Next Article in Journal
A Method of Chestnut Forest Identification Based on Time Series and Key Phenology from Sentinel-2
Previous Article in Journal
Scale Information Enhancement for Few-Shot Object Detection on Remote Sensing Images
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Remote Sensing
Volume 15
Issue 22
10.3390/rs15225373
remotesensing-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

The Relationship between Land Surface Temperature and Air Temperature in the Douro Demarcated Region, Portugal

by
Filipe Adão
*,
Helder Fraga
,
André Fonseca
,
Aureliano C. Malheiro
and
João A. Santos
Centre for the Research and Technology of Agro-Environmental and Biological Sciences (CITAB), University of Trás-os-Montes e Alto Douro (UTAD), Quinta de Prados, 5000-081 Vila Real, Portugal
*
Author to whom correspondence should be addressed.
Remote Sens. 2023, 15(22), 5373; https://doi.org/10.3390/rs15225373
Submission received: 20 October 2023 / Revised: 9 November 2023 / Accepted: 12 November 2023 / Published: 16 November 2023
(This article belongs to the Section Remote Sensing in Agriculture and Vegetation)
Browse Figures
Versions Notes

Abstract

:
Climatic studies of agricultural regions normally use gauge-based air temperature datasets, which are produced with interpolation methods. The informative quality of these datasets varies depending on the density of the weather stations in a particular region. A way to overcome this limitation is to use the land surface temperature calculated from satellite imagery. To show this, the MODIS land surface temperature was compared with the PTHRES gridded dataset for air temperature in the Douro Demarcated Region (Portugal) between the years 2002 and 2020. The MODIS land surface temperature was able to detect a more pronounced maritime–continental gradient, a higher lapse rate, and thermal inversions in valley areas in winter. This information could prove to be crucial for farmers looking to adapt their practices and crops to extreme events, such as heat waves or heavy frost. However, the use of land surface temperature in climate studies should consider the differences in air temperature, which, on some occasions and locations, can be up to ten degrees in the summer.

1. Introduction

The viability and sustainability of agriculture rely primarily on favorable climatic conditions for the proper phenological development and growth of existing crops [1]. Temperature is one of the most important climate variables in agriculture [2]. Furthermore, it is known that temperature-related phenomena, such as thermal inversions, can cause frost on crops and, consequently, partially ruin economic output [3]. Therefore, it is important, if not crucial, that existing crops are well adapted to their environment with particular concern for temperature.
Access to detailed temperature information is not always easy, however. Many studies looking at temperature in agricultural regions use gridded observational datasets [4,5,6,7,8,9]. Datasets like these are built with spatial and temporal interpolation methods using gauge-based data acquired at weather stations. The distribution of these stations normally has varying densities and, depending on topography and atmospheric circulation, the data can have a considerable bias towards the thermal gradients formulated by the interpolation methods. The effect of this limitation has been reported, and it normally results in lower representability of smaller-scale heterogeneities due to a lack of information [10,11,12]. Furthermore, studies typically look at the mean values of temperature in a given period calculated from monthly or annual datasets. Daily datasets are not often used due to the disparity in spatial and temporal construction and the quality control methodologies of daily acquired data [13]. The E-OBS [14] and Iberia01 [15] are two daily datasets that are available for Europe and meet the necessary criteria for reliability. However, the production of these datasets did not consider all of the available information to focus on the longest time series acquired [13]. Therefore, such gridded datasets are more appropriate for studies that focus on giving a general picture of a region than a detailed one.
To obtain a more detailed idea of temperature heterogeneities in a particular region, one could always increase the number of weather stations. However, this comes with great financial cost. Thus, an alternative solution has to be considered. One of those solutions is remote sensing data. This is data captured with spectrometers mounted on geostationary or polar satellites that measure the Earth’s radiation. Using algorithms that relate its spectral bands with temperature, one can derive the temperature at the surface of the Earth with great accuracy [16]. Such type of data has been successfully applied to climate studies [17], forestry and fires [18], and agrometeorology [19]. Although the land surface temperature (LST) does not translate directly to the air temperature measured by weather stations, it can still indicate how the temperature changes spatially and with time.
The objective of the current study is to demonstrate how the LST derived from remote sensing data can provide a clearer picture of temperature heterogeneities across an agricultural region, namely the Douro Demarcated Region (DDR). The DDR is located in northern Portugal and is an important region for the primary sector of the Portuguese economy. It is the main producing region of wines in the country [20] and is also a strong contributor to olive and almond production [8,21]. Furthermore, it has been classified by UNESCO as a world heritage site for its long-standing history in wine production and its natural wonder, which has resulted in growth for rural and agritourism [22,23]. It is, therefore, a region that warrants attention. The study was performed by comparing NASA’s Moderate Resolution Imaging Spectroradiometer (MODIS) LST satellite products with the PTHRES gridded dataset for the air temperature at two meters height (2mT) and analyzing the main differences. The two datasets were chosen due to their matching spatial resolution and temporal availability. The Iberia01 gridded dataset was also considered, but its spatial resolution does not match that of the MODIS LST, so it was discarded. Lastly, the previously mentioned atmospheric phenomenon of thermal inversion was investigated in the Vilariça Valley, a known area that is prone to frost events [24].

