., 2017. Forest conservation and the private sector: stakeholder perceptions towards payment for ecosystem service schemes in the tobacco and sugarcane sectors in Malawi. Sustain. Sci. 12 (5), 727e746. https://doi.org/10.1007/s11625-017-0469-6. Conrad, C., Dech, S., Dubovyk, O., Fritsch, S., Klein, D., Low, F., Schorcht, G., Zeidler, J., 2014. Derivation of temporal windows for accurate crop discrimination in heterogeneous croplands of Uzbekistan using multitemporal RapidEye images. Comput. Electron. Agric. 103, 63e74. https://doi.org/10.1016/j.compag.2014.02. 003. Defourny, P., Bontemps, S., Bellemans, N., Cara, C., Dedieu, G., Guzzonato, E., Hagolle, O., Inglada, J., Nicola, L., Rabaute, T., Savinaud, M., Udroiu, C., Valero, S., gue , A., Dejoux, J.-F., El Harti, A., Ezzahar, J., Kussul, N., Labbassi, K., Be Lebourgeois, V., Miao, Z., Newby, T., Nyamugama, A., Salh, N., Shelestov, A., Simonneaux, V., Traore, P.S., Traore, S.S., Koetz, B., 2019. 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. Environ. 221, 551e568. https://doi.org/10.1016/j.rse. 2018.11.007. Downing, T.E., Lüdeke, M., Reynolds, J.F., Stafford Smith, D.M., 2002. International desertification, social geographies of vulnerability and adaption. In: Reynolds, J.F., Stafford Smith, D.M. (Eds.), Global Desertification: Do Humans Cause Deserts? Dahlem University Press, pp. 233e252. EC, 2009a. Directive 2009/28/EC of the European Parliament and of the Council of 23 April 2009 on the Promotion of the Use of Energy from Renewable Sources and Amending and Subsequently Repealing Directives 2001/77/EC and 2003/ 30/EC. EC, 2009b. Council Regulation (EC) No 73/2009 of 19 January 2009 Establishing Common Rules for Direct Support Schemes for Farmers under the Common Agricultural Policy and Establishing Certain Support Schemes for Farmers, Amending Regulations (EC) No 1290/2005, (EC) No 247/2006, (EC) No 378/2007 and Repealing Regulation (EC) No 1782/2003. Edrisi, S.A., Abhilash, P.C., 2016. Exploring marginal and degraded lands for biomass and bioenergy production: an Indian scenario. Renew. Sustain. Energy Rev. 54, 1537e1551. https://doi.org/10.1016/j.rser.2015.10.050. Egoh, B., Rouget, M., Reyers, B., Knight, A.T., Cowling, R.M., van Jaarsveld, A.S., Welz, A., 2007. Integrating ecosystem services into conservation assessments: a review. Ecol. Econ. 63 (4), 714e721. https://doi.org/10.1016/j.ecolecon.2007.04. 007. Estel, S., Kuemmerle, T., Alc antara, C., Levers, C., Prishchepovd, A., Hostert, P., 2015. Mapping farmland abandonment and recultivation across Europe using MODIS NDVI time series. Remote Sens. Environ. 163, 312e325. https://doi.org/10.1016/j. rse.2015.03.028. Evans, S.G., Ramage, B.S., DiRocco, T.L., Potts, M.D., 2015. Greenhouse gas mitigation on marginal land: a quantitative review of the relative benefits of forest recovery versus biofuel production. Environ. Sci. Technol. 49 (4), 2503e2511. https://doi.org/10.1021/es502374f.

