Abstract

The authors explore a new approach to monitoring of desertification that is based on use of results on the relation between albedo and surface temperature for the Sonoran Desert in northwestern Mexico. The criteria of predominance of radiation by using the threshold value of Advanced Very High Resolution Radiometer (AVHRR) and Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index (NDVI) were determined. The radiation mechanism for regulating the temperature of the surface and the definition of threshold values for AVHRR and MODIS NDVI have an objective justification for the energy budget, which is based on the dominance of radiation surface temperature regulation in relation to evapotranspiration. Changes in the extent of arid regions with AVHRR NDVI of <0.08 and MODIS NDVI of <0.10 can be considered to be a characteristic in the evolution of desertification in the Sonoran Desert region. This is true because, in a certain year, the time span of the period when radiation factor predominates is important for the desertification process.

1. Introduction

The correlation between albedo and dry-land surface temperature can serve as an indicator of processes that modulate the surface temperature in dry lands. The term dry land is used in reference to arid, semiarid, and dry subhumid regions whose aridity index, defined as the ratio between annual mean precipitation and Thornthwaite (1948) potential evapotranspiration, ranges between 0.05 and 0.65 (see the 1994 United Nations Convention to Combat Desertification in Those Countries Experiencing Serious Drought and/or Desertification, Particularly in Africa; online at http://www.unccd.int/Lists/SiteDocumentLibrary/conventionText/conv-eng.pdf).

For dry lands, there are three competing factors that underlie surface temperature modulation (Becker et al. 1988):

  1. The first factor is radiation. As albedo increases (decreases), the radiation absorbed by the surface decreases (increases), which causes cooling (warming) of the surface.

  2. The second factor is evapotranspiration. Prolonged lack of precipitation, as well as anthropogenic effects, usually causes reduction of vegetation. This, in turn, results in increased albedo and decreased evapotranspiration. This leads to an increase in surface temperature, and vice versa. Modulation of temperature by evapotranspiration and aerodynamic modulation are closely linked through roughness parameters.

  3. The third factor is aerodynamic modulation. As the density of short vegetation (such as grasses and shrubs) decreases, the surface becomes smoother (the roughness parameter decreases). This causes a decrease in vertical fluxes of sensible and latent heat and an increase in diurnal surface temperature. Here, this factor is not considered because field measurements to assess it are unavailable.

Interest in different ratios between the temperature-modulating factors was raised by several researchers (Jackson and Idso 1975; Ripley 1976a,b; Idso 1977, 1981) after numerical experiments by Charney (1975), who studied the impact of albedo of dry land on regional climate. Charney’s albedo mechanism is based on changes in the net radiation at the top of the atmosphere that induce vertical motion to keep thermal balance. Therefore, it is a mechanism that involves the whole atmospheric column response to change in albedo at the surface. According to Charney, radiation is predominant in circumstances in which anthropogenic effects cause reduction of vegetation, which leads to increased albedo and decrease in temperature (negative correlation). As a result, convection, cloudiness, and precipitation decrease, which causes a further increase in albedo. The theoretical explanation of positive albedo–temperature feedback, on the basis of the thermal balance equation, is given by Idso (1981). Later, a linear statistical model of generation of positive feedback was suggested by Avissar and Pielke (1989).

The regional positive albedo–precipitation feedback or the regional albedo desertification mechanism (Charney) may contribute to human-induced desertification. Note that a mesoscale albedo desertification mechanism was suggested one year earlier by Otterman (1974). Its understanding is important for monitoring of desertification, because the duration of the prevalence of radiation plays an important role in maintaining desertification.

According to Wendler and Eaton (1983) and Zolotokrylin (1986), the albedo feedback mechanism reveals itself only in circumstances of highly reduced vegetation, which is typical of deserts and less typical of semideserts. In areas where phytomass storage is larger than in deserts, however, radiation is suppressed by evapotranspiration. This develops a negative albedo–precipitation feedback, which impedes the desertification process.

In situ observations of albedo and surface temperature were first published in the 1980s, followed by publication of remote sensing observations. Not only did these publications indicate a temperature drop with albedo increase in deserts, but also some temperature increase in highly developed vegetation during anthropogenic or natural reduction (Wendler and Eaton 1983; Goward et al. 1985; Otterman and Tucker 1985; Zolotokrylin 1986; Vukovich et al. 1987; Seguin et al. 1989; Menenti et al. 1988). Other investigators (Bryant et al. 1990; Michalek et al. 2001) who used study sites along the Arizona–Sonora border also found significant differences in vegetation cover, albedo, and surface temperature in small field plots over relatively large areas. These results suggest that it may be difficult to assess whether there are significant differences in biophysical properties of semiarid grassland between United States and Mexico in the vicinity of the border.

An in-depth analysis of in situ, helicopter, and satellite data gives insight into how the albedo-to-temperature ratio varies in deserts and semideserts of different dry regions of Asia and Africa (Zolotokrylin 2003). Desertification is essentially the interaction of two factors: environmental effects (aridization) and human-induced activities (degradation of vegetation). Human-induced degradation is caused by excessive grazing and other anthropogenic impacts. All these processes constitute yet another source of aridization.

The main factor that modulates aridization is regional climate. The varying vegetation plays the most important role in the aridization process, especially within the areas of human-induced degradation. In turn, variability of vegetation leads to mesoscale patchiness of albedo. The albedo of degraded areas increases while the surface temperature drops. This is typical for surface temperature modulation by radiation. According to Otterman (1974), the mesoscale convection over these areas weakens, which causes a drop in precipitation rate and an increase in albedo. The aridization scale changes as well. In addition to the regional scale, mesoscale aridization occurs, which is caused by radiation-induced modulation of surface temperature within the degraded areas.

The mechanism of mesoscale aridization is essentially the positive albedo–precipitation feedback, which decreases local precipitation; hence, to accomplish the task of monitoring desertification, one identifies the area of prevalence of radiation-induced temperature modulation. Therefore, it is important to determine the vegetation condition for which the prevalence of radiation-induced temperature modulation takes place.

Threshold values of the normalized difference vegetation index (NDVI) may be used as an indicator of switches in surface temperature–modulating factors within a seasonal cycle (Zolotokrylin 2003). NDVI is a satellite-based, remotely sensed measure of the “greenness” of the vegetation cover. Its use and applications simplify the monitoring of desertification and restrict it to the NDVI data from Advanced Very High Resolution Radiometer (AVHRR) and Moderate Resolution Imaging Spectroradiometer (MODIS) (Watts et al. 2007; Vivoni et al. 2008; Mendez-Barroso et al. 2009; Lizárraga-Celaya et al. 2010).

Therefore, this study focuses on seasonal cycles of parameters that modulate surface temperature in the Sonoran Desert in northwestern Mexico. Understanding of this process is important for monitoring desertification and determining the threshold value of AVHRR and MODIS NDVI on the basis of the dominance of radiation surface temperature regulation in relation to evapotranspiration. This is so because, in a given year, the time span during which the radiation factor predominates is important for the desertification process.

2. Study area and data

The Sonoran Desert is one of the wettest and hottest of the four deserts in North America (Chihuahuan, Great Basin, Mojave, and Sonoran). It covers a large area in the states of Baja California, Baja California Sur, and Sonora in Mexico as well as parts of Arizona and California in the United States (Shreve and Wiggins 1964; Weiss and Overpeck 2005).

The study area includes the Sonoran Desert and adjacent areas (Fig. 1) within a range of 23°–35°N and 109°–117°W. Our focus is on the Mexican high-plains part of the region and, in particular, the rectangular study areas 30°–31°N, 112°–113°W and 31°–32°N, 113°–114°W in the state of Sonora, which extends to the Gulf of California. The study area is dominated by desert shrub (Balling et al. 1998). Figure 2 illustrated the typical shrub vegetation that predominates in the two study areas.

Fig. 1.

The Sonoran Desert study area, displaying the location of 40 weather stations. The larger area includes part of the states of Sonora and Baja California (29°–32°N, 111°–116.5°W). The two smaller study areas are 30°–31°N, 112°–113°W and 31°–32°N, 113°–114°W. The numbers correspond to the codes for stations Sonoyta (26096) and Pitiquito (26093). The irregular dotted area in the inset represents the extent of the Sonoran Desert.

Fig. 1.

The Sonoran Desert study area, displaying the location of 40 weather stations. The larger area includes part of the states of Sonora and Baja California (29°–32°N, 111°–116.5°W). The two smaller study areas are 30°–31°N, 112°–113°W and 31°–32°N, 113°–114°W. The numbers correspond to the codes for stations Sonoyta (26096) and Pitiquito (26093). The irregular dotted area in the inset represents the extent of the Sonoran Desert.

Fig. 2.

Typical scrub vegetation in the Sonoran Desert (photograph taken by E. Alcántara Razo, M.Sc.).

Fig. 2.

Typical scrub vegetation in the Sonoran Desert (photograph taken by E. Alcántara Razo, M.Sc.).

Precipitation in the Sonoran Desert is 100–300 mm each year, mostly as rain. Rainfall is primarily received during summer, and so there is a strong seasonal signature; its interannual variability is high. Daytime temperatures can reach or exceed 40°C during May–September. There are two distinct rainy periods: winter (December–March) and summer (July–September), both determined by regional weather circulation and climatic variations (Bryant et al. 1990). The most significant feature of the summer season is the abrupt change in moisture conditions from hot and dry in May and June to relatively cool and rainy in July. July–September contribute 40%–60% of the annual precipitation, with a maximum in August on the mainland and in August and September near the coast of the Baja California Peninsula (Vivoni et al. 2008). Maximum precipitation in August is near the coast on the continental side, and this has been documented as a regular feature of the monsoon circulation in western Mexico (Brito-Castillo et al. 2010). Winter rainfall is associated with passing midlatitude systems and their fronts over the North American continent (Jáuregui 1995). The humid airflow transports the showers that are mainly caused by the monsoon system from the tropical Pacific Ocean and the Gulfs of Mexico and California to the Mexican territory (Douglas et al. 1993; Higgins et al. 2003). Dry periods occur in early summer (May–June) and early autumn.

Mean NDVI and calibrated albedos from AVHRR channels 1 and 2 and calibrated temperatures from AVHRR channels 4 and 5, with a 0.15° spatial resolution, collected from the National Oceanic and Atmospheric Administration (NOAA)-9 and NOAA-11 satellites (April 1985–September 1994) were used in this study (Gutman et al. 1995). Discontinuities and residual trends can be traced in time series of NDVI and temperature from channel 4. Discontinuities resulted from the switch from NOAA-9 to NOAA-11 in 1988 and the Mount Pinatubo eruption in 1991. Trends are a combined effect of satellite orbit drift and a possible persistent error in postlaunch calibration of channels. The orbit drift affects the solar and thermal IR channels through systematic variation of illumination geometry and diurnal heating/cooling of the surface and atmosphere, respectively. We also used monthly means of albedo and surface temperature data (Global AVHRR-Derived Land Climatology) prepared by the NOAA National Environmental Satellite, Data, and Information Service and the National Geophysical Data Center. This is one of the few climatological datasets that are brought up to being practically usable and are considered to be a reference because of their meticulous preparation. Ground verification of these albedo and temperature data, as well as comparisons with helicopter data, were carried out in the deserts of central Asia (Zolotokrylin 2003).

In addition, we used newer data distributed by the U.S. Geological Survey/National Aeronautics and Space Administration Land Processes Distributed Active Archive Center (LP DAAC; see https://lpdaac.usgs.gov/ for information and data descriptions)—to be specific, the MODIS products from 2005 to 2009—to obtain new results and to compare them with AVHRR data. Some of the actual data also can be downloaded from the Oak Ridge National Laboratory (daac.ornl.gov). The following three MODIS Land Subset Products were used.

  1. The first product is version 5 of the “MODIS/Terra Land Surface Temperature and Emissivity (LST/E) Monthly L3 Global 0.05Deg CMG” (here L3 denotes the processing level and CMG indicates climate modeling grid). Per the LP DAAC, “[t]he MODIS/Terra Land Surface Temperature and Emissivity products provide per-pixel temperature and emissivity values in a sequence of swath-based to grid-based global products. The MODIS/Terra LST/E Monthly L3 Global 0.05Deg CMG (short name: MOD11C3) is configured on a 0.05° latitude–longitude CMG.”

  2. The second product is version 5 of the MODIS/Terra “Vegetation Indices Monthly L3 Global 0.05Deg CMG” (short name: MOD13C2), which contains values of the normalized difference vegetation index (NDVI). Per the LP DAAC, “[g]lobal MODIS vegetation indices are designed to provide consistent spatial and temporal comparisons of vegetation conditions. Blue, red, and near-infrared reflectance, centered at 470-nanometers, 648-nanometers, and 848-nanometers, respectively, are used to determine the MODIS daily vegetation indices. The MODIS NDVI complements NOAA’s AVHRR NDVI products and provides continuity for historical applications. The MODIS NDVI product contains atmospherically corrected bi-directional surface reflectance masked for water, clouds, and cloud shadows…Global MOD13C2 data are cloud-free spatial composites of the gridded 16-day 1-kilometer MOD13A2, and are provided monthly as a level-3 product projected on a 0.05 degree (5600-meter) geographic CMG. Cloud-free global coverage is achieved by replacing clouds with the historical MODIS time series climatology record.”

  3. The third product is version 5 of the Terra/Aqua combined global “BRDF-Albedo Model Parameters 16-day L3 0.05Deg CMG” (here BRDF is bidirectional reflectance distribution function), which contains calculated albedo. Per the LP DAAC, “[t]he MODIS BRDF/Albedo Model Parameters product (MCD43C1) contains the weighting parameters for the models used to derive the albedo and NBAR [nadir BRDF-adjusted reflectance] products (MCD43C3 and MCD43C4). The models support the spatial relationship and parameter characterization best describing the differences in radiation due to the scattering (anisotropy) of each pixel, relying on multi-date, atmospherically corrected, cloud-cleared input data measured over 16-day periods.”

Surface albedo was computed using the data sensed in seven spectral bands from 0.47 to 1.24 μm. The monthly means of albedo were computed by averaging the two 16-day averages with starting days as follows: May (yeardays 121 and 137), June (yeardays 153 and 169), July (yeardays 185 and 201), and August (yeardays 217 and 233).

Climate data for the area were assembled by the Servicio Meteorologico Nacional of Mexico and were further processed with a program developed by the Instituto Mexicano de Tecnología del Agua (ERIC 2008) containing precipitation and daily maximum and minimum temperatures, along with totals, for many stations throughout Sonora. These daily data had undergone quality-assurance testing in Mexico and should be of reasonable quality for use in this study. The intra-annual cycle of annual monthly mean sums of precipitation was computed from monthly mean sums of precipitation at each station. This procedure was done for two periods: 1985–94 and 2005–09. For 1985–94, we used data from 23 stations (Fig. 1) within the region 27°–33°N, 108°–118°W (ERIC 2008).

The annual monthly means from each station were linearly interpolated into the 0.5° × 0.5° grid nodes within the region 29°–32°N, 111°–115°W. The annual monthly mean averaged over this region was computed from the gridded data, excluding the nodes whose elevation exceeded 500 m. The same method was used to estimate the mean precipitation within the area 30°–31°N, 112°–113°W, for which we did not have precipitation data (Fig. 3a).

Fig. 3.

The intra-annual cycle of average monthly precipitation sums from the large study area (solid line) and the small study areas (dashed lines) for (a) 1985–94 and (b) 2005–09.

Fig. 3.

The intra-annual cycle of average monthly precipitation sums from the large study area (solid line) and the small study areas (dashed lines) for (a) 1985–94 and (b) 2005–09.

A similar method for computing precipitation was used for 40 stations in the 2005–09 period within the region 26.8°–32.4°N, 109°–115°W (Fig. 1). Data between 2005 and 2008 were obtained from the ERIC (2008) database. The data after 2008 were provided by the Servicio Meteorologico Nacional of Mexico. Next, within the region 29°–32°N, 111°–115°W, monthly means of precipitation were estimated from the gridded data. The intra-annual cycle of annual monthly mean sums of precipitation within the regions 30°–31°N, 112°–113°W and 31°–32°N, 113°–114°W was defined by the data collected at stations 26093 and 26096 (Fig. 3b).

Comparison of Figs. 3a and 3b revealed that precipitation during May–September 2005–09 within region 29°–32°N, 111°–115°W was higher (131 mm) than for the same months in 1985–94 (118 mm). For July–August 2005–09, rainfall was higher at station 26093 (30°–31°N, 112°–113°W; 90 mm) than at station 26096 (31°–32°N, 113°–114°W; 64 mm). As we shall see, the differences in rainfall patterns may be vital in interpreting the relationship between albedo and surface temperature.

3. Results

The main features of the ratio between albedo and surface temperature are discussed by analyzing monthly means (albedo, temperature, and NDVI) in Sonora within the two box-shaped areas: 30°–31°N, 112°–113°W and 31°–32°N, 113°–114°W.

a. Ratio between monthly means of albedo and monthly surface temperature

Month-by-month inspection of data helps to elucidate the relationship between albedo and surface temperature for 1985–94 (Fig. 4) and 2005–09 (Figs. 5 and 6 ). In dry months of 1985–94 (May and June), radiation modulating the surface temperature is predominant (Figs. 4a,b). In the figures, there is a negative correlation between these parameters (−0.75), and the AVHRR NDVI values decline as albedo increases and temperature falls.

Fig. 4.

Relationship between albedo and surface temperature in the Sonoran Desert (30°–31°N, 112°–113°W) during a year: (a) May, (b) June, (c) July, and (d) August. The points are labeled with the AVHRR NDVI values and refer to the average over an area of 16 km × 16 km (grid point). The AVHRR data are averaged over the period 1985–94 (Gutman et al. 1995). A box of 1° × 1° has a 16 km × 16 km grid or 39 values.

Fig. 4.

Relationship between albedo and surface temperature in the Sonoran Desert (30°–31°N, 112°–113°W) during a year: (a) May, (b) June, (c) July, and (d) August. The points are labeled with the AVHRR NDVI values and refer to the average over an area of 16 km × 16 km (grid point). The AVHRR data are averaged over the period 1985–94 (Gutman et al. 1995). A box of 1° × 1° has a 16 km × 16 km grid or 39 values.

Fig. 5.

Relationship between albedo and surface temperature in the Sonoran Desert. Points are labeled with the MODIS NDVI values. The MODIS data are averaged over the period 2005–09. Shown are (a) May, (b) June, (c) July, (d) August, and (e) September for the study area of 30°–31°N, 112°–113°W.

Fig. 5.

Relationship between albedo and surface temperature in the Sonoran Desert. Points are labeled with the MODIS NDVI values. The MODIS data are averaged over the period 2005–09. Shown are (a) May, (b) June, (c) July, (d) August, and (e) September for the study area of 30°–31°N, 112°–113°W.

Fig. 6.

As in Fig. 5, but for the study area of 31°–32°N, 113°–114°W.

Fig. 6.

As in Fig. 5, but for the study area of 31°–32°N, 113°–114°W.

The opposite occurs in the humid months of July and August (Figs. 4c,d). The response of vegetation occurred entirely (in a week or so) within July and August; the AVHRR NDVI increased over most of the territory, peaking at 0.19 (Figs. 4c,d); that is, the transpiring green phytomass grew in quantity. The portion of the radiation balance consumed by evapotranspiration becomes larger while the portion of the energy that goes into turbulent heating of the surface boundary layer decreases. At the end of the period, modulation of the surface temperature by evapotranspiration dominates over most of the study area; the correlation between albedo and surface temperature is positive (left side of the approximation curve in Fig. 4d).

At the same time, some areas with scant green phytomass remained in the study area (where AVHRR NDVI did not exceed 0.09). This corresponded to the rightmost approximation point on the curve (right side of Fig. 4d). Most likely, this is a manifestation of evapotranspiration and radiation. In September, when AVHRR NDVI dropped to 0.50, the albedo–temperature correlation weakened, as compared with August when the correlation was 0.90; yet the positive sign of the correlation remained.

Hence, during May–June 1985–2004, radiation was predominant as a surface temperature–modulating factor. It was followed by evapotranspiration in August, when rainfall peaked. In other words, the conditions became favorable for weaker climatic desertification during the humid season.

Because rainfall within the Sonoran Desert varies considerably across the region and over time, it would be unreasonable to expect that the pattern would occur in the same region (30°–31°N, 112°–113°W) in 2005–09 a second time. This statement is illustrated by Fig. 5, which shows a weak correlation between albedo and surface temperature in May–June and an increased positive correlation from July to September. Unlike the period of 1985–94, this example does not reveal the dominant role of radiation during the dry season. In May–July, correlation is weak, indicating a balance between the radiation and evapotranspiration factors. Instead, we noticed a gradual dominance of evapotranspiration over radiation during the humid season. Therefore, we suggest that the process of climatic desertification within the study area weakened from wetter conditions in 2005–09 when compared with 1985–94 (Fig. 3).

In the adjoining region (31°–32°N, 113°–114°W), where rainfall in 2005–09 was less than the other box (Fig. 3b), we see from Fig. 6 that radiation was dominant and that its role diminished from May to August; the negative correlation decreased from 0.69 to 0.48. Evapotranspiration did not become dominant until September. The one-month lag was a consequence of the shift in the rainfall peak to September within region 29°–32°N, 111°–115°W and of precipitation in region 31°–32°N, 113°–114°W being close to its maximum in September (Fig. 3).

b. Duration of periods with predominant surface temperature modulation

It is possible to identify three periods in which radiation, evapotranspiration, or radiation–evapotranspiration prevails. Radiation was predominant in April–June, prior to the summer rainfall. During this dry period, the amount of phytomass declines to a minimum and the surface temperature correlates with albedo. This creates favorable conditions for generation and preservation of desert climate, until summer rainfall creates a new situation.

Modulation by evapotranspiration dominates during a short time period (August–September) when soil moisture content and water storage capacity of plants markedly increased, as a result of summer precipitation. Vivoni et al. (2008) showed seasonal progression of the link between evapotranspiration and soil moisture in northwestern Mexico. Greater green phytomass adds to the loss of moisture from evapotranspiration. This reduces the portion of energy spent on turbulent heating of the bottom of the atmosphere and heating of the ground surface. Radiation is also reduced by the increase in plant cover (leaves) and by evapotranspiration, but it is still important. During this season, the climate of the area resembles semidesert conditions. Where rocky surfaces are common and the vegetation is scant, however, local desert climate was preserved.

In the remaining part of the year (October–March), neither radiation nor evapotranspiration prevails (not shown). The latter is mainly characterized by evaporation from soil. Reduced solar radiation diminishes the impact of radiation, but evapotranspiration is boosted by increased precipitation in winter. Hence, the equilibrium between temperature-modulating factors is maintained while the correlation between albedo and temperature drops.

c. Average threshold NDVI: Lower limits allow possibility of predominance of radiation when the values are less than this limit

For the purpose of monitoring, it is important to determine the threshold value of monthly mean AVHRR and MODIS NDVI, which is a characteristic of predominance of radiation as a factor in the Sonoran Desert. This value can vary, depending on the spatial scale. We considered histograms of AVHRR and MODIS NDVI (Fig. 7) in May, June, and July for the regions 30°–31°N, 112°–113°W and 31°–32°N, 113°–114°. From the comparative analysis of the histograms, average AVHRR NDVI decreased from 0.103 ± 0.013 in May to 0.086 ± 0.01 in June to 0.08 ± 0.01 in July. This situation is possible if the effective precipitation, that is, the precipitation sufficient for growth of a grass cover, falls mostly in middle to late July. Because a grass cover develops slowly, for example, a month in deserts of central Asia (Zolotokrylin 2003), greening is observed by satellites no sooner than August. As a result, the threshold value of 0.08 ± 0.01 is acceptable for this region. At lower values, temperature-modulating radiation becomes predominant. It is evident that the estimate of the threshold value from MODIS NDVI must be higher than that of AVHRR NDVI. For MODIS NDVI, this estimate ranges between 0.10 and 0.11 (Fig. 8).

Fig. 7.

Histogram of the AVHRR NDVI in the Sonoran Desert study area of 30°–31°N, 112°–113°W for (a) May, (b) June, and (c) July. The period is 1984–95.

Fig. 7.

Histogram of the AVHRR NDVI in the Sonoran Desert study area of 30°–31°N, 112°–113°W for (a) May, (b) June, and (c) July. The period is 1984–95.

Fig. 8.

Histogram of the MODIS NDVI in the Sonoran Desert for May for the period 2005–09 for (a) a study area of 30°–31°N, 112°–113°W and (b) a study area of 31°–32°N, 113°–114°W.

Fig. 8.

Histogram of the MODIS NDVI in the Sonoran Desert for May for the period 2005–09 for (a) a study area of 30°–31°N, 112°–113°W and (b) a study area of 31°–32°N, 113°–114°W.

It is significant that the part of the Sonoran Desert where AVHRR NDVI is lower than or equal to the threshold value of 0.08 peaks in July. In the most humid month (August), this area begins to shrink. Hence, it is safe to assume that in these regions desertification intensified through albedo–precipitation positive feedback.

The change in coverage with MODIS NDVI < 0.10 was analyzed within the two regions (30°–31°N, 112°–113°W and 31°–32°N, 113°–114°W). As expected, the maximum for the area 31°–32°N, 113°–114°W occurred in August. At the same time, within the other area, 30°–31°N, 112°–113°W, conditions remained stable.

4. Discussion

There are numerous attempts to use AVHRR NDVI data for monitoring desertification and land cover (Maselli et al. 1998; Karnieli and Dall’Olmo, 2003; Piao et al. 2005; Helldén and Tottrup 2008). Maselli et al. (1998) develop and test a procedure that is based on classification and regression analysis techniques for generating an NDVI dataset with the spatial resolution of Landsat Thematic Mapperimages and the temporal resolution of NOAA AVHRR maximum-value composites. The usefulness of combining scatterplot analysis of NDVI data and land surface temperature for monitoring desertification, drought, and phenology in the Sinai and Negev was provided by Karnieli and Dall’Olmo (2003). This encouraged researchers to use the relationship between NDVI and rainfall for monitoring desertification—for example, by using a linear relationship between average rainy-season NDVI and annual precipitation. Tucker et al. (1991) and Tucker and Nicholson (1999) propose that the threshold NDVI value corresponding to an annual rainfall of 200 mm delineates the boundary of the Sahara. Later, the threshold NDVI values or the boundaries between arid and semiarid regions (NDVI = 0.219) and semiarid and subhumid regions (NDVI = 0.323) were determined for dry lands in China (Piao et al. 2005). Helldén and Tottrup (2008) use time series of integrated and standardized annual NDVI anomalies to generate and compare a corresponding rainfall dataset (1981–2003). These were used to monitor desertification on regional and global scales.

Inferences about seasonal variability of temperature-modulating factors of dry lands may become a background for understanding dynamics of desertification and monitoring. We suggest monitoring desertification of the Sonoran region by tracking changes in the extent of arid area where AVHRR NDVI < 0.08 and MODIS NDVI < 0.10, a characteristic of desertification. An extended increase of dry areas is an indicator of strengthening of desertification, whereas an extended decrease of dry areas is suggestive of its weakening.

The main contribution of our study was to determine the threshold NDVI for monitoring desertification on the basis of radiation for regulating temperature of the surface and the definition of threshold values for AVHRR and MODIS NDVI, which is an objective measure of the dominance of radiation surface temperature regulation in relation to evapotranspiration. In this sense, they are universal factors.

It is well known that current desertification is a result of the interaction of aridity and anthropogenic degradation of arid lands. Aridity drains a territory of vegetation, mainly from decreasing regional atmospheric moisture and more-frequent droughts. When degradation of vegetation from aridity and human activities reaches a threshold value, which happens when green biomass is ~0.5 t ha−1 dry weight for desert conditions in Asia and Africa (Zolotokrylin 2003), it quickly increases the role of radiation regulation of surface temperature relative to that of evapotranspiration. The dominance of radiation control is characterized by an increased negative correlation between albedo and surface temperature, which also serves as a sign of the formation of a regional (or mesoscale) positive feedback, albedo–precipitation, and a supporting desertification. In other words, the value of biomass or its surrogate (AVHRR or MODIS NDVI) is considered the threshold by changing the sign of correlation between albedo and surface temperature. If large areas (more than mesoscale) with AVHRR or MODIS NDVI are below the threshold, we assume that these areas are affected by desertification. Desertification is supported by positive feedback of albedo–precipitation. The threshold value of AVHRR or MODIS NDVI is determined by the frequency distribution of the probabilities of NDVI (the smallest scale of the distribution) in the case of a negative correlation between albedo and surface temperature.

Two aspects are important here. The first is related to the investigation of a switching mechanism between the temperature-modulating factors, on the basis of the energy balance. Our study states that the switch is possible at a certain value of NDVI. To better understand the switching mechanism, more specific experiments need to be conducted that are based on energy-balance studies and calculation of radiation and vertical turbulent fluxes of heat and moisture in the atmosphere boundary layer and soil.

The second aspect of the study is related to the switch indicator. In strict terms, green phytomass may serve as such an indicator, that is, the amount of assimilating vegetation per square unit. The work on formulation and definition of such an indicator will take many years of multiple experiments among different types of vegetation. This indicator can be viewed as a base (reference), which will subsequently be transformed into a remote NDVI indicator because green phytomass and NDVI are highly correlated (Tucker et al. 1985). A reference indicator is also essential for applying corrections to the NDVI data obtained by different radiometers (AVHRR and MODIS). One of the first and foremost tasks is to obtain uniform NDVI time series. This issue arises because the MODIS radiometer replaced the AVHRR radiometer on new satellite series.

5. Summary

The analysis of synchronous time series of albedo, surface temperature, and AVHRR and MODIS NDVI has shown that the dominating temperature-modulating factors can switch within the year in the study area. Radiation is dominant in dry months, when the threshold values are 0.08 for AVHRR NDVI and 0.10 for MODIS NDVI in most parts of the study area.

In these circumstances, the surface temperature is negatively correlated with albedo. This can cause generation of positive albedo–precipitation feedback, which in turn contributes to the desertification process.

The evapotranspiration temperature-modulating factor prevails in the most humid months, August–September. Positive correlation between albedo and temperature, which occurs at this time, creates conditions for generation of negative albedo–precipitation feedback, which impedes desertification.

In the autumn and winter months, equilibrium is achieved between the radiation and evapotranspiration types of temperature modulation. During these periods, the correlation between albedo and surface temperature is weak.

It is important for the monitoring of desertification 1) to estimate the extent of the area’s AVHRR and MODIS NDVI values that are less than the threshold value both in dry and humid months and 2) to outline the trend of the changes. This will be the scope of a separate paper.

Acknowledgments

The authors appreciate the comments of the anonymous reviewers. Special thanks are given to Ira Fogel for editorial revisions. This study was funded by University of Guadalajara grant PROINPEP 2010, Consejo Nacional de Ciencia y Tecnología of Mexico (CONACYT Grants J50757-F and 83433-CB), and CIBNOR Grants PC 1.0 and PC 0.3. Thanks are given to Enrique Vivoni for his comments to the previous version of this manuscript and to the Hydrometeorological Disasters and Climate Network (REDESCLIM) of CONACYT for its partial support.

REFERENCES

REFERENCES
Avissar
,
R.
, and
R. A.
Pielke
,
1989
:
A parameterization of heterogeneous land surface for atmospheric numerical models and its impact on regional meteorology
.
Mon. Wea. Rev.
,
117
,
2113
2136
.
Balling
,
R. C.
, Jr.
,
J. M.
Klopatec
,
M. H.
Hilderbrant
,
C. K.
Moritz
, and
C. J.
Watts
,
1998
:
Impacts of land degradation on historical temperature records from the Sonoran Desert
.
Climatic Change
,
40
,
669
681
.
Becker
,
F.
,
H. J.
Bolle
, and
P. R.
Rowntree
,
1988
:
The International Satellite Land-Surface Climatology Project. Free University of Berlin Press, 100 pp
.
Brito-Castillo
,
L.
,
E. R.
Vivoni
,
D. J.
Gochis
,
A.
Filonov
,
I.
Tereshchenko
, and
C.
Monzon
,
2010
:
An anomaly in the occurrence of the month of maximum precipitation distribution in northwest Mexico
.
J. Arid Environ.
,
74
,
531
539
.
Bryant
,
N. A.
,
L. F.
Jonson
,
A. J.
Brazel
,
R. C.
Balling
,
C. F.
Hutchinson
, and
L. R.
Beck
,
1990
:
Measuring the effect of overgrazing in the Sonoran Desert
.
Climatic Change
,
17
,
243
264
.
Charney
,
J. G.
,
1975
:
Dynamics of deserts and drought in Sahel
.
Quart. J. Roy. Meteor. Soc.
,
101
,
193
202
.
Douglas
,
M. W.
,
R. A.
Maddox
,
K.
Howard
, and
S.
Reyes
,
1993
:
The Mexican monsoon
.
J. Climate
,
6
,
1665
1677
.
ERIC
,
2008
:
Extractor rápido de información climatológica (Rapid extractor of climatological information). Instituto Mexicano de Tecnología del Agua, Morelos, Mexico, CD-ROM, v4. [Available online at http://www.imta.gob.mx/index.php?option=com_virtuemart&page=shop.browse&category_id=28&Itemid=58.]
Goward
,
S. N.
,
G. D.
Cruickshanks
, and
A. S.
Hope
,
1985
:
Observed relation between thermal emission and reflected spectral radiance of a complex vegetated landscape
.
Remote Sens. Environ.
,
18
,
137
146
.
Gutman
,
G.
,
D.
Tarpley
,
A.
Ignatov
, and
S.
Olson
,
1995
:
The enhanced NOAA global land data set from the Advanced Very High Resolution Radiometer
.
Bull. Amer. Meteor. Soc.
,
76
,
1141
1156
.
Helldén
,
U.
, and
C.
Tottrup
,
2008
:
Regional desertification: A global synthesis
.
Global Planet. Change
,
64
,
169
176
,
doi:10.1016/j.gloplacha.2008.10.006
.
Higgins
,
R. W.
, and
Coauthors
,
2003
:
Progress in pan American CLIVAR research: The North American monsoon system
.
Atmósfera
,
16
,
29
65
.
Idso
,
S. B.
,
1977
:
A note on some recently proposed mechanisms of genesis of deserts
.
Quart. J. Roy. Meteor. Soc.
,
103
,
369
370
.
Idso
,
S. B.
,
1981
:
Surface energy balance and genesis of deserts
.
Arch. Meteor. Geophys. Bioklimatol.
,
30
,
253
260
.
Jackson
,
R. D.
, and
S. B.
Idso
,
1975
:
Surface albedo and desertification
.
Science
,
189
,
369
370
.
Jáuregui
,
E.
,
1995
:
Rainfall fluctuations and tropical storm activity in Mexico
.
Erkunde
,
49
,
39
48
.
Karnieli
,
A.
, and
G.
Dall’Olmo
,
2003
:
Remote-sensing monitoring of desertification, phenology, and droughts
.
Manage. Environ. Qual.
,
14
,
22
38
.
Lizárraga-Celaya
,
C.
,
C. J.
Watts
,
J. C.
Rodriguez
,
J.
Garatuza-Payan
,
R. L.
Scott
, and
J.
Sáiz-Hernández
,
2010
:
Spatio-temporal variation in surface characteristics over the North American monsoon region
.
J. Arid Environ.
,
74
,
540
548
.
Maselli
,
F.
,
M. A.
Gilbert
, and
C.
Conese
,
1998
:
Integration of high and low resolution NDVI data for monitoring vegetation in Mediterranean environments
.
Remote Sens. Environ.
,
63
,
208
218
.
Mendez-Barroso
,
L. A.
,
E. R.
Vivoni
,
C. J.
Watts
, and
J. C.
Rodriguez
,
2009
:
Seasonal and interannual relations between precipitation, surface soil moisture and vegetation dynamics in the North American monsoon region
.
J. Hydrol.
,
377
,
59
70
.
Menenti
,
M.
,
W.
Bastiaassen
,
D.
van Eick
, and
M. A.
Abd el Karim
,
1988
:
Linear relationships between surface reflectance and temperature and their application to map actual evaporation of groundwater
.
Adv. Space Res.
,
9
,
165
176
.
Michalek
,
J. L.
,
J. E.
Colwell
,
N. E. G.
Roller
,
N. A.
Miller
,
E. S.
Kasishke
, and
W. H.
Schlezenger
,
2001
:
Satellite measurements of albedo and radiant temperature from semi-desert grassland along the Arizona/Sonora border
.
Climatic Change
,
48
,
417
425
.
Otterman
,
J.
,
1974
:
Baring high-albedo soils by overgrazing: Hypothesized desertification mechanism
.
Science
,
186
,
531
533
.
Otterman
,
J.
, and
C. J.
Tucker
,
1985
:
Satellite measurements of surface albedo and temperatures in semi-desert
.
J. Climate Appl. Meteor.
,
24
,
228
235
.
Piao
,
S.
,
J.
Fabg
,
H.
Liu
, and
B.
Zhu
,
2005
:
NDVI-indicated decline in desertification in China in the past two decades
.
Geophys. Res. Lett.
,
32
,
L06402
,
doi:10.1029/2004GL021764
.
Ripley
,
E. A.
,
1976a
:
Drought in the Sahara: Insufficient biogeophysical feedback?
Science
,
191
,
100
102
.
Ripley
,
E. A.
,
1976b
:
Comment on the paper “Dynamics of deserts and drought in Sahel” by J. G. Charney
.
Quart. J. Roy. Meteor. Soc.
,
102
,
466
467
.
Seguin
,
B.
,
E.
Assad
,
P.
Freteaud
,
J
.
Imbernon
,
Y.
Kerr
, and
J. P.
Lagouarde
,
1989
:
Use of meteorological satellites for water balance monitoring in Sahelian regions. Int. J. Remote Sens., 10, 1101–1117
.
Shreve
,
F.
, and
I. L.
Wiggins
,
1964
:
Vegetation and Flora of the Sonoran Desert. Stanford University Press, 1740 pp
.
Thornthwaite
,
C. W.
,
1948
:
An approach toward a rational classification of climate
.
Geogr. Rev.
,
38
,
55
94
.
Tucker
,
C. J.
, and
S. E.
Nicholson
,
1999
:
Variations in the size of the Sahara Desert from 1980 to 1997
.
Ambio
,
28
,
587
591
.
Tucker
,
C. J.
,
C. L.
Vanpraet
,
M. J.
Sharman
, and
G.
Van Ittersum
,
1985
:
Satellite remote sensing of total herbaceous biomass production in the Senegalese Sahel: 1980–1984
.
Remote Sens. Environ.
,
17
,
233
249
.
Tucker
,
C. J.
,
H. E.
Dregne
, and
W. W.
Newcomb
,
1991
:
Expansion and contraction of the Sahara Desert from 1980 to 1990
.
Science
,
253
,
299
301
.
Vivoni
,
E. R.
,
H. A.
Moreno
,
G.
Mascaro
,
J. C.
Rodriguez
,
C. J.
Watss
,
J.
Garatuza-Payan
, and
R. L.
Scott
,
2008
:
Observed relation between evapotranspiration and soil moisture in the North American monsoon region
.
Geophys. Res. Lett.
,
35
,
L22403
,
doi:10.1029/2008GL036001
.
Vukovich
F. M.
,
D. L.
Toll
, and
R. E.
Murphy
1987
:
Surface temperature and albedo relationship in Senegal derived from NOAA-7 satellite data
.
Remote Sens. Environ.
,
22
,
413
421
.
Watts
,
C. J.
,
R. L.
Scott
,
J.
Garatuza-Payan
,
J. C.
Rodriguez
,
J. H.
Pruegger
,
W. P.
Kustas
, and
M.
Douglas
,
2007
:
Changes in vegetation condition and surface fluxes during NAME 2004
.
J. Climate
,
20
,
1810
1820
.
Weiss
,
J. L.
, and
J. T.
Overpeck
,
2005
:
Is the Sonoran Desert losing its cool?
Global Change Biol.
,
11
,
2065
2077
.
Wendler
,
G.
, and
F.
Eaton
,
1983
:
On the desertification of the Sahel zone: Part 1. Ground observations
.
Climate Change
,
4
,
365
380
.
Zolotokrylin
,
A. N.
,
1986
:
The relationships between surface albedo and temperature in arid regions (in Russian). Data Meteor. Studies, 10, 38–45
.
Zolotokrylin
,
A. N.
,
2003
:
Climatic Desertification (in Russian). Nauka, 246 pp
.