2.1 Climatic criteria

The climatic criteria selected follow the literature review in studies that estimate the potential of photovoltaic solar energy using AHP in conjunction with GIS and the information availability. The criteria selected for this study are temperature (^{Tahri et al., 2015}; ^{Suh and Brownson, 2016}; ^{Al Garni and Awasthi, 2018}; ^{Ruiz et al., 2020}; ^{Sengupta et al., 2021}), cloud index (^{Noorollahi et al., 2016}; ^{Zoghi et al., 2017}; ^{Sengupta et al., 2021}), and solar irradiance (^{Tahri et al., 2015}; ^{Watson and Hudson, 2015}; ^{Noorollahi et al., 2016}; ^{Suh and Brownson, 2016}; ^{Asakereh et al., 2017}; ^{Zoghi et al., 2017}; ^{Al Garni and Awasthi, 2018}; ^{Mohamed, 2020}; ^{Haddad et al., 2021}; ^{Ríos and Duarte, 2021}; ^{Sengupta et al., 2021}; ^{Coruhlu et al., 2022}).

The quality of the terrestrial observations in Mexico is questionable because the sensors installed in meteorological stations that measure surface solar irradiance (the most important variable in this study) do not receive constant calibration. Therefore, the databases usually have errors, therefore in this situation, the reliability is low when characterizing the spatial distribution patterns of surface solar irradiance (^{Huang et al., 2019}). In this case, satellite information is established as a great alternative source of information in areas where the quality of terrestrial observations is poor.

Daily annual average values derived from the Land Surface Temperature (LST) product were utilized for the temperature criterion. Through Python, the data was downloaded hourly throughout 2020 via AWS by NOAA (^{NOAA, 2022}) and processed.

The cloud index criterion was estimated with radiance values of the Advanced Baseline Imager (ABI) sensor from Geostationary Operational Environmental Satellites, R Series, better known as GOES 16, using Heliosat methodology (^{Diabaté et al., 1989}). The radiance values were downloaded every hour throughout 2020, processed, and then used to calculate the daily annual average cloud index. Subsequently, the cloud index was employed to calculate the atmospheric transmission factor, which in turn was used to estimate global surface solar radiation using the Heliosat methodology. Finally, the daily annual average of solar irradiation was computed. Radiance values were obtained from AWS by NOAA (^{NOAA, 2022}) and processed using Python scripts.

2.2 Geographic Information Systems processing

The characteristics of climatic criteria are shown in Table I. The variables used share practically the same characteristics, therefore their GIS processing steps are almost identical. The GIS data were processed and edited using ArcGIS 10.8 software, employing various tools.

The first step involved interpolating the data layers using the “Natural neighbor” tool. The resulting layers were classified using the “Natural Jenks” methodology, which maximizes the distinctions between categories. It is important to note that the temperature criterion was an exception and was classified manually. Then the reclassification process changed the input values of the climatic criteria to a numerical scale. In this study, the scale range is from 1 to 5. At the same time, this numerical scale corresponds to a specific type of technology (Table II) to associate the values of each climatic criterion with corresponding technology types based on the electrical characteristics provided in Tables SIV and SV of the supplementary material.

The reclassification process (Table III) was determined based on the information provided in Tables II, SIV, and SV. When the average solar irradiance is high, it is advisable to opt for solar panel technologies with lower efficiency, and conversely, when it is low, technologies with higher efficiency are recommended (see Table SIV). For regions with high average temperatures, it is advisable to use solar panels with the lowest negative temperature coefficient (see Table SV). Regarding the reclassification of the cloud index, we considered solar panel technologies capable of utilizing diffuse radiation, such as monocrystalline silicon and polycrystalline silicon. The cloud index is a normalized value that quantifies the atmosphere’s reflective properties concerning the ground’s reflective properties. Values close to or below 0 indicate clear conditions, while values close to 1 suggest cloudy conditions. A high cloudiness index signifies greater diffuse radiation and lower temperatures.

2.3 Analytical Hierarchical Process

AHP has established itself as the most widely used multi-criteria decision-making (MCDM) method when used with GIS for determining specific objectives in regional areas, as demonstrated by the previously cited studies. This study follows ^{Saaty (2001)} in the development of AHP methodology.

The first step is to determine the objective and the criteria to use. The objective is to recommend solar panel technologies based on different climatic criteria (solar irradiance, cloud index, and temperature) in Mexico. The second step is to determine priorities among climatic criteria based on a pairwise comparison matrix (Table IV). The priorities were established through a literature review, where the number of authors citing the selected climatic criteria was observed. The total number of articles using each climatic criterion determined its priority in this study (Table V). The values used for each element in these pairwise comparison matrices are used by ^{Saaty (1987)}.

The third step is calculating the weight of each climatic criterion. The previous pairwise comparison matrix must be normalized to calculate the weight for each criterion from an average of each row (Table VI). The weight demonstrates the importance of each climatic criterion to achieve the goal. To finish this step, it is necessary to calculate suitability scores using Eq. (1).

where Ss _i represents the suitability scores, x _i , is the criterion value assigned by Saaty’s criteria score (^{Saaty, 1987}), w _i is the weight calculated for each criterion, and n is the number of selected criteria.

The last step is the calculated consistency ratio to verify inconsistencies in the pairwise comparison matrices. This step is crucial to ensure a correct valuation. Eqs. (2), (3), and (4) were used to calculate the consistency ratio (Table VII).

where:

CI is the consistency index, m the matrix size, RI the random index value is according to ^{Saaty (1987)}, CR the consistency ratio, which must be < 0.10 to determine consistency, and λ _max the maximum eigenvalue of the pairwise comparison matrix.

To finalize the methodology chapter, the weighted overlay tool in ArcGIS was employed. This tool incorporates the reclassification values (Table III) to homogenize the varying numbering ranges of the climatic criteria, considering the input values of the criteria and the electrical characteristics of solar panel technology types (Tables SIV and SV). The reclassified values are subsequently multiplied by the relative weights obtained through the AHP method for each criterion. Finally, the resulting values are summed together to generate an output raster, which is then used to create the suitability map.

3.1 Climatic criteria maps

The technological recommendation for commercial solar panels in each climatic criterion is influenced by several factors. These include the electrical parameters of the solar panel (Tables SIV and SV), the spatiotemporal variation of the climatic criteria (Figs. 4-6), and the chosen data classification method. With these considerations in mind, the following results are obtained:

According to the solar irradiance criterion (Table III), CIGS and CdTe emerge as the most recommended commercial solar panel technologies, covering approximately 32.51 and 31.81% of Mexico’s total area, respectively. Areas where CIGS and CdTe are recommended exhibit a daily annual average solar irradiance ranging from 448 to 497 W m^-2. The solar irradiance map (Fig. 3) allows us to identify that the states with the highest daily annual average solar irradiance include Morelos (526 W m^-2), Guerrero (504 W m^-2), and Colima (498 W m^-2), and the lowest values are found in Quintana Roo (421 W m^-2), Veracruz (430 W m^-2), and Yucatán (432 W m^-2) (for the rest of the states see the supplementary material).

Fig. 3 Map of solar irradiance daily annual average (W m^-2) in Mexico, 2020. 

Based on the temperature criterion (Table III), monocrystalline silicon is the most recommended commercial solar panel technology, covering 41.89% of Mexico’s total area. Monocrystalline silicon is recommended for areas with a daily annual average temperature below 24 ºC. For the remaining areas of the country, solar panels with lower temperature coefficients are advisable. The temperature map (Fig. 4) allows us to identify that the states with the highest daily annual average temperature are Baja California Sur (28 ºC), Sonora (27 ºC), and Sinaloa (27 ºC), and the lowest values are found in Tlaxcala (18 ºC), Mexico City (18 ºC), and Mexico State (18 ºC) (for the rest of the states see the supplementary material).

Fig. 4 Map of temperature daily annual average (ºC) in Mexico, 2020. 

Temperature is the most significant factor affecting the degradation of electrical power delivered by solar panels. The electrical power calculated in a photovoltaic system depends on the series-parallel arrangement of the electrical connections between the panels and their quantity, determined by the required current and voltage output. Losses occurring at the point of maximum power due to temperature increases result in a reduction of this parameter, causing the actual value to decrease in comparison to the calculated one. In states with the highest average temperatures, it is advisable to use solar panels containing solar cells with lower temperature coefficient values to minimize these losses. In this scenario, amorphous silicon and CdTe solar panels are recommended due to their relatively lower temperature coefficient values. Alternatively, oversizing the photovoltaic system by increasing the number of solar panels can also be considered as a solution for other types of technology.

According to the cloud index criterion (Table III), CIGS is the most recommended commercial solar panel technology with 52.72% of Mexico’s total area. CIGS is recommended for areas with a daily annual average cloud index ranging from > 0.126 to 0.161. The cloud index map (Fig. 5) allows us to identify that the states with the highest daily annual average cloud index values are Quintana Roo (0.21), Chiapas (0.20), Veracruz (0.20), and Yucatán (0.20). Conversely, the states with the lowest cloud index values are Baja California (0.11), Morelos (0.11), and Sonora (0.12) (for the rest of the states see the supplementary material).

Fig. 5 Map of cloud index daily annual average (dimensionless) in Mexico, 2020. 

On the other hand, in states with a high cloud index, the use of bifacial solar cells is advisable since they harness the diffuse component of light on their rear face. Bifacial solar cells based on silicon technology have shown the most promising results to date and are considered the leading technology in this field.

3.2 Suitability map

The spatial distribution of recommended solar panel technologies in the suitability map of Mexico reveals a discernible pattern (Fig. 6). Monocrystalline and polycrystalline silicon technologies, which share similar characteristics, are primarily recommended in states around the Gulf of Mexico and northern Chihuahua. These areas are characterized by a high cloud index and low solar radiation values. In contrast, CdTe and CIGS technologies are predominantly recommended in states around the Gulf of Cortés and some areas in central and northern Mexico, which feature high temperatures, low cloud index values, and high-intensity solar radiation. Amorphous silicon exhibits the most versatile distribution pattern in Mexico, making it well-suited for a wide range of climatic conditions. Complementary details on irradiance, cloud index, and temperature results in the 32 States of the Mexican Republic and experimental data of the characterization of solar panels and cells in different parts of the country can be found in the supplementary material.

Fig. 6 Suitability map of solar panel technologies in Mexico. 

According to Table VIII, CIGS and amorphous silicon are the most suitable solar panel technologies for Mexico. This determination is primarily attributed to how climatic criteria were classified and reclassified within the GIS process, as well as the weights estimated in the AHP methodology. It is worth noting that if an alternative method for data classification had been chosen, the results might have been different. However, given the absence of direct references from similar works in Mexico, this study represents the pioneering effort to recommend the most suitable solar panel technology for the country based on the evaluated climatic criteria.

The characteristics of the solar panel technologies analyzed in this study suggest that in areas where monocrystalline silicon is deemed suitable, polycrystalline silicon can also be a viable recommendation, and vice versa. Similarly, this holds true for CIGS and CdTe technologies. Consequently, the choice between technologies may hinge on additional factors such as available investment or market availability.