Fathollahi-Fard, A.M., Hajiaghaei-Keshteli, M., 2016. Red Deer Algorithm (RDA); a new optimization algorithm inspired by Red Deers' mating. In: Proceedings of 12th International Conference on Industrial Engineering, IEEE, pp. 33e34. Fathollahi-Fard, A.M., Hajiaghaei-Keshteli, M., Tavakkoli-Moghaddam, R., 2018a. The social engineering optimizer (SEO). Eng. Appl. Artif. Intell. 72, 267e293. https:// doi.org/10.1016/j.engappai.2018.04.009. Fathollahi-Fard, A.M., Hajiaghaei-Keshteli, M., Tavakkoli-Moghaddam, R., 2018b. A bi-objective green home health care routing problem. J. Clean. Prod. 200, 423e443. https://doi.org/10.1016/j.jclepro.2018.07.258. Feng, X., Fu, B., Yang, X., Lü, Y., 2010. Remote sensing of ecosystem services: an opportunity for spatially explicit assessment. Chin. Geogr. Sci. 20 (6), 522e535. https://doi.org/10.1007/s11769-010-0428-y. Fensholt, R., Rasmussen, K., Nielsen, T.T., Mbow, C., 2009. Evaluation of earth observationbased long term vegetation trendsdintercomparing NDVI time series trend analysis consistency of Sahel from AVHRR GIMMS, Terra MODIS and SPOT VGT data. Remote Sens. Environ. 113, 1886e1898. https://doi.org/10. 1016/j.rse.2009.04.004. Fiorese, G., Guariso, G., 2010. A GIS-based approach to evaluate biomass potential from energy crops at regional scale. Environ. Model. Softw 25, 702e711. https:// doi.org/10.1016/j.envsoft.2009.11.008. Gaglio, M., Aschonitis, V.G., Mancuso, M.M., Reyes Puig, J.P., Moscoso, F., Castaldelli, G., Fano, E.A., 2017. Changes in land use and ecosystem services in tropical forest areas: a case study in Andes mountains of Ecuador. Int. J. Biodivers. Sci. Ecosyst. Serv. Manag. 13 (1), 264e279. https://doi.org/10.1080/ 21513732.2017.1345980. Gaglio, M., Tamburini, E., Lucchesi, F., Aschonitis, V., Atti, A., Castaldelli, G., Fano, E.A., 2019. Life cycle assessment of maize-germ oil production and the use of bioenergy to mitigate environmental impacts: a gate-to-gate case study. Resources 8 (2), 60. https://doi.org/10.3390/resources8020060. Gissi, E., Siciliano, G., Reho, M., 2011. Biomass production and land use management in the Italian context: regulations, conflicts, and impacts. In: 51st Congress of the European Regional Science Association: "New Challenges for European Regions and Urban Areas in a Globalised World", 30 August - 3 September 2011, Barcelona, Spain. European Regional Science Association (ERSA), Louvain-laNeuve. https://www.econstor.eu/bitstream/10419/120273/1/ERSA2011_1413. pdf. (Accessed 4 April 2019). Gissi, E., Gaglio, M., Reho, M., 2014. Trade-off between carbon storage and biomassbased energy sources ecosystem services, the case study from the province of Rovigo (Italy). Ann. Bot. (Rome) 4, 73e81. https://doi.org/10.4462/annbotrm11814. Gissi, E., Gaglio, M., Reho, M., 2016. Sustainable energy potential from biomass through ecosystem services trade-off analysis: the case of the Province of Rovigo (Northern Italy). Ecosyst. Serv. 18, 1e19. https://doi.org/10.1016/j.ecoser. 2016.01.004. Gissi, E., Gaglio, M., Aschonitis, V., Fano, E.A., Reho, M., 2018. Soil-related ecosystem services trade-off analysis for sustainable biodiesel production. Biomass Bioenergy 114, 83e89. https://doi.org/10.1016/j.biombioe.2017.08.028. Gissi, E., Garramone, V., 2018. Learning on ecosystem services co-production in decision-making from role-playing simulation: comparative analysis from Southeast Europe. Ecosyst. Serv. 34 (B), 228e253. https://doi.org/10.1016/j. ecoser.2018.03.025. Gr adinaru, S.R., Kienast, F., Psomas, A., 2019. Using multi-seasonal Landsat imagery for rapid identification of abandoned land in areas affected by urban sprawl. Ecol. Indicat. 96 (2), 79e86. https://doi.org/10.1016/j.ecolind.2017.06.022. Haines-Young, R., Potschin, M., 2011. Common International Classification of Ecosystem Services (CICES): 2011 Update. Report to the European Environmental Agency, Nottingham. Hastings, A., Clifton-Brown, J., Wattenbach, M., Mitchell, C., Smith, P., 2009a. The development of miscanfor, a new Miscanthus crop growth model: towards more robust yield predictions under different climatic and soil conditions. GCB Bioenergy 1, 154e170. https://doi.org/10.1111/j.1757-1707.2009.01007.x. Hastings, A., Clifton-Brown, J., Wattenbach, M., Mitchell, C., Stampfl, P., Smith, P., 2009b. Future energy potential of Miscanthus in europe. GCB Bioenergy 1, 180e196. https://doi.org/10.1111/j.1757-1707.2009.01012.x. Hastings, A., Tallis, M.J., Casella, E., Matthews, R.W., Henshall, P.A., Milner, S., Smith, P., Taylor, G., 2014. The technical potential of Great Britain to produce ligno-cellulosic biomass for bioenergy in current and future climates. GCB Bioenergy 6, 108e122. https://doi.org/10.1111/gcbb.12103. Holland, R.A., Eigenbrod, F., Muggeridge, A., Brown, G., Clarke, D., Taylor, G., 2015. A synthesis of the ecosystem services impact of second generation bioenergy crop production. Renew. Sustain. Energy Rev. 46, 30e40. https://doi.org/10. 1016/j.rser.2015.02.003. Huete, A.R., 1988. A soil-adjusted vegetation index (SAVI). Remote Sens. Environ. 25, 295e309. https://doi.org/10.1016/0034-4257(88)90106-X. Jager, H.I., Efroymson, R.A., 2018. Can upstream biofuel production increase the flow of downstream ecosystem goods and services? Biomass Bioenergy 114, 125e131. https://doi.org/10.1016/j.biombioe.2017.08.027. de Jong, R., de Bruin, S., de Wit, A., Schaepman, M.E., Dent, D.L., 2011. Analysis of monotonic greening and browning trends from global NDVI time-series. Remote Sens. Environ. 115, 692e702. https://doi.org/10.1016/j.rse.2010.10.011. Kang, S., Post, W.M., Nichols, J.A., Wang, D., West, T.O., Bandaru, V., Izaurralde, R.C., 2013. Marginal lands: concept, assessment and management. J. Agric. Sci. 5 (5). https://doi.org/10.5539/jas.v5n5p129. Klingebiel, L.A.A., Montgomery, P.H., 1961. Land Capability Classification. USDA Handbook 210. USDA, Washington, DC.

D. Longato et al. / Journal of Cleaner Production 237 (2019) 117672 Lobell, D.B., Asner, G.P., Ortiz-Monasterio, J.I., Benning, T.L., 2003. Remote sensing of regional crop production in the Yaqui Valley, Mexico: estimates and uncertainties. Agric. Ecosyst. Environ. 94 (2), 205e220. https://doi.org/10.1016/ S0167-8809(02)00021-X. Low, F., Michel, U., Dech, S., Conrad, C., 2013. Impact of feature selection on the accuracy and spatial uncertainty of per-field crop classification using support vector machines. ISPRS J. Photogramm. 85, 102e119. https://doi.org/10.1016/j. isprsjprs.2013.08.007. Low, F., Fliemann, E., Abdullaev, I., Conrad, C., Lamers, J.P.A., 2015. Mapping abandoned agricultural land in Kyzyl-Orda, Kazakhstan using satellite remote sensing. Appl. Geogr. 62, 377e390. https://doi.org/10.1016/j.apgeog.2015.05. 009. Malladi, K.T., Sowlati, T., 2017. Optimization of operational level transportation planning in forestry: a review. Int. J. For. Eng. 28. https://doi.org/10.1080/ 14942119.2017.1362825. Maragno, D., Gaglio, M., Robbi, M., Appiotti, F., Fano, E.A., Gissi, E., 2018. Fine-scale analysis of urban flooding reduction from green infrastructure: An ecosystem services approach for the management of water flows. Ecol. Model. 386, 1e10. https://doi.org/10.1016/j.ecolmodel.2018.08.002. Mauerhofer, V., 2018. Legal aspects of ecosystem services: an introduction and an overview. Ecosyst. Serv. 29, 185e189. https://doi.org/10.1016/j.ecoser.2017.11. 002. Milner, S., Holland, R.A., Lovett, A., Sunnenberg, G., Hastings, A., Smith, P., Wang, S., Taylor, G., 2016. Potential impacts on ecosystem services of land use transitions to second-generation bioenergy crops in GB. GCB Bioenergy 8, 317e333. https:// doi.org/10.1111/gcbb.12263. Noon, C.E., Daly, M.J., 1996. GIS-based biomass resource assessment with BRAVO. Biomass Bioenergy 10, 101e109. https://doi.org/10.1016/0961-9534(95)000658. Pagani, V., Guarneri, T., Busetto, L., Ranghetti, L., Boschetti, M., Movedi, E., CamposTaberner, M., Garcia-Haro, F.J., Katsantonis, D., Stavrakoudis, D., Ricciardelli, E., Romano, F., Holecz, F., Collivignarelli, F., Granell, C., Casteleyn, S., Confalonieri, R., 2018. A high-resolution, integrated system for rice yield forecasting at district level. Agric. Syst. 168, 181e190. https://doi.org/10.1016/j.agsy. 2018.05.007. Pan, X., Xu, L., Yang, Z., Yu, B., 2017. Payments for ecosystem services in China: policy, practice, and progress. J. Clean. Prod. 158, 200e208. https://doi.org/10. 1016/j.jclepro.2017.04.127. Pascucci, S., Carfora, M.F., Palombo, A., Pignatti, S., Casa, R., Pepe, M., Castaldi, F., 2018. A comparison between standard and functional clustering methodologies: application to agricultural fields for yield pattern assessment. Remote Sens. 10 (4), 585. https://doi.org/10.3390/rs10040585. Pettorelli, N., 2013. The Normalized Differential Vegetation Index. Oxford University Press, Oxford, UK. Pettorelli, N., Laurance, W.F., O'Brien, T.G., Wegmann, M., Nagendra, H., Turner, W., 2014. Satellite remote sensing for applied ecologists: opportunities and challenges. J. Appl. Ecol. 51, 839e848. https://doi.org/10.1111/1365-2664.12261. Pointereau, P., Coulon, F., Girard, P., Lambotte, M., Stuczynski, T., Sanchez Ortega, V., Del Rio, A., 2008. Analysis of farmland abandonment and the extent and location of agricultural areas that are actually abandoned or are in risk to be abandoned. In: Anguiano, E., Bamps, C., Terres, J.-M. (Eds.), JRC Scientific and Technical Reports (EUR 23411 EN). ri, M., 2014. The effect of bioenergy Popp, J., Lakner, Z., Harangi-R akos, M., Fa expansion: food, energy, and environment. Renew. Sustain. Energy Rev. 32, 559e578. https://doi.org/10.1016/j.rser.2014.01.056. Power, A.G., 2010. Ecosystem services and agriculture: tradeoffs and synergies. Phil. Trans. R. Soc. B 365, 2959e2971. https://doi:10.1098/rstb.2010.0143. Prishchepov, A.V., Radeloff, V.C., Dubinin, M., Alcantara, C., 2012. The effect of Landsat ETM/ETMþ image acquisition dates on the detection of agricultural land abandonment in Eastern Europe. Remote Sens. Environ. 126, 195e209. https://doi.org/10.1016/j.rse.2012.08.017. Provincia di Rovigo, 2010. Programma Energetico Provinciale. http://cdn1.regione. veneto.it/alfstreaming-servlet/streamer/resourceId/df725670-7c02-442b-aa8f8ca6bf6baad2/piano_energetico.pdf. (Accessed 3 April 2019). Pulighe, G., Bonati, G., Colangeli, M., Morese, M.M., Traverso, L., Lupia, F., Khawaja, C., Janssen, R., Fava, F., 2019. Ongoing and emerging issues for sustainable bioenergy production on marginal lands in the Mediterranean regions. Renew. Sustain. Energy Rev. 103, 58e70. https://doi.org/10.1016/j.rser.2018.12. 043. Regione del Veneto, 2015. Programma di Sviluppo Rurale per il Veneto 2014-2020. Approvato ai sensi del Regolamento (UE) n. 1305/2013 del Parlamento Europeo e del Consiglio. Decisione di esecuzione Commissione C(2015) 3482 del 26.5.2015. Regolamenti (UE) n. 1303/2013 e n. 1305/2013. Deliberazione/CR n. 71 del 10/06/2014. DELIBERAZIONE DELLA GIUNTA REGIONALE n. 947 del 28 luglio 2015. Rembold, F., Atzberger, C., Savin, I., Rojas, O., 2013. Using low resolution satellite imagery for yield prediction and yield anomaly detection. Remote Sens. 5 (4),

15

1704e1733. https://doi.org/10.3390/rs5041704. Richmond, A., Kaufmann, R.K., Myneni, R.B., 2007. Valuing ecosystem services: a shadow price for net primary production. Ecol. Econ. 64, 454e462. https://doi. org/10.1016/j.ecolecon.2007.03.009. Sahebjamnia, N., Fathollahi-Fard, A.M., Hajiaghaei-Keshteli, M., 2018. Sustainable tire closed-loop supply chain network design: hybrid metaheuristic algorithms for large-scale networks. J. Clean. Prod. 196, 273e296. https://doi.org/10.1016/j. jclepro.2018.05.245. Sallustio, L., Pettenella, D., Merlini, P., Romano, R., Salvati, L., Marchetti, M., Corona, P., 2018. Assessing the economic marginality of agricultural lands in Italy to support land use planning. Land Use Policy 76, 526e534. https://doi. org/10.1016/j.landusepol.2018.02.033. Samela, C., Troy, T.J., Manfreda, S., 2017. Geomorphic classifiers for flood-prone areas delineation for data-scarce environments. Adv. Water Resour. 102, 13e28. https://doi.org/10.1016/j.advwatres.2017.01.007. Samela, C., Albano, R., Sole, A., Manfreda, S., 2018. A GIS tool for cost-effective delineation of flood-prone areas. Comput. Environ. Urban Syst. 70, 43e52. https://doi.org/10.1016/j.compenvurbsys.2018.01.013. Sarker, B.R., Wu, B., Paudel, K.P., 2019. Modeling and optimization of a supply chain of renewable biomass and biogas: processing plant location. Appl. Energy 239, 343e355. https://doi.org/10.1016/j.apenergy.2019.01.216. Semeraro, T., Mastroleo, G., Pomes, A., Luvisi, A., Gissi, E., Aretano, R., 2019. Modelling fuzzy combination of remote sensing vegetation index for durum wheat crop analysis. Comput. Electron. Agric. 156, 684e692. https://doi.org/10. 1016/j.compag.2018.12.027. Sims, R.E.H., Mabee, W., Saddler, J.N., Taylor, M., 2010. An overview of second generation biofuel technologies. Bioresour. Technol. 101, 1570e1580. https://doi. org/10.1016/j.biortech.2009.11.046. Smith, A.M.S., Drake, N.A., Wooster, M.J., Hudak, A.T., Holden, Z.A., Gibbons, C.J., 2007. Production of Landsat ETM1 reference imagery of burned areas within southern African savannahs: comparison of methods and application to MODIS. Int. J. Remote Sens. 28, 2753e2775. https://doi.org/10.1080/ 01431160600954704. Soldatos, P., 2015. Economic aspects of bioenergy production from perennial grasses in marginal lands of south europe. Bioenerg. Res. 8, 1562e1573. https://doi.org/ 10.1007/s12155-015-9678-y. Stefanski, J., Kuemmerle, T., Chaskovskyy, O., Griffiths, P., Havryluk, V., Knorn, J., Korol, N., Sieber, A., Wastke, B., 2014. Mapping land management regimes in western Ukraine using optical and SAR data. Remote Sens-Basel 6, 5279e5305. https://doi.org/10.3390/rs6065279. Tilman, D., Socolow, R., Foley, J.A., Hill, J., Larson, E., Lynd, L., Pacala, S., Reilly, J., Searchinger, T., Somerville, C., Williams, R., 2009. Beneficial biofuels e the food, energy, and environment trilemma. Science 325, 270e271. https://doi.org/10. 1126/science.1177970. Valentine, J., Clifton-Brown, J., Hastings, A., Robson, P., Allison, G., Smith, P., 2012. Food vs. Fuel: the use of land for lignocellulosic ‘next generation’ energy crops that minimize competition with primary food production. GCB Bioenergy 4, 1e19. https://doi.org/10.1111/j.1757-1707.2011.01111.x. Verma, M., Friedl, M.A., Richardson, A.D., Kiely, G., Cescatti, A., Law, B.E., Wohlfahrt, G., Gielen, B., Roupsard, O., Moors, E.J., Toscano, P., Vaccari, F.P., Gianelle, D., Bohrer, G., Varlagin, A., Buchmann, N., Van Gorsel, E., Montagnani, L., Propastin, P., 2013. Remote sensing of annual terrestrial gross primary productivity from MODIS: an assessment using the FLUXNET La Thuile data set. Biogeosciences 10, 11627e11669. https://doi.org/10.5194/bgd-1011627-2013. Villa, P., Stroppiana, D., Fontanelli, G., Azar, R., Brivio, P.A., 2015. In-season mapping of crop type with optical and X-band SAR data: a classification tree approach using synoptic seasonal features. Remote Sens. 7, 12859e12886. https://doi.org/ 10.3390/rs71012859. € ttcher, K., Carrara, P., Bordogna, G., Brivio, P.A., 2011. Weissteiner, C., Boschetti, M., Bo Spatial explicit assessment of rural land abandonment in the Mediterranean area. Glob. Planet. Chang. 79, 20e36. https://doi.org/10.1016/j.gloplacha.2011. 07.009. Wessels, K.J., van den Bergh, F., Scholes, R.J., 2012. Limits to detectability of land degradation by trend analysis of vegetation index data. Remote Sens. Environ. 125, 10e22. https://doi.org/10.1016/j.rse.2012.06.022. Zhang, F., Johnson, D.M., Wang, J., 2015. Life-cycle energy and GHG emissions of forest biomass harvest and transport for biofuel production in Michigan. Energies 8 (4), 3258e3271. https://doi.org/10.3390/en8043258. Zhang, W., Brandt, M., Tong, X., Tian, Q., Fensholt, R., 2018. Impacts of the seasonal distribution of rainfall on vegetation productivity across the Sahel. Biogeosciences 15, 319e330. https://doi.org/10.5194/bg-15-319-2018. Zurlini, G., Petrosillo, I., Aretano, R., Castorini, I., D'Arpa, S., De Marco, A., Pasimeni, M.R., Semeraro, T., Zaccarelli, N., 2014. Key fundamental aspects for mapping and assessing ecosystem services: predictability of ecosystem service providers at scales from local to global. Ann. Bot. 4, 53e63. https://doi.org/10. 4462/annbotrm-11754.