Deforestation and Its Effect on Surface Albedo and Weather Patterns

Abstract

Deforestation is an important environmental problem and a key promoter of regional climate change through modifying the surface albedo. The objective of this research was to characterize the impact of deforestation and land use changes on surface albedo (α) and climate patterns in a tropical highland region of Mexico, between the years 2014 and 2021. The main land cover types are coniferous forests (CF), oak and gallery woodlands (OGW), and annual agriculture (AA), which represent more than 88% of the regional territory. We used 2014 and 2021 Landsat 8 OLI images with topographic and atmospheric correction in order to develop an inventory of albedo values for each land cover type in both time scenarios. Albedo images were generated by using the equation proposed by Liang in 2001, which is based on the reflectance of the bands 2, 3, 4, 5, and 7. Differences in albedo values were calculated between the years 2014 and 2021, and those differences were correlated with variations in climate parameters, for which we used climate data derived from the WRF model. In addition, the different land use changes found were classified in terms of triggers for increasing or decreasing surface albedo. We used the Mann–Whitney U Test to compare the 2021 − 2014 climatic deviations in two samples: Sample A, which included sites without albedo change in 2021; and Sample B, including sites with albedo change in 2021. Results showed that between 2014 and 2021, at least 38 events of land use change or deforestation occurred, with albedo increments between 1 and 11%, which triggered an average increment of 2.16% (p < 0.01; Mann–Whitney U Test) of the regional surface albedo in comparison to the 2014 scenario. In this period, the albedo for CF, OGW, and AA also increased significantly (p < 0.001; Mann–Whitney U Test) by +79, +12, and +9%, respectively. In addition, the regional albedo increment was found to be significant and negatively correlated (p < 0.01 Spearman’s coefficient) with relative humidity (RH), maximum temperature (Tmax), and minimum temperature (Tmin), and correlated (p < 0.01) positively with diurnal temperature range (DTR). The Mann–Whitney U Test revealed that 2021 climatic variations in Sample B sites are statistically different (p < 0.05) to 2021 climatic variations in Sample A sites, which demonstrates that albedo changes are linked to a decrease in minimum temperature and relative humidity and an increase in DTR. Conversion of CF and OGW into AA, perennial agriculture (PA), or grassland (GR) always yielded an albedo increment, whilst the conversion of AA to irrigation agriculture (IA) or PA triggered a decrease in albedo, and finally, the pass from GR or AA to protected agriculture (PA) caused albedo to increase or decrease, depending on the greenhouse covers materials. Reducing deforestation of CF and OGW, conversion of AA or GR into PA, and selecting adequate greenhouse covers could help to mitigate regional climate change.

1. Introduction

Forests provide environmental services such as carbon sequestration, fauna refuge, soil generation and conservation, being part of the trophic chain, and climate regulation, among others [1]. Nevertheless, deforestation and changes in land use constitute a constant threat to the effectiveness of these services and the stability of ecosystems in the urban–rural gradients. According to FAO [2], cropland expansion is the main driver of deforestation, causing almost 50 percent of global deforestation, followed by livestock grazing, accounting for 38.5 percent.

Surface albedo is an important parameter in the land cover energy balance and is assumed as a primary essential climate variable. Variations in surface albedo can be used as a diagnostic tool for local climate change [3]. The albedo of forests is low compared to other plant covers; hence, forests absorb more solar radiation [4]; when deforestation occurs, the first environmental service affected is climate regulation via an immediate alteration of the surface albedo value, affecting the radiative balance of the ecosystem [5,6]. This begins a chain of climatic effects, including a variation in evapotranspiration rates and environmental humidity levels, thus modifying the water cycle, water catchment, and carbon sequestration [7,8,9,10].

By decreasing the amount of atmospheric humidity, the climate regulation capacity of forests is reduced; thus, deforestation also contributes to a greater incidence of extreme temperature events, with new maximum temperature records and more presence of meteorological frosts; as a consequence, generally, the diurnal temperature range increases [11,12]; moreover, an alteration in the number of cloudy and rainy days may occur, as well as an increase in the intensity of precipitations and floods [5,8]. In addition, when the forest cover is removed, CO₂ and volatile organic compounds are released into the atmosphere, contributing to global climate change [9], which culminates in the deterioration of environmental comfort for different living organisms.

Local/regional climate change caused by deforestation and land use changes synergize with global warming causing a more dynamic climate change process [13]. The climatic effects of deforestation and albedo changes depend on the latitude where deforestation occurs; thus, tropical deforestation is generally found to warm the climate, whereas high-latitude deforestation is generally found to cool the climate [1,7,9,14,15]. However, the climatic effects of deforestation in tropical highland forests have not been rigorously investigated, or else they have been overlooked in many studies where the relationship between deforestation and climate change has been analyzed. An empirical study developed for six coniferous forests in tropical and subtropical highland areas of Mexico revealed that deforested areas have higher temperatures than wooded areas (+0.43 to +0.69 °C), and that the type of land cover and forest species influence the diurnal temperature range (DTR) [16]. Once deforestation occurs, subsequent land cover changes also trigger albedo changes and climatic effects, both of which can be of variable magnitude and direction (increase or decrease), depending on the reflectivity properties of the new land cover [3].

The objective of this research was to characterize the albedo and climatic effects of deforestation and land use changes in a tropical highland region of western México.

2. Materials and Methods

2.1. Study Area

The study area was the municipality of Tapalpa, which is located in the southwest of the state of Jalisco, México, between 19°36′49″ and 20°05′54″ north latitude, and between 103°36′20″ and 103°54′00″ west longitude, at an average altitude of 1950 m. It has a territory of 442.15 km², including 17,715 ha of temperate woodlands, mainly represented by coniferous and pine–oak forests [17]. The predominant climate type according to the Köppen–García system [18] is Cb(w″₂)(i′)g, which is described as a subhumid temperate climate with rainfalls occurring in summer and autumn, a percentage of winter rain between 5 and 10.2%, and with little annual thermal oscillation (5 to 7 °C). The warmest month is May and the coldest one is January [19]. The annual mean precipitation is 891.2 mm, and the annual mean temperature fluctuates between 8.5 and 23.7 °C [19].

2.2. Supplies

Land cover and topographic datasets were used from Landsat 8 remote sensors for the periods 2014 and 2021. Images with little or no cloud cover (<10%) and a resolution of 30 m were selected and downloaded from the dataset Landsat Collection 2 on the website USGS Earth Explorer (https://earthexplorer.usgs.gov/, accessed on 23 May 2023). Information from bands 2, 3, 4, 5, and 7 (blue, green, red, near-infrared, and shortwave infrared 2) was used to estimate albedo [1,20]. Other inputs used were a Terrain Elevation Model (TEM) from the Mexican Elevation Continuum 3.0 (CEM 3.0), Municipal Political Division Vector dataset 2020 [21,22], and the SAMOF 2016 land use and land cover type map [23], and monthly climatic rasters were prepared from the daily data of the Weather Research and Forecasting (WRF) model provided by the National Laboratory of Modeling and Remote Sensing of the Instituto Nacional de Investigaciones Forestales Agrícolas y Pecuarias [24]. Prior to the calculation of albedo values, the satellite images were subjected to processes of atmospheric and topographic correction and conversion from narrowband to broadband [20], for which the images were processed and analyzed with the QGIS 3.24, ERDAS Image 2015, and IDRISI 17.0 software.

We made an inventory of the area (hectares) corresponding to each land cover in Tapalpa. For this, the vector dataset of the municipal political division and land use and a map of land cover types from SAMOF 2016 [23] were used.

2.3. Preliminary Images Process

2.3.1. Topographic Correction

It eliminates the distortions given by the irregular shape of the terrain, since the shading and lighting effects, respectively, produce lower and higher reflectance [25]. It also includes correcting for geometric distortion, which involves correcting the image according to a given geographic projection system. The methods to carry out the topographic correction depend on the degree of distortion, which is moderate in the case of the satellites of the Landsat series (multi-spectral sensor). There are no standardized models for topographic correction processes [25,26]. In this research, the shading effect was corrected with the C-correction method, by using the program ERDAS Imagine 2014; the C-correction method is based on lighting conditions, which is why it uses a terrain elevation model to calculate the angle of incidence of radiation [25,26]. The Equations (1) and (2) is as follows:

$$L_h=L_T\times \frac{COS Z+C}{COS i+C},$$

(1)

where:

• L_h = reflectance of a horizontal surface;
• L_T = reflectance from an inclined surface;
• COS Z = cosine of the solar zenith angle;
• COS i = cosine of incident local angle;

$$C=\frac{b}{m} for L_T=m\times \mathit{COS} i+b$$

(2)

where:

• m = linear regression gradient: L_T-COS i;
• b = linear regression interception: L_T-COS i.

2.3.2. Atmospheric Correction with Solar Angle

Atmospheric correction eliminates or reduces the distortion associated with gases and particles in the atmosphere, which scatter solar radiation captured by satellite sensors. It is accomplished by using spectral characteristics for each band of the sensor expressed in digital levels (DN) [1,27,28]; they are proportional to the original radiance captured by the sensor; they are in 16-bit unencrypted format, so it is necessary to transform them to radiance values of the top of the atmosphere (TOA) starting from the sensor calibration coefficients and the scale factor of 0.1 (Equation (3)) [20,27]; and they are normalized to transform into reflectance values expressed in percentage given that the real reflectance of a roof captured by the sensor is conditioned by the behavior of the atmosphere and the angle of observation; for this the conversion method with solar angle (Equation (4)) [20,27] was used. This method improves the precision of the reflectance of the surface, in comparison with other methods. For the process, a model was created in the ERDAS Image 2014 software (Figure 1).

$${\rho }_{{\lambda }^{\prime }}=M_{\rho }{Q}_{cal}+A_L$$

(3)

where:

• ${\rho }_{{\lambda }^{\prime }}$ = Spectral radiance TOA;
• $M_{\rho }$ = Band—specific multiplicative rescaling factor;
• $A_L$ = Band—specific additive rescaling factor;
• $Q_{cal}$ = Quantified and calibrated standard product pixel values (DN);
• $M_{\rho },A_L$ are the specific calibration data for each band available in the image metadata;

$$P_{\lambda }=\frac{P_{{\lambda }^{\prime }}}{\cos {(\theta }_{SZ})}=\frac{P_{{\lambda }^{\prime }}}{\sin {(\theta }_{SE})},$$

(4)

where:

• ${\rho }_{\lambda }$ = Reflectance TOA;
• ${\theta }_{SZ}$ = Local solar zanital angle; ${\theta }_{SZ}=90°-{\theta }_{SE}$;
• ${\theta }_{SE}$ = Sun elevation angle; information contained in the image metadata. *
• * refers to the sun elevation angle from the center of the scene in degrees.

2.4. Albedo Calculation

Surface albedo values can be calculated based on the corrected surface reflectance from Landsat 8 OLI/TRIS 2014 and 2021 satellite images using the blue, green, red, near-infrared (NIR), and shortwave infrared (SWIR) bands 1 and 2 [1,27,29]. To obtain albedo images in this study, Equation (5) proposed by Liang [29] was used:

Albedo = 0.356 αBLUE + 0.130 αGREEN + 0.373 αRED + 0.085 αNIR + 0.072 αSWIR2 − 0.0018

(5)

where:

• αBLUE, αGREEN, αRED, αNIR, and y αSWIR2 are the reflectance data corresponding to bands 2, 3, 4, 5, and 7, respectively.

In order to explain possible albedo variations between 2014 and 2021 and their implications for changing weather patterns, areas with and without land cover change were located by inspecting high-resolution images from Google Earth.

The different land use changes found were classified in terms of triggers for increasing or decreasing surface albedo by calculating the percentage difference between albedo 2021 and albedo 2014. Positive differences meant increased albedo, whilst negative ones were taken as decreasing albedo. This classification enabled us to evaluate which types of land use changes could eventually contribute to mitigating the regional climate change caused by deforestation or land use change practices.

2.5. Relationship between Albedo Changes and Variation of Climatic Variables

In order to estimate the impact of the surface albedo changes on the climatic conditions, we used the following as indicator variables: relative humidity (RH), maximum temperature (Tmax), minimum temperature (Tmin), and diurnal temperature range (DTR). They were calculated monthly and annually for the years 2014 and 2021 by using daily data derived from the WRF model. The variable DTR is considered an indicator of the climatic regulation capacity of an ecosystem [30,31,32] and is calculated as (Equation (6)):

DTR = Tmax − Tmin.

(6)

For RH, Tmax, Tmin, and DTR, we made monthly and annual raster layers throughout processes of interpolation in the software, QGIS 3.24, and with the method, IDW (Inverse Distance Weighting), which estimates the values of the cells by calculating averages of the values of the sample data points in the vicinity of each processing cell [33]. Punctual and georeferenced values of the four climatic variables were extracted from the raster layers generated in order to structure a data matrix in an Excel spreadsheet. Albedo values for 2014 and 2021 were also added to this data matrix to proceed with statistical analysis.

2.6. Statistical Analysis

Before proceeding with the statistical analyses, the albedo and climatic data were tested for normality using the Shapiro–Wilk test. The results of this test showed no normality for all data analyzed. This is why we used non-parametric statistics for subsequent statistical analyzes.

Statistical analysis included correlation analysis with Spearman’s statistic; albedo changes 2021 − 2014 were correlated versus climatic variables changes 2021 − 2014. Analysis of means comparison was made by using the Mann–Whitney U Test. In this case, albedo values from 2014 and 2021 were analyzed to declare significant or not significant statistical differences. The software used to perform these analyses was IBM SPSS Statistics 19. For the whole municipality of Tapalpa, 633 sites with WRF climatic data were located; thus, statistical analysis at municipal scale were made with those data. Accordingly, we only used the albedo data of the pixels corresponding with those 633 WRF sites when studying the relationship between albedo changes and climatic variations.

To prove that the climatic variations from 2014 to 2021 could indeed be due to a change in surface albedo, two samples of 48 sites each were used: sample A, which included sites with no albedo change in those seven years; and sample B, which included sites with land use and albedo. The Mann–Whitney U test was used for this purpose.

3. Results

3.1. Representativeness of Land Covers in the Study Area

Figure 2 shows the land covers present in the study area, and their representativeness (percentage of the total study area) is described in

Table 1. Types of land cover and their representativeness in terms of area occupied in the Municipality of Tapalpa, Jalisco, México.

	Area
Land Cover Types	Hectares	%
Urban area (UA)	773.09000	1.24
Tular (TL)	59.99484	0.09
Bare ground (BG)	32.5345	0.052
Natural grassland (NGR)	197.84891	0.31
Low deciduous and sub-deciduous forest (LDF)	4371.1777	7.05
Cultivated or induced grassland (CGR)	1214.52896	1.96
Cloud forest and low evergreen forest (CEGF)	1.35978	0.0021
Oak and gallery forest (OGF)	12,094.98695	19.52
Cultivated and induced forest (CIF)	11.69106	0.018
Coniferous forest (CF)	26,356.0703	42.55
Water bodies (WB)	483.7009	0.78
Perennial agriculture (PA)	5.98163	0.01096
Annual agriculture (AA)	16,337.47	26.37
Total Municipality of Tapalpa	61,940.44	100.00

. As can be seen, the most important types of land cover are coniferous forests (CF) at 42.55% of the territory, annual agriculture (AA) at 26.37%, and oak and gallery forest (OGF) at 19.52%. Other types of land cover maintain a low representation.

3.2. Albedo Values for Types of Land Cover

Table 2. Maximum (Max) and Average (Ave) albedo values (%) for types of land cover in 2014 and 2021, and significance p-value for the Mann–Whitney U Test comparing albedo means for 2014 and 2021 in the municipality of Tapalpa, Jalisco, México.

	Albedo Values
					Difference
	2014		2021		Ave 2021 − Ave 2014
Land Cover Types	Max	Ave	Max	Ave	Diff	p-Value
Cloud and low evergreen forest	7.15	6.12	7.87	6.46	+0.34	0.191
Cultivated and induced forest	7.18	6.34	8.85	5.53	−0.81	0.011
Perennial agriculture	8.40	6.84	9.29	6.52	−0.32	0.0001
Tular	8.54	5.69	8.28	7.6	+1.91	0.005
Water bodies	10.01	4.72	10.55	4.69	−0.03	0.0001
Natural grassland	10.61	7.13	12.19	7.8	+0.67	0.0001
Low deciduous and sub-deciduous forest	10.92	6.13	12.99	6.94	+0.81	0.0001
Cultivated and induced grassland	12.97	7.11	19.63	5.77	−1.34	0.0001
Coniferous forest	13.39	5.37	14.89	9.6	+4.23	0.0001
Urban area	15.42	7.99	17.29	8.61	+0.62	0.0001
Bare ground	16.01	9.51	16.67	7.81	−1.70	0.340
Oak and Gallery forest	16.27	5.77	21.20	6.47	+0.70	0.0001
Annual Agriculture	21.46	7.82	27.01	8.52	+0.70	0.0001
Municipality of Tapalpa	21.46	6.22	27.01	6.79	+2.16	0.0001

shows the maximum and average albedo values for each land cover in the study area in the years 2014 and 2021. A simple comparison can be made by looking at the 2014 and 2021 average albedo values; thus, in 2014, the three land covers with the highest albedo were BG, UA, and AA with 9.51, 7.99, and 7.82%, respectively; however, in 2021, the three highest values corresponded to CF, UA, and AA with 9.6, 8.61, and 8.52%, respectively. In contrast, the three types of land cover with the lowest albedo value in 2014 were WB, CF, and TL with 4.72, 5.37, and 5.69%, respectively; whereas in 2021, they were WB, CIF, and CGR with 4.69, 5.53, and 5.77%, respectively (Table 2).

Table 2 also shows that all land covers varied their albedo values during the period 2014 through 2021, which indicates the presence of alterations or changes in land use. Eight positive cases are noted (this is an increase in albedo), and five cases in which the albedo decreased. However, making a balance based on the data in Table 1 and Table 2, we can derive that more than 95% of the territory has experienced an increase in surface albedo, which has produced an increment of 2.16% for the study area albedo (p < 0.0001; Mann–Whitney U Test).

Figure 3 allows evidence of the surface albedo increment in most of the land cover types by the year 2021. Out of CEGF and BG, all land covers showed an albedo change significant (p < 0.05; Mann–Whitney U Test). It is remarkable that from the eight cases of albedo increment, the highest augment corresponded to CF land cover (+4.23%).

In Figure 4, the spatial distribution of several ranges of albedo values for 2014 and 2021 scenarios can be seen and compared. The maps enable a simple but illustrative comparison of both scenarios and elucidate the extent of deforestation and land-use changes’ effects on the variation of surface albedo.

3.3. Land Cover Changes between 2014 and 2021 and Their Impact on Albedo Value

Table 3. 38 Polygonal land cover changes detected between 2014 through 2021 and their impact on albedo value in the Municipality of Tapalpa, Jalisco, México.

Albedo Change in Land Cover Types
Change Year	Land Cover Type 2014	Land Cover Type 2021	Albedo 2014 (%)	Albedo 2021 (%)	Difference 2021 − 2014 (%)
2015	Annual agriculture	Protected agriculture	9.10	11.38	+2.27
2015	Annual agriculture	Protected agriculture	8.43	7.49	−0.94
2016	Coniferous forest	Perennial agriculture	4.99	9.97	+4.97
2016	Grassland	Protected agriculture	7.47	16.84	+9.37
2016	Coniferous forest	Annual agriculture	5.61	9.77	+4.15
2016	Coniferous forest	Grassland	5.25	9.99	+4.73
2016	Oak forest	Annual agriculture	5.10	10.56	+5.46
2016	Oak forest	Annual agriculture	6.08	8.11	+2.02
2016	Annual agriculture	Protected agriculture	9.40	14.39	+4.98
2016	Annual agriculture	Protected agriculture	7.91	7.91	0.00
2016	Annual agriculture	Protected agriculture	10.31	10.11	−0.19
2016	Annual agriculture	Protected agriculture	8.63	9.35	+0.72
2016	Annual agriculture	Irrigation agriculture	10.40	8.43	−1.97
2016	Annual agriculture	Protected agriculture	6.27	14.24	+7.96
2017	Annual agriculture	Protected agriculture	8.73	11.05	+2.32
2017	Grassland	Protected agriculture	7.34	10.54	+3.19
2017	Coniferous forest	Perennial agriculture	5.02	6.54	+1.52
2017	Annual agriculture	Perennial agriculture	6.78	6.25	−0.53
2017	Grassland	Annual agriculture	9.61	15.76	+6.15
2017	Annual agriculture	Protected agriculture	8.52	10.05	+1.53
2018	Coniferous forest	Perennial agriculture	5.75	7.86	+2.10
2019	Oak forest	Annual agriculture	5.85	9.65	+3.79
2019	Coniferous forest	Annual agriculture	5.63	8.26	+2.62
2019	Coniferous forest	Annual agriculture	5.34	7.12	+1.77
2020	Coniferous forest	Bare ground	5.34	9.49	+4.14
2020	Oak forest	Annual agriculture	7.64	9.94	+2.30
2020	Coniferous forest	Annual agriculture	5.39	9.91	+4.52
2020	Coniferous forest	Annual agriculture	5.48	7.92	+2.43
2020	Oak forest	Annual agriculture	5.90	9.00	+3.10
2020	Coniferous forest	Grassland	5.17	8.90	+3.72
2020	Oak forest	Grassland	5.65	12.16	+6.50
2020	Oak forest	Secondary vegetation	5.07	7.21	+2.13
2020	Annual agriculture	Protected agriculture	7.42	12.87	+5.45
2020	Annual agriculture	Protected agriculture	8.65	18.68	+10.03
2020	Annual agriculture	Protected agriculture	7.60	17.97	+10.36
2020	Annual agriculture	Protected agriculture	10.54	11.99	+1.45
2021	Perennial agriculture (trees < 1-year-old)	Perennial agriculture (trees > 7-year-old trees	7.09	5.70	−1.39
2021	Perennial agriculture (trees < 1-year-old)	Perennial agriculture (trees > 7-year-old trees	7.42	6.15	−1.27

and Figure 5 permit the monitoring of 38 specific cases of land use changes detected during the period 2014 to 2021 through Google Earth image analyses. These cases imply 38 polygons, which total 658 ha, which is 1.06% of the municipality’s total area. This amount of affected hectares is distributed as follows: 220 ha from coniferous forest to perennial agriculture; 165 ha from annual agriculture to sheltered agriculture; 125 ha from coniferous forest to annual agriculture; 53 ha from annual agriculture to perennial agriculture; 47 ha from perennial agriculture (trees < 1-year-old) to perennial agriculture (trees > 7-year-old); 26 ha from grassland to sheltered agriculture; 18 ha from oak and gallery forest to grassland; 3.4 ha from coniferous forest to bare soil; and 1.2 ha from coniferous forest to grassland.

As can be seen, the land cover changes reported in Table 3 occurred from 2015 to 2021, and they denote different impacts on albedo value, although most of them represent an increase in albedo value from +0.72 to +10.36%. Only 6 of the 38 cases were found to be a decrease in albedo value, and they were in the range of −1.97 to −0.19 (Table 3).

3.4. Relationship between Albedo Change and Climate Variables

At the municipality scale, albedo changes (α2021 − α2014) correlated negatively and significantly (p < 0.01) with 2021 − 2014 deviations in RH, Tmax, and Tmin, and positively and significantly (p < 0.01) with 2021 − 2014 DTR deviations (

Table 4. Spearman’s correlation between albedo changes and climatic variables changes from 2014 through 2021 and means comparison test between 2014 and 2021 climatic variables in the Municipality of Tapalpa, Jalisco, México.

	Maximum Temperature (°C)	Minimum Temperature (°C)	Diurnal Temperature Range (DTR) (°C)	Relative Humidity (%)
Spearman rho	−0.262	−0.232	0.264	−0.204
p-value	0.0001	0.0001	0.0001	0.0001
2014	26.6	9.7	16.9	49.4
2021	26.5	9.1	17.4	46.7
Mann–Whitney U	0.00002	0.000015	72470	17175
p-value	0.235	0.0001	0.0001	0.0001

). Although Spearman’s correlation values ranged from −0.2 to 0.3, which, according to LaMorte [34], evidence a weak association between variables, in spite of this weakness, the significance of the correlations indicates that albedo changes are possibly triggering a decrease in RH, as well as in Tmax and Tmin, although this effect is higher in Tmin, giving rise to a DTR increase.

Out of maximum temperature, the Mann–Whitney U test revealed significant differences (p < 0.001) between 2014 and 2021 values for the climatic variables analyzed (Table 4). However, it should be considered that in this case, we are comparing weather statistics of two individual years; thus, these differences may be partially due to the inter-annual variability of climate. Nevertheless, it is interesting to note that the regional average relative humidity in 2021 was lower than that of 2014, even though in 2021, the accumulated annual precipitation (984 mm) was 23% higher than that of 2014 (798 mm) (Figure 6).

Table 5. Results of comparing 2021 − 2014 climatic deviations between sites without albedo change (A) and sites with albedo change (B).

Statistics	Maximum Temperature (°C)	Minimum Temperature (°C)	DTR (°C)	HR (%)	Albedo (%)
N Sample A	38	38	38	38	38
N Sample B	38	38	38	38	38
A 2014 mean value	26.41	9.45	16.96	49.42	5.19
A 2021 mean value	26.48	9.05	17.43	46.74	5.28
A 2021 − 2014 deviation	0.07	−0.40	0.47	−2.68	0.09
B 2014 mean value	26.02	8.97	17.06	48.91	7.73
B 2021 mean value	26.09	8.38	17.71	46.05	9.65
B 2021 − 2014 deviation	0.07	−0.59	0.65	−2.86	1.92
Mann–Whitney U Test
Mann–Whitney U	1062	662	802	896	0
p-value	0.528	0.0002	0.01	0.05	0.0001

shows the results of comparing 2021 − 2014 climatic deviations between sites without albedo change (Sample A) and sites with albedo increase (Sample B) during the period 2014–2021. According to the U Test, A sites are different from B sites in albedo, minimum temperature, DTR, and relative humidity deviations (p-value < 0.05), indicating that increasing albedo triggers a decrease in minimum temperature and relative humidity, and an increase in DTR.

Table 6. Monthly and annual diurnal temperature range (DTR) for years 2014 and 2021 and results of the Mann–Whitney U Test for DTR means comparison between 2014 and 2021.

Monthly/Annually	DTR 2014 (°C)	DTR 2021 (°C)	DTR 2021–DTR 2014 (°C)	Mann–Whitney U Test	p-Value
January	17.14	20.35	+3.21	10,975	0.0001
February	20.38	19.59	-0.79	57,995	0.0001
March	20.67	19.59	-1.07	33,746	0.0001
April	20.56	18.79	-1.77	1848	0.0001
May	16.91	19.52	+2.61	2300	0.0001
June	14.29	14.29	0.00	196,788	0.585
July	14.22	13.84	-0.39	48,754	0.0001
August	14.13	13.70	-0.43	16,655	0.0001
September	12.98	13.60	+0.62	19,809	0.0001
October	15.14	15.55	+0.40	69,646	0.0001
November	16.74	19.56	+2.82	164	0.0001
December	19.06	19.14	+0.07	1,777,151	0.0001
Annual	16.85	17.29	+0.43	104,679	0.0001

shows that most of the monthly and annual DTR values of 2021 differ statistically (p < 0.01; Mann–Whitney U Test) from those of 2014. By 2021, DTR increased in January, May, September, October, November, and December, whilst a DTR decrease was observed for February, March, April, July, and August. The summary of these monthly effects yielded an annual DTR increment of +0.43 °C as an average value for the whole study area. The spatial distribution of several ranges of deviations 2021 − 2014 for DTR and minimum temperature are shown in Figure 7.

4. Discussion

The values of albedo surveyed for types of land cover in the study area are reasonably similar to those reported in previous studies [1,4,13,27], showing that the methodology used to determine the surface albedo is adequate. According to the albedo calculation for the images 2014 and 2021, surface reflectivity has been increasing in recent years both on a regional scale and particularly for the different land covers present in the study area. In the map of the spatial distribution of land cover types, it is evident that annual agriculture has a presence close to coniferous forest, which indicates that the fragmentation of this kind of forest is mainly due to the expansion of the agricultural frontier. This is a common pattern of land use change, which has been reported by other researchers in different parts of the world [13,35]. This type of land use change yields one of the greatest impacts on surface albedo [13], and it is one of the most common causes of local/regional climate change [35]. In fact, the results of detecting cases of land use changes revealed that most of the area affected (52.46%) corresponded to coniferous forests displaced by agricultural lands.

The albedo calculation for each land cover showed a majority trend of increasing albedo for the year 2021. Seven land covers, which represent more than 95% of the studied area, showed a significant (p < 0.05) positive albedo difference α2021 – α2014, leading to an increase in albedo of 2.16% on average for the whole study area (Mann–Whitney U Test, p < 0.05). The deforestation process that Tapalpa has experienced during the last years could explain this increment in albedo [10,36,37].

The temporal and spatial variation of albedo has been closely related to global climate change and regional weather and environment [38,39]; in this research, the increment in albedo (2.16%) for the study area was found to be associated with a decrease in maximum and minimum temperature, as well as relative humidity (Spearman rho, p < 0.0001); hence, it was linked to an increase in the diurnal temperature range (Spearman rho, p < 0.0001). Moreover, the results of the Mann–Whitney U Test (p < 0.05) revealed that the regional increase in albedo is related to a decrease in minimum temperature and relative humidity, as well as an increase in DTR, which are indicators of a decrease in the climatic regulation capacity of the study area [40]. In the context of this study, the decrease in relative humidity can be in connection with deforestation and changes in land use, as has been shown in similar studies [41]; however, the drop in relative humidity has also been explained as a global warming effect [42].

Since the decrease in the minimum temperature is greater than that of the maximum temperature, the DTR has increased during the period 2014–2021. The possible explanation in this regard lies in the fact that the climatic effects of deforestation and albedo changes depend on the latitude where deforestation occurs; thus, tropical deforestation is generally found to warm the climate, whereas high-latitude deforestation is generally found to cool the climate [1,7,9,14,15]. In the case of Tapalpa, geographically it is considered a tropical region, but according to its altitude, it is described as a temperate thermal zone, with an annual mean temperature between 12 and 18 °C [18,19,43].

According to the effects on albedo and local climate by the types of land cover changes analyzed, it may be concluded that some of them may contribute to climate change mitigation in previously deforested areas; for example, we observed that albedo decreased for certain land cover changes, such as annual agriculture conversion to perennial agriculture, or changing annual agriculture to irrigation agriculture. We also observed that the conversion of traditional agriculture systems to sheltered agriculture units produced both conditions of increased and decreased albedo, depending on the nature of the covers used in greenhouses. This fact indicates the feasibility of adequately designing these protected agriculture systems according to the desired climatic effects.

The establishment of avocado orchards in areas where currently annual agriculture is practiced was also shown to gradually reduce surface albedo as fruit tree growth occurred. As observed, the growth of avocado trees in seven years yielded a decrease of 17% in albedo. The effect of reducing albedo by orchards has already been reported [44,45,46]. The reduction of albedo when changing from annual agriculture to sheltered agriculture can be explained by considering that there are materials used in greenhouse structures capable of absorbing large amounts of solar radiation, but, in counterpart, increasing the heat island effect [39]. In the case of a change from annual agriculture to irrigation agriculture, the decrease in albedo is explained by the increase in soil moisture and the constant greening of the cover [47,48].

The annual and monthly DTR increase observed in Tapalpa has been reported previously for other regions in México, Bolivia, Patagonia, Madagascar, Indonesia, and central Russia—more than 2  °C since approximately the 1960s [11,49,50,51]. Conversely, five of the twelve months reported a decrease in DTR, a pattern that has also been reported for different regions in the American continent [11,51] as well as in other continents [12,52]. The above denotes the seasonal differentiated climatic effects of the change in surface albedo.

5. Conclusions

The surface albedo of the study area increased 2.16% on average between 2014 and 2021 as a result of deforestation events and changes in land use that occurred in that period. The magnitude of the albedo increase varied according to the type of land cover change; conversion from forest to annual agriculture or conversion from conventional types of agriculture to protected agriculture produced the highest albedo increments. Conversely, the establishment of avocado orchards and irrigated crops on previously impacted lands led to a decrease in surface albedo.

The albedo increment detected for the 2021 scenario was found to be associated with a decrease in the minimum temperature and relative humidity, thus triggering an increase in the diurnal temperature range, which indicates a decline in the regional climate regulation capacity.

The findings of this research enable us to conclude that both deforestation and land cover changes are triggering regional climate changes, which, by creating synergy with global climate change, could represent worrying future environmental scenarios for tropical highland regions.

Reducing deforestation of all types of forests, the conversion of annual agriculture or grasslands into perennial agriculture and the adequate selection of greenhouse covers for new protected agriculture systems could help to mitigate regional climate change.