2. Materials and Methods

2.1. Study Area

The study focused on the DDR in northern Portugal (Figure 1). Its area is covered by around 18% of vineyards, which are mostly located on the steep Douro River (and its tributaries) margins [25]. Other orchards in the DDR include apple, olive, and almond trees [26]. These plantations are mixed in with forested areas in the western and central parts and sparsely vegetated areas in the eastern parts [27]. The geology of the DDR is mainly composed of schistose-layered rock and granitic outcrops. In terms of elevation, it peaks at around 1400 m and has its lowest point at 40 m [28].
The climate in the DDR is a warm temperate Mediterranean-type, which is characterized by moderately cool and rainy winters and hot and dry summers but with a noteworthy decrease in precipitation and an increase in continentality from west to east as we move away from the Atlantic coastline [29]. The topography is decisive in this regard, as the Marão and the Montemuro mountainous ranges to the west shelter humid air from the Atlantic Ocean from coming in at a constant rate throughout the year. Moreover, the various deep valleys and tall hills promote the formation of distinct features of the local climate, mainly due to mesoscale circulation systems within the Iberian Douro/Duero catchment (e.g., katabatic and anabatic winds). One example of such a valley is the already mentioned Vilariça Valley, where the formation of morning frost is common due to the thermal inversion of air masses [24]. This valley is represented in Figure 1 by the letters A (longitude: 7.11°W; latitude: 41.22°N) and B (longitude: −7.15°W; latitude: 41.24°N). This spatial variability allows for the specialized production of fruit and wine, with their properties differing between the different meso and microclimates.

2.2. MODIS Land Surface Temperature

MODIS is NASA observation equipment built to measure the Earth’s outgoing radiation from the visible to near-infrared and thermal infrared spectrum (0.4 to 14.4 µm). It is mounted on the Terra and Aqua satellites that have a low, circular sun-synchronous polar orbit with a period of 99 min [30]. The acquired imagery is used to make a series of products of surface temperature on land and water, with their application ranging from climate studies to energy–water cycles to vegetation water stress assessment, among others [31]. For this study, the MYD11_L2v061 product for LST was considered [32]. It is produced with Aqua satellite imagery via the generalized split-window algorithm and has a spatial resolution of 1 km [33]. It has been available since 2002 and is produced two times per day in the DDR: once around 1 p.m. and once around 2 a.m. Both available times will hereon be referred to as day and night times.
The LST products (hereon referred to as data) were retrieved from the NASA Earthdata search engine for the winter (December, January, and February) and summer (June, July, and August) seasons, at day and night times, in the period between 2002 and 2020. The spring and autumn seasons were not analyzed for the sake of conciseness. The data with a clear sky over the study area were selected, and, upon visual inspection, those which appeared distorted were discarded. In total, 120 raster image files were used for the winter (2.60% of the available images) and 330 raster image files were used for the summer (6.78% of the available images).
After the data retrieval and selection, the following processing steps were taken: firstly, the data were converted from their original HDF5 format to NetCDF format to be georeferenced. The conversion was performed using HEGtool software (v2.12) provided by NASA, choosing the nearest neighbor resampling and geographic projection. Secondly, the NetCDF converted files were reprojected to EPSG:4326–WGS84 and cut to an area around the DDR using the ‘gdalwarp’ raster program from the Geospatial Data Abstraction Library (GDAL) [34]. Thirdly, as some data still showed residual cloud cover, represented by isolated empty pixels, these were interpolated using the ‘gdal_fillnodata.py’ raster program, also from GDAL. For the winter, 0.42% of the daytime data and 1.12% of the nighttime data were interpolated. As for the summer, 0.02% of the daytime data and 1.18% of the nighttime data were interpolated. Fourthly, the mean of the LST data was calculated for each season and both day and night times, and the resulting rasters were cut into the DDR shape. Table 1 presents the minimum, mean, and maximum values of the mean LST in the DDR for the considered period. Finally, the DDR LST mean raster images were feature-scaled by subtracting the corresponding mean values from each of the pixel values.

2.3. PTHRES Air Temperature

The PTHRES air temperature is a climate, gridded dataset with approximately 1 km spatial resolution that provides information on daily minimum, mean, and maximum temperatures at any point in mainland Portugal at a height of two meters. It was produced from the coarser-resolution E-OBS gridded dataset provided by Copernicus using a downscaling two-step approach [35]. It is available for the same period as the MODIS LST (2002–2020) and seasons (winter and summer). The daytime temperatures were retrieved from PT.TX.HRES (maximum temperature), while the nighttime temperatures were retrieved from PT.TN.HRES (minimum temperature). These temperatures refer to different times of the day in comparison with the retrieved MODIS LST. Daytime 2mT should correspond to 1 p.m. in winter and 4 p.m. in summer, and nighttime 2mT should correspond to 6 a.m.
The processing steps applied to the PTHRES dataset mirrored those applied to the MODIS LST dataset. The only exceptions were the NetCDF conversion and the interpolation to fill in empty pixel cloud cover, which were not necessary. Table 2 presents the minimum, mean, and maximum values of the produced mean 2mT in the DDR in the considered period.

2.4. Analysis and Comparison of the MODIS Land Surface Temperature and PTHRES Air Temperature in the Douro Demarcated Region

The analysis of the temperature difference between MODIS LST and PTHRES 2mT in the DDR was achieved firstly by looking at the subtraction of the feature-scaled 2mT rasters from the LST rasters (LST–2mT) for each season and time of day. This delta allowed for an understanding of where the major feature-scaled temperature differences are distributed. Then, scatterplots of the LST versus 2mT, LST–2mT versus elevation, and LST and 2mT versus elevation were plotted, including linear regression models, to see the relationship between the different variables for each season and time of day. Furthermore, the temperature evolution of the LST and 2mT throughout the winter and summer for points A and B in the Vilariça Valley were analyzed. In this part of the analysis, the temperature is representative of the centered moving means for eleven days, which used the value for each day that was previously averaged across the considered period. Some days are missing as there were no LST data that passed the selection criteria described in Section 2.2. Lastly, non-parametric Spearman correlation coefficients (r) were calculated to aid in the previous analyses.

3. Results

3.1. MODIS Land Surface Temperature and PTHRES Air Temperature Mean Differences

The feature-scaled mean temperature differences (LST–2mT) in the DDR showed two distinct patterns for day and night times (Figure 2). The daytime differences across the region showed a distinct longitudinal gradient with positive temperature differences in the eastern part, reaching +5 °C in winter and +8 °C in summer, and negative temperature differences in the western part, reaching −5 °C in winter and −10 °C in summer. The nighttime differences showed a vertical temperature gradient effect, both in winter and summer, with positive temperature differences at lower elevations reaching +3 °C in winter and +4 °C in summer (except for some valleys, like the Vilariça Valley) and negative temperature differences at higher elevations reaching −3 °C in winter and −4 °C in summer.
The linear regression models fitted to the LST versus 2mT (Figure 3) and their coefficients showed low predictive power except regarding the summer nighttime (r2 = 0.33), which showed a non-negligible relationship between the data in this period. All of the models were statistically significant (p-value < 0.05) except for the model corresponding to summer daytime, which had a p-value of 0.27. Regarding the Spearman coefficients in the winter, these were positive and low during the daytime (ρ = 0.25) and moderate during the nighttime (ρ = 0.43), and both were statistically significant. In the summer, these were negative and very low during the daytime (ρ = −0.11) and positive and moderate during the nighttime (ρ = 0.63) and were both statistically significant.

3.2. MODIS Land Surface Temperature and PTHRES Air Temperature versus Elevation

The temperature differences (LST–2mT) versus elevation showed a circular distribution around zero in the winter, especially during the nighttime, and a wider distribution in the summer (Figure 4). The fitted linear regression models have very low predictive power, except for the one regarding the summer nighttime (r2 = 0.329), which showed a non-negligible relationship between the data in this period. All of the models were statistically significant, except for the model corresponding to winter nighttime, which had a p-value of 0.06. Regarding the Spearman coefficients, in the winter correlations, they were negative and low during the daytime (ρ = −0.10) and positive and very low during the nighttime (ρ = 0.03). In the summer, the coefficients were negative and low during the daytime (ρ = −0.17) and moderate during the nighttime (ρ = −0.52). All of the coefficients were statistically significant, except for winter nighttime (p-value = 0.23).
Regarding the individual relationship of LST and 2mT with elevation, the LST had a more dispersed distribution of values than the 2mT, but both were inversely proportional with elevation (Figure 5). The linear regression models certify this, with negative slopes, which also indicate the lapse rates. The lapse rates were higher for the LST, except for winter nighttime. All of the models were statistically significant.

3.3. MODIS Land Surface Temperature and PTHRES Air Temperature in the Vilariça Valley

3.3.1. Winter Season

The centered moving means of the LST and 2mT in points A and B in the Vilariça Valley (Figure 6) and their Spearman correlation coefficients were analyzed for the winter. During the daytime, the LST was higher in A than in B, with the difference between the coldest and hottest days being around +9 °C. During the nighttime, the LST was lower in A than in B, with the difference between the coldest and hottest days being around +3 °C. This switch between day and night indicates a thermal inversion. As for the 2mT, it was higher in A than in B for both day and night times, with differences between these coldest and hottest days being around +7 °C during the daytime and +4 °C during the nighttime. Contrary to the LST, there was no observable thermal inversion.
When comparing the LST and 2mT, during the daytime, the LST was higher than the 2mT in A and B, both temperatures remained fairly stable until DOY (day of the year) 11, and, afterwards, they increased. During the nighttime, the LST was lower than the 2mT in A, while in B, it was higher until DOY 11, and, after that, it was lower. The LST was fairly constant until DOY 41, and it started increasing afterwards. As for the 2mT, it started increasing earlier in DOY 358. Lastly, the Spearman coefficients indicated a strong correlation between the two temperatures for the daytime (ρ > 0.84) and a mild to strong correlation for the nighttime (ρ > 0.66) and were all statistically significant (Table 3).

3.3.2. Summer Season

The centered moving means of the LST and 2mT at points A and B in the Vilariça Valley (Figure 7) and their Spearman correlation coefficients were analyzed for the summer. During the daytime, the LST at A was consistently lower than the LST at B. During the nighttime, the LST at A was mostly higher than the LST at B. Contrary to the winter, there was no thermal inversion of the LST when comparing day and night times. As for the 2mT, it was higher at A than at B during both day and night times.
When comparing the LST and 2mT, during the daytime, the temperatures had both a slow increase until DOY 212 and a moderate decrease after that. During the nighttime, the same behavior was observed. However, during the daytime, the LST was around +13 °C higher than the 2mT, while during the nighttime, the LST was some degrees lower than the 2mT. Lastly, the Spearman coefficients indicated a mild to strong correlation between the two temperatures for the daytime (ρ > 0.67) and a strong correlation for the nighttime (ρ > 0.92) and were all statistically significant (Table 4).

4. Discussion

The MODIS LST and PTHRES 2mT were successfully extracted for the DDR in the winter and summer seasons in the years between 2002 and 2020. The chosen methodology of the analysis proved useful in showing how remote sensing data can reveal several details about the temperature, under clear-sky conditions, in an important agricultural region in Portugal when comparing it with a gauge-based temperature dataset product of interpolation. Looking at the results, four main findings can be discussed. Following their discussion, a subchapter on the limitations of the study is also presented.

4.1. Maritime Influence on Temperature

The daytime feature-scaled mean temperature differences (LST–2mT) showed an east–west gradient, where positive differences were observed in the eastern part and negative differences were observed in the western part (Figure 2). This gradient likely reflected the maritime influence on the region, which is caused by the humid air travel circulation from the Atlantic Ocean towards the DDR [36]. The humidity reduces the thermal inertia of the air and results in lower fluctuations of the temperature within a day [37]. According to [38], diurnal temperature ranges are inversely proportional to moisture in the air. Therefore, as air masses travel from the ocean into the Iberian Peninsula and lose their moisture, thermal inertia increases, and diurnal temperature ranges increase. Such effects have been reported for Egypt and in the US, where the temperature oscillated less in locations closer to the sea/ocean in comparison to locations far from it [37,39].
The observed differences, however, show that the two databases reflect the maritime influence differently, with MODIS LST being able to capture it better than PTHRES. It should be noted that because the daytime LST and 2mT data corresponded to different times of the day, there is a lag bias in the results. If the LST’s acquisition time was at the hour when the maximum temperatures are observed, which in the summer is around 4 p.m., the gradient’s transition from negative to positive should be more to the west, and the positive differences would be higher, and the negative ones would be lower. If the 2mT corresponded to the acquisition hour of the LST, the transition would be more to the east and the positive differences would be lower, and the negative ones would be higher. Nevertheless, even with a lag between the temperatures, the result is valid as it would still occur if the LST and 2mT corresponded to the same time, as previously explained. Two main reasons for the observed pattern are, firstly, the limited number of weather stations (<10) in northern Portugal used to calculate the gridded dataset [35] and, secondly, the complexity of the terrain in the DDR. It is known that terrain complexity highly influences atmospheric circulation and, thus, can affect the amount of humidity transported from the ocean to the DDR [40,41,42]. Because the PTHRES dataset was made to be a general-purpose dataset, it did not consider the complexity of the terrain and considered only the distance to the coastline to estimate the maritime influence in its interpolation process. Therefore, because of the way the data was interpolated, the positive temperature differences in the west and the negative differences in the east indicate an underestimation and overestimation of the maritime influence in the DDR, respectively, by PTHRES 2mT.
Regarding the seasonal effect, the observed range of the daytime differences during the summer was two times higher than in the winter (Figure 2). This shows that the DOY is also a determinant of the maritime influence on the temperature in the DDR. During the winter, the LST is overall closer to the estimated 2mT, while in the summer, the difference can be much lower or much higher depending on the distance to the coast. This confirms previous reports that moisture availability in the air is seasonally dependent due to the different degrees of insolation, and, therefore, the time of year can be a predictor of the degree to which a nearby ocean can affect the temperature ranges of a region such as the DDR [43].

4.2. Lapse Rate

The nighttime feature-scaled mean temperature differences (LST–2mT) showed a vertical gradient, both for the winter and summer (Figure 2). The positive differences were observed mostly at lower elevations, and negative differences were observed at higher elevations, with some exceptions in some valleys like the Vilariça Valley, where the opposite was observed. The observed general pattern shows what seems to be a partial disagreement between the two temperature datasets concerning the lapse rate despite the observed correlation of the temperature differences with elevation being low overall, except for summer nighttime (Figure 4). This natural atmospheric process determines that the air temperature decreases with elevation, resulting from the decrease in atmospheric pressure [44], but it can vary depending on the location and weather conditions [45]. The reason for the disagreement is likely related to the conjugation of these two factors. The interpolation process in the production of PTHRES 2mT is considered a uniform lapse rate regime, with no bias towards any particular weather condition or terrain complexity [35]. This is commonly the case for gridded temperature datasets [46,47,48]. Therefore, the fact that the nighttime temperature differences exhibited a decrease with elevation points towards a higher lapse rate in reality than that estimated for PTHRES 2mT. This is particularly the case for the summer nighttime, as the temperature differences are higher on both ends of the interval in comparison with the winter, and the estimated lapse rate for the LST was two times that for the 2mT (Figure 5). It should be noted that the estimated winter nighttime lapse rate for the LST is similar to that of the 2mT. This is likely related to the higher occurrence of thermal inversions in the winter when the atmospheric stability is higher, especially in mountainous regions like the DDR [49]. Also, if the LST and 2mT corresponded to the same time, the positive differences would be lower, and the negative differences would be higher, as minimum temperatures are normally reached around 6 p.m. or before sunrise. But, just as was explained in the previous subsection, the result is still valid, nevertheless.

4.3. Thermal Inversion in the Vilariça Valley

The moving average values for the LST mean values for each of the considered DOYs at points A and B in the Vilariça Valley showed a clear inversion from day to night in the winter (Figure 6). This result along with the feature-scaled mean temperature differences (LST–2mT) (Figure 2) and the estimated lapse rates (Figure 5) seems to confirm previous claims that, in this area, the typical lapse rate regime is not observed all year. The pattern of the temperature differences was more heterogeneous in the winter nighttime than in the summer nighttime, and the lapse rate for the LST was practically that of the 2mT, which is contrary to what was estimated for the summer. Thermal inversions typically occur in regions with temperate climates with high atmospheric stability conditions and are characterized by radiative imbalances along with katabatic winds and cool air drainage that result in the accumulation of cold air masses at low-elevation areas with warmer air above them [50]. Regions with complex terrain that stifle air circulation, such as the DDR, are especially prone to this atmospheric phenomenon, especially in the winter, as was observed [49]. Studies of the lower atmosphere that focused on the Colorado Plateau in the United States, the Chamonix–Mont Blanc valley in France, and Shropshire in the United Kingdom have also confirmed this [51,52,53].
As for the 2mT, as explained before, the interpolation process used to create PTHRES considered a uniform lapse rate, and it rendered it practically unable to detect thermal inversions in Portugal. Only with the inclusion of more information from weather stations and terrain complexity could it perhaps capture these phenomena in the DDR.

4.4. Land Surface Temperature and Air Temperature

The summer LST and 2mT values for the considered DOYs at points A and B showed that the LST reached around 45 °C, while the 2mT reached around 35 °C (Figure 7). Although it is a fairly obvious assessment that the LST is not always a straight representative of the 2mT, the observed difference is significant. Therefore, it is important to state that this fundamental difference between the LST and air temperature should be considered in future climate studies using remote sensing data.

4.5. Limitations of the Study

The MODIS LST data were easily accessible through the NASA Earthdata search engine. However, many of them could not be used for this study due to cloud cover or a lack of quality. This particularly affected the winter period. Furthermore, the acquisition time of the LST corresponding to the day and night times did not match perfectly with that of the maximum and minimum temperatures for PTHRES. The maximum temperatures are usually reached around 1 p.m. in the winter and 4 p.m. in the summer. As for the minimum temperatures, these are usually reached around 6 a.m. This time difference affected the display of the maritime influence and the lapse rate results by introducing a lag between the temperatures (Figure 2). Another limitation is that the uncertainty associated with the temperature values for each of the 1 km2 pixels from both MODIS LST and PTHRES 2mT was not considered. Lastly, the study showed clearly that remote sensing data can provide greater insight into the temperature heterogeneity of an agricultural region. However, its direct use in climatic studies should consider the differences in the air temperature, which can be significant (Figure 7). This difference could be estimated using empirical models [54,55,56] or statistical approaches [57,58] that relate the two temperatures to make a fairer assessment of the climate of a particular region, though these transfer functions should be site-specific.

5. Conclusions

The MODIS remote sensing land surface temperature was used to investigate temperature heterogeneities in the Douro Demarcated Region in Portugal by comparing them with the PTHRES gridded air temperature dataset for Portugal. The main findings showed a higher ability of the remote sensing land surface temperature to detect the maritime-continental contrast in temperature, a higher lapse rate, and thermal inversions in deep and sheltered valleys in the winter. This difference in PTHRES’s air temperature is most likely due to two reasons: the limited number of weather stations used to produce it and the interpolation method, which did not account for the impact of the terrain complexity. The maritime influence and lapse rate are influenced by the type of terrain, which can influence the air circulation, both horizontally and vertically. For PTHRES, these effects were taken to be linearly dependent on the distance to the coast and the elevation, respectively. Without further information from weather stations, a region like the DDR with complex terrain would be hard to describe with all the necessary detail. Thus, it can only serve as a general-purpose dataset. While the MODIS land surface temperature proved to be insightful and useful for climate studies of agricultural regions, its use should be approached with caution due to the significant deviations from the air temperature that can occur, particularly in the summer.

Author Contributions

Conceptualization, methodology, writing—review and editing: all authors. Data curation, formal analysis, investigation, software, validation, writing—original draft preparation, visualization: F.A. Resources: F.A. and A.F. Supervision: J.A.S., A.C.M., H.F. and A.F. Project administration, funding acquisition: J.A.S. and A.C.M. All authors have read and agreed to the published version of the manuscript.

Funding

This work was funded by national funds from the FCT—Portuguese Foundation for Science and Technology under project number UIDB/04033/2020.

Data Availability Statement

The remote sensing land surface temperature data from NASA “MODIS/Aqua Land Surface Temperature/Emissivity 5-Min L2 Swath 1 km V061” is available at https://search.earthdata.nasa.gov/search, accessed on 22 July 2023. The PTHRES air temperature gridded dataset is available upon request from the authors.

Acknowledgments

Project AgