Mapping optimal biorefinery plant locations using Earth observation data in the Mediterranean region

Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Climate, Infrastructure and Environment Executive Agency. Neither the European Union nor the granting authority can be held responsible for them. Mapping optimal biorefinery plant locations using Earth observation data in the Mediterranean region Erico Kutchartt, Francesco Pirotti, José Salgado-Rojas, Álvaro Cortés-Molino, Núria Aquilué, Neus Puy Earth Sensing Summer School 2025 San Vito di Cadore, September 8, 2025 www.pysolo.eu Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 2 Introduction •Through thermo-chemical processes, forest biomass can be converted into versatile solid, liquid,or gaseous biofuel forms and high-value chemicals. •Solar irradiation can be used for these thermochemical proccesses, reinforcing the concept of green renewable energies. •Agood balance between solar-power energy and forest biomass resources must be found to establish biorefinery plants in southern Europe. •Forest biomass can be converted into bioenergy products and strengthen the circular bioeconomy in rural areas, providing renewable energy for industrial, commercial, and domestic use. www.pysolo.eu 3Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 Main goals •Identify variables that can be considered impractical for extracting and storing forest biomass, and then apply restrictions based on these limitations. •Develop a multi-criteria decision support system analysis based on vegetation and accessibility features, weighting them in different scenarios. •Compute the ideal number and localization of biorefinery plants using the simulated annealing algorithm combined with a cost-function analysis. General Identify the most suitable areas regarding forest biomass availability and solar irradiation to establish biorefinery plants in the Mediterranean region. Specific objectives www.pysolo.eu 4Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 General overview Exist a relationship between forest biomass and solar irradiation? Answer: there is a trend, but with a low fitting www.pysolo.eu 5Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 Pirotti et al. (2023) Santoro et al. (2022) There are several biomass datasets at pan-European and global scale What dataset is more convenient in our analysis? www.pysolo.eu 6Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 Methodology Constraints Multi-Criteria Analysis (MCA) Two-step approaches: limitations and weights www.pysolo.eu 7Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 Normalisation of variables from 0 to 1 using the ramp function (Spain) www.pysolo.eu 8Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 Minimum and maximum values based on the 1º and 99º percentile Country AGB availability (Mg/ha) Net primary productivity (kgC/ha/year) Forest cover (%) Slope terrain (%) Road proximity (log)* Spain 5.30 - 215.50 3962 -12789 50.43 - 97.47 3.74 - 49.39 1.00 - 3.22 Italy 8.69 - 296.55 4351 -12777 50.97 - 98.48 6.14 - 49.57 1.00 - 3.07 Greece 8.02 - 242.99 3524 -13296 50.66 - 99.27 8.52 - 49.54 1.00 - 3.26 It is acrucial step to identify the maximum and minimum values before normalisation/weights to avoid skewed values (outliers), which could result in unrealistic low scores and misrepresent areas with high potential. *This variable was originally in meters (m); however, it was transform to logarithmic scale to simplify the analysis. www.pysolo.eu 9Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 Pairwise variable combinations were performed to examinate the influence of one variable against another •Black dot is the median sensitivity, red ribbon the interquartile range, and grey ribbon the 10th -90th percentile range. •These pairwise combinations help to identify how sensitive one variable was against another, examining how y-value changes when x-value increases. •The highest sensitivity was identified in the variables of biomass, forest cover, and road proximity, which help to distribute the weights in the eight scenarios suggested. www.pysolo.eu 16Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 Road network was integrated to optimise the biomass transport The road network was used to calculate the maximum distance (100 km), which was defined as the threshold and incorporated into the total suitability index computation. www.pysolo.eu 17Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 Using simulated annealing algorithm to identify the optimal number of clusters and hub locations www.pysolo.eu 18 Results from the identified clusters: Case of Italy City Potential Biomass (Mg) Effective Biomass (Mg) Proximal Biomass (Mg) Without Natura 2000 (Mg) Bolzano 2 .83e+08 9 .37e+07 (33.1%) 1 .03e+07 (3.6%) 7 .09e+06 (2.5%) Campobasso 5 .02e+07 2 .76e+07 (54.9%) 2 .33e+06 (4.6%) 4 .48e+05 (0.9%) Catania 2 .29e+07 1 .25e+07 (54.5%) 3 .83e+06 (16.7%) 8 .29e+05 (3.6%) Catanzaro 6 .84e+07 4 .40e+07 (64.3%) 1 .37e+07 (20.0% ) 1 .03e+07 (15.0%) Genoa 7 .62e+07 4 .68e+07 (61.4%) 6 .40e+06 (8.4%) 5 .05e+06 (6.6%) Potenza 6 .13e+07 3 .79e+07 (61.9%) 6 .83e+06 (11.2%) 2 .41e+06 (3.9%) Sassari 1 .22e+07 7 .38e+06 (60.3%) 9 .69e+05 (7.9%) 6 .89e+05 (5.6%) Siena 1 .11e+08 7 .47e+07 (67.5%) 1 .03e+07 (9.3%) 6 .19e+06 (5.6%) Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 The table show the eight clusters determined in Italy as the optimal number and the location of hubs (cities) to store the forest biomass, highlighting the amount of biomass completely free of restrictions. www.pysolo.eu 19Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 Proximal biomass and areas without Natura 2000 restrictions were classified by tree species Proximal Biomass (Mg) Without Natura 2000 (Mg) City Pines Oaks Others Pines Oaks Others Mg %Mg %Mg %Mg %Mg %Mg % Bolzano 8,549,000 83 0 0 1,751,000 17 6,026,500 85 70,900 1 992,600 14 Campobasso 0 0 1,234,900 53 1,095,100 47 0 0 300,160 67 147,840 33 Catania 421,300 11 3,255,500 85 153,200 4 265,280 32 472,530 57 91,190 11 Catanzaro 1,781,000 13 0 0 11,919,000 87 927,000 9 0 0 9,373,000 91 Genoa 0 0 960,000 15 5,440,000 85 50,500 1 656,500 13 4,343,000 86 Potenza 0 0 5,327,400 78 1,502,600 22 0 0 2,241,300 93 168,700 7 Sassari 9,690 1 959,310 99 0 0 0 0 689,000 100 0 0 Siena 1,030,000 10 5,150,000 50 4,120,000 40 309,500 6 3,218,800 52 2,661,700 42 The pixels identified in the proximal biomass and with no Natura 2000 restrictions were masked with the tree species map provided by the European Forest Institute (EFI). www.pysolo.eu 20 Spanish Forest Map EFI Tree Species Map Artificial Intelligence Map Bonnanella et al.2022Brus et at. 2011Spanish Ministry (2010) Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 It is the previous table correct with the biomass (Mg) assigned to the three groups of species? Let’s check the three following examples in Pinus sylvestris: www.pysolo.eu 21Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 How is the biomass distributed by tree species? Are the species found in the hubs the most suitable for the pyrolysis proccess? Some concerns arise when promising cities accumulate high amout of biomass with other species, which are not neccesarily pines or oaks. For example, Catanzaro (south of Italy) showed good potential between biomass availability and solar-power energy, but further studies regarding pyrolysis proccess in unkown species must be carry out. www.pysolo.eu 22 Conclusions Earth Sensing Summer School, Erico Kutchartt, 11.09.2025 •It is possible to find a good balance between forest biomass and solar-power energy, identifying minimun DNI values to carry out the thermo-chemical proccess. •The hubs selected in each cluster must be explore by local data (forest inventories and road networks) before to take the decision of allocate or not a biorefinery plant. •Target tree species must be study very well before to stablish the biorefinery plants, because wood anatony is completely different between species (coniferous vs broadleaves) and it can affect the pyrolysis proccess. •The market must be analyse before to stablish any biorefinery plant because it is important to avoid competition with other forest wood uses (timber, paper, etc.). If not, forest biomass can be not available or too expensive for the bioenergy products. www.pysolo.eu 23 Thank you! Contact Dr. Erico Kutchartt [email protected]