Study site

Zoquiapan Forestry Experimental Station is located in the Northwest of the volcano Iztaccíhuatl and the Southwestern part of the Zoquiapan National Park, at the following coordinates 19°12'30" and 19°20'00" N, and 98°30'00" and 98º42'30" W. It occupies an area of 1 624.23 ha in the altitudinal range of 3 080 to 3 670 m (^{Blanco et al., 1981}) in the municipalities of Ixtapaluca and Chalco, State of Mexico (Figure 1).

Figure 1 Map of the study area location. 

Homogeneous cartographic landscape units

The methodology proposed by ^{Gómez et al. (2008)} was used to generate geoforms, climate, and vegetation maps with a Geographic Information System (GIS, ArcMap 10.1) to characterize the homogeneous cartographic landscape units. In this study, we define homogeneous landscape units as those areas that share the same physiographic attributes with minimal spatial variation. The three maps were merged to generate the homogeneous cartographic landscape units, which were assigned a name beginning with soil and vegetation attributes, climate and geoform. The process is described in the following paragraphs.

Geoforms map. The Digital Elevation Model (^{INEGI, 2016c}) of Mexico, Puebla and Tlaxcala state was used to highlight altitudinal ranges, slopes and contour lines. The altitudinal ranges used were delineated every 100 m. The ranges of slope followed the criterion established by the United Nations Environment Program and other institutions in the procedure of digital databases on soils and terrain (^{van Engelen and Dijkshoorn, 2013}) based on the evaluation of soil degradation (Global Assessment of Human-induced Soil Degradation or GLASOD), and the contour lines were every 10 m. Next, the map of geological structures (^{INEGI, 2016b}) and the map of water bodies (^{INEGI, 2016a}) were superimposed to characterize the geomorphological units of the study zone.

Climate map. The climate map was characterized using the methodology of ^{Gómez et al. (2008)} by determining annual and monthly temperature and precipitation data. For the first variable, simple linear interpolation methods were complemented with statistical analysis. For the second, a graphic method was used with expert criteria, making analogies with neighbouring areas for which information on this variable was available.

First, with the Extractor Rápido de Información Climática (ERIC, version 1.0) (^{IMTA, 2016}), maximum and minimum temperatures were obtained, and mean annual precipitation from the 22 climatological stations closest to the study area with complete data of at least the last 20 years. The watershed located west (Atoyac River Basin) and within the Central Plateau was considered to determine the influence of winds and select the climatological stations of influence (^{Juárez et al., 2005}).

Temperature map. Data on monthly maximum and minimum temperatures from the climatological stations were used to obtain each station's average annual and monthly temperatures. This aimed to generate simple linear regression equations to calculate the ranges of variation from the elevation for each month and the yearly average. The annual linear regression equation was used to construct the isothermal map at one degree Celsius. Then, the monthly increase was calculated following temperature ranges. Polygons were created by superimposing the contour line map over that of temperature and elevation and calculated with the simple linear regression model in GIS Arc Map 10.1.

Precipitation map. The annual precipitation was obtained from the 22 climatological stations to georeferenced them, according to the coordinates in GIS ArcMap 10.1. With the Digital Elevation Model and the data on annual precipitation from the meteorological stations, the amount of rainfall that could exist in the study area was estimated using analogies with the regions with climatological stations considering the natural vegetation and the influence of the orographic units in the occurrence of rainfall. The procedure is best described in ^{Gómez et al. (2008)}.

Map of areas of climatic influence. The areas of climatic influence are defined as those with the same annual temperature and precipitation (^{Gomez et al., 2008}). We intersected the maps of mean annual isotherm and mean annual isohyet in GIS ArcMap 10.1 to create the areas of climatic influence. Monthly and yearly values for temperature and precipitation were estimated for each area to determine the climate type using the Köppen climate classification system modified by ^{García (2004)}.

Final climate map. Once the climate type was determined, the polygons with the same climate type were merged to obtain the climate map.

Vegetation map. A detailed delineation of land use and vegetation types was done using Series V of Land Use and Vegetation (^{INEGI, 2016d}) scale 1:1 000 000. Also, Google Earth satellite images were used to define the vegetation stands in the study area, which was named by land use and vegetation type and verified in the field (through field trips to all areas).

The land use and vegetation types registered by the ^{INEGI (2016d)} Series V in the study area are fir forest, pine forest, pine-oak forest and induced grassland. Comparing the land use and vegetation units of Series V with Google Earth satellite images, we observed that the borders of the ^{INEGI (2016d)} polygons do not coincide with the vegetation conglomerates. Thus, we refined and elaborated the boundaries with greater detail and precision.

Map of homogeneous cartographic landscape units. Finally, the three maps previously defined in GIS ArcMap 10.1 (geoforms, climate and vegetation) were intersected. From this merging, the homogeneous landscape units were obtained.

Site criteria

Since it is not possible to establish a monitoring system in all the resulting areas, it was necessary to select just three sites. The main reason was economic limitation (Table 1). Thus, the following are the site criteria for choosing the three homogeneous cartographic landscape units where the meteorological stations were physically located.

Representativity. The sites have characteristics representative of the geoforms, climate and vegetation type of the study area, which should interest the research users. To this end, an elevation transect was performed to observe weather behaviour in the three sites with different altitudes, vegetation and geoforms. The region where the study area installations are located (intermountain valley) was used as the base point to give continuity to the meteorological data collected previously at this site.

Cleared site. The sites are free of natural or artificial obstacles that might obstruct the free transit of the meteorological elements; preferably a terrain cleared in a radius of 20 m.

Levelled terrain. The sites do not have depressions or slopes that could limit the stations' installation and make access for data collection difficult during the rainy season.

Distance to the observer. The stations are close to the place where the operator is located.

Easy access. The sites have an open access road to facilitate operation, maintenance, systematic data collection, and constant surveillance of the stations.

The first criterion was established to collect valuable meteorological helpful information in the research projects of the Zoquiapan Forestry Experimental Station. The other criteria for clearing, level ground, proximity to the observer and easy access were defined according to WMO (^{Romo and Arteaga, 1989}).

Calibration testing of the meteorological stations

The monitoring system consists of three meteorological stations denominated EMZ-01, EMZ-02 and EMZ-03. Stations EMZ-01 and EMZ-02 are automatic meteorological stations of the brand Davis Instruments, model Vantage Pro2 Plus wireless (^{DAVIS, 2019}). They have a Vantage Pro2 console and a WeatherLink datalogger for receiving, visualizing and storing data, a seven-watt solar panel and a six-volt-1.4 ampere battery to provide the console with constant energy. The console sensors are described in Table 2. EMZ-03 is an automatic meteorological station built with an Arduino microcontroller. It is attached to a 10-watt solar panel and a 12-volt, 7-ampere battery for a constant energy supply. A microSD module functions as a data logger.

The WeatherLink datalogger of the stations EMZ-01 and EMZ-02 reports inside and outside temperature and humidity, wind chill and dew point, current and trending atmospheric pressure, current precipitation and daily, monthly and annual accumulated precipitation, rain intensity, solar and UV radiation, evapotranspiration, wind direction and velocity, wind gusts, lunar phase and time of sunrise and sunset. In the Arduino microcontroller of the station EMZ-03, the sensors have been programmed with Arduino integrated development environment to obtain simple data on temperature, humidity, current and daily precipitation, atmospheric pressure, and wind direction and speed. The cost of the stations and expenses incurred are shown in Table 1.

There are two ways to verify the sensors of the meteorological stations. The first compares the sensor response with a known reference measurement when the sensor and the reference device are subjected to the same environmental conditions. The second consists of subjecting the sensor to an artificial state in which the response is theoretically predictable (^{INE, 2010}).

We chose to use the reference equipment of the Department of Meteorology of the Universidad Autónoma Chapingo, a Davis Instruments station, Vantage Vue wireless model, with the same characteristics as stations EMZ-01 and EMZ-02. To compare the measurement of the stations, they were placed 5 m from the reference station for one week.

The comparison tests were conducted within one week with sampling every 30 minutes, resulting in a sample of 336 data recorded on an Excel 2010 spreadsheet. Data were compared using a t-test for two samples, assuming unequal variances. The tested sensors were temperature, humidity, pressure, precipitation, solar radiation, UV radiation, and wind direction and speed.

Installation of the meteorological stations

It was necessary to delineate an area of 25 m² in each site around which a cyclone fence with a gate to install the meteorological stations. The fence is 1.5 m high from ground level and has an entrance with a lock to safeguard the stations. The stations were installed in the centre of each fenced-in area. Tripods were anchored to the ground with bases constructed for that purpose, adjusting the tripod screws so that they were firmly anchored to the ground. The weathervanes were oriented toward the north. Solar cells for the equipment were installed facing the south to guarantee efficient use of solar radiation since, most of the year, the sun is south of the latitude in which the stations are found. The rain gauges were levelled with a manual level to ensure that the stand was not inclined to the sides and that the data collection was accurate. The solar radiation and UV radiation sensors were levelled correctly according to the level bubble it is equipped with so that solar radiation reaches the sensors adequately.

Station EMZ-01 was installed on July 22, 2016, and stations EMZ-02 and EMZ-03 were established on August 16, 2016. Once the fences were installed, the stations were placed and configured to begin operation. In addition, data from the first 20 days of operation (August 20 to September 8, 2016) are presented to compare the weather in the three selected cartographic units.

Homogeneous cartographic landscape units

An altitudinal range map shows altitudes from 3 000 m to 3 700 masl, based on the Digital Elevation Model of ^{INEGI (2016c)}. From the slopes map, it can be observed that moderately steep slopes predominate in the southern part of the study area, and in the central to western parts, there are very steep slopes that diminish toward the east. Moreover, the central part is almost flat, and moderately tilted slopes predominate toward the north.

The above maps were superimposed on maps of geological structures (^{INEGI, 2016b}) and water bodies (^{INEGI, 2016a}) to characterize the geological units of the study zone. The geoforms defined are shown in Figure 2. Results show geoforms are as follows: 4.82 % steep volcanic apparatus (SVA), 1.41 % moderately inclined glens (MIG), 0.46 % almost flat tops of volcanic domes (AFTVD), 2.40 % inclined tops of volcanic domes (ITVD), 1.15 % moderately inclined tops of volcanic domes (MITVD), 8.11 % steep hillsides (StH), 3.87 % sloped hillsides (SlH), 13.57 % descended hillside dissected by streams (SlHDS), 14.07 % moderately steep hillsides (MStH), 10.74 % moderately sloped hillsides (MSlH), 1.73 % moderately sloped hillsides dissected by streams (MSlHDS) and 37.66 % almost flat intermountain valley dissected by streams (AFIVDS).

Figure 2 Geoforms map of the study area. 

Map of areas of climatic influence. The polygons of areas of climatic influence are shown in Figure 3. The climate types according to Köppen climate classification modified by ^{García (2004)} are: Cb'(w₂)(w)ig (22.37 %), Cc(w₂)ig (26.38 %) and Cc(w₂)(w)ig (51.26 %) (Table 3). The climate in the study area is temperate. The location and height influence the course of the temperature, while the location and orientation influence the precipitation.

Figure 3 Climate map of the study area. 

Isotherm map. The estimated annual and monthly temperature ranges from the simple linear regression models for the study area are presented in Table 4. In the areas of lower altitudes, the mean annual temperature estimated is in the range of 10-11 °C; at the highest altitude, its range is 6-7 °C.

Isohyet maps. The dominant winds associated with the formation of the cloud systems that generate most of the precipitation in the EFEZ come mainly from the Gulf of Mexico with a Northeast to Southwest trajectory. Annual mean precipitation values of 1 000 to 1 100 mm were estimated at the intermontane valley. Here, the winds descend from slopes located to the North and Northeast of the area of study and then are forced to ascend following the contour of the slopes of the South-southeast where condensation of the water-air vapour increase and then the amount of precipitation with a higher range estimated of 1 300-1 400 mm.

Data on the annual precipitation of each polygon was generated with a table showing average monthly rainfall values, monthly precipitation intervals and estimated monthly precipitation calculated for each range. The intervals of mean monthly precipitation for each range of annual mean precipitation were assigned in accord with the trend and average precipitation values of each group of climatological stations, establishing ranges within which the averages of the different groups of stations were included. The results are shown in Table 5.

Vegetation map. Seven categories were obtained: 0.29 % water (W), 16.60 % fir (F), 2.69 % grassland (G), 49.64 % pine (P), 14.63 % pine-alder (PA), 1.89 % pine-oak (PO) and 14.26 % pine-fir (PF); for more information see Figure 4.

Figure 4 Vegetation map of the study area. 

Map of homogeneous cartographic landscape units. By merging the geoforms, climate and vegetation maps, 38 homogeneous cartographic landscape units were obtained and assigned names following vegetation attributes, climate type and geoform (Figure 5). Table 6 shows the site criteria used to select the sites where the meteorological stations were installed.

Figure 5 Homogeneous cartographic landscape units and meteorological stations location. 

The weather stations were installed considering the desire to give continuity to prior records of meteorological data. First, station EMZ-01 was located at an altitude of 3 273 masl in an almost flat intermountain valley intersected by streams with grassland vegetation and a temperate semi-cold climate.

The station EMZ-02 is located on a steep hillside with an altitude of 3 413 masl, 140 m higher than the EMZ-01. This is relevant for knowledge of climate behaviour in the thermal belt that can occur in this unit's altitudinal range. The vegetation is an Oyamel Fir Forest. This species is significant for the research carried out here in ecological landscape planning (^{Lomas-Barrié et al., 2005}), forest research (^{Martínez-Santiago et al., 2017}) or land use change (^{Paredes-González et al., 2018}). It has the same climate type as the latter.

The station EMZ-03 is located on a steep hillside at an altitude of 3 575 masl, 162 m higher than EMZ-02 and 302 m higher than EMZ-01. The station's location is relevant for knowledge of the climate behaviour in this altitudinal range to have the meteorological information in an altitudinal transect. The natural vegetation cover is a forest with a pine-alder association, which is also essential for the research carried out in the area. Figure 1 shows the location of the three stations in the sites that satisfied the site criteria.

Calibration tests of the meteorological stations

Calibration tests of the stations were conducted with reference equipment belonging to the Department of Meteorology of the Universidad Autónoma Chapingo. Using a t-test for two samples, assuming unequal variances for each of the sensors with 30-minute intervals between data collections for a week, we obtained 336 samples for statistical analysis. The tests comprehended the periods from July 8 to July 14, 2016, for stations EMZ-01 and EMZ-02, while for station EMZ-3, the period was June 2 to June 8, 2016.

For stations EMZ-01 and EMZ-02, calibration tests of the temperature, humidity, solar radiation, UV radiation, air pressure, wind speed and direction sensors were conducted. The data from the test station were compared with the data of the reference station. The statistical t-value for the sensors of the two stations was far below the critical t-value of the two tails (1.96). It is thus assumed that the test of means demonstrated that the data are statistically equal since there are no significant differences between samples.

Calibration tests were conducted on the sensors of each station. We found significant differences in temperature and humidity sensors and calibrated them. Calibration is a systematic process for modelling climatic variables (^{Torres-Quezada et al., 2021}) or agronomic crop prediction (^{Corvino et al., 2018}). The statistical analysis did not significantly differ in the air pressure, wind speed and direction sensors. The precipitation sensor data were validated with the reference station using an artificial rainfall test with rain simulator equipment because no rain was recorded during the testing period. Moreover, the artificial rainfall test offers more certainty in collecting data.

There was a significant difference between samples in the temperature sensor, so an adjustment of -0.78 ºC was made in the Arduino programming code. It was again tested for one week, and the statistical analysis was repeated. The samples showed no significant differences, and the sensor was considered calibrated (Figure 6). For the humidity sensor, a significant difference was found between samples. An adjustment of +2.34 % was made in the Arduino programming code, and the calibration tests were again run for one week. We found that the statistical t-value is lower than the two-tailed critical t-value and thus assumed no significant differences between the samples.

Prueba t para dos muestras suponiendo varianzas desiguales = Two-sample t-test assuming unequal variances; Referencia = Reference; Media = Mean; Varianza = Variance; Observaciones = Observations; Diferencia hipotética de las medias = Hypothetical difference of means; Grados de libertad = Degrees of freedom; Estadístico t = t statistic; P(T=˂t) una cola = One-tail P(T<=t); Valor crítico de t (una cola) = Critical t value (one tail); P(T=˂t) dos colas = Two-tailed P(T<=t); Valor crítico de t (dos colas) = Critical t value (two tails).

Figure 6 Calibration test of temperature sensors. 

The statistical t-values for the air pressure, wind speed, and direction sensors were far below the critical t-value (two-tailed). We thus assumed that the test of means demonstrates that there are no significant differences in the samples. As for the precipitation sensor, the data were validated by comparing the readings of the artificial rainfall equipment with those recorded by station EMZ-03, resulting in similar recorded amounts of rain.

First data calibration

After installing the stations, we configured the WeatherLink consoles of stations EMZ-01 and EMZ-02. For station EMZ-03, the Arduino program code was downloaded. Later, we verified that the data collection was correct. Figures 7 and 8 present the data from the meteorological stations' first 20 days of operation. In the graphs, weather differences in each unit can be observed, underlining those caused by different altitudes, vegetation and geoform at which the stations were installed.

Figure 7 Data from the three stations of a) Temperature, b) Humidity, c) Solar radiation, and d) UV radiation. 

Figure 8 Data from the three stations of a) Pressure, b) Precipitation, c) Wind speed, and d) Wind direction. 

A data-collection schedule was programmed for collection during the first week of each month. Recovery of the data is manual. The operator must download the data monthly to a laptop with a USB port. Data are transferred automatically by connecting the USB cable of the datalogger to a portable computer. For station EMZ-03, the microSD memory of the station must be removed to copy the data in the computer and later return it to the station and again program it with the Arduino programming code.

The results pay off in developing open-source climate monitoring systems, fostering collaboration and improving infrastructure (^{Lin and Zini, 2008}). The use of free software allows weather stations to be adapted by modifying and adjusting their components (^{Stallman, 2004}). In addition, using the Arduino system allowed reaching the goal without being an expert programmer or having high financial resources, which is one of its advantages (^{Ferdoush and Li, 2014}; ^{Katyal et al., 2016}). The results will establish bases for a climate monitoring system in the experimental area of Zoquiapan, providing detailed information and construction of better climate, forest fire and meteorological risk maps (^{Meulenert et al., 2005}; ^{Adepoju et al., 2020}).

Installing automatic equipment has proven to be effective in obtaining climatic information. As shown, it is possible to use automatic stations programmed with Arduino. In areas of complex relief and difficult access, they are a viable option (^{Martínez del Castillo et al., 2012}). It was noticed that the records were sufficient to show the representativeness of the complexity of the landscape and climate in the three sites studied. Representativeness is a fundamental criterion for urban and rural planning because it makes processes more efficient (^{Yang and Regan, 2014}). In addition, installing equipment with Arduino allows for updating weather stations with many years of operation (^{Faugel and Bobkov, 2013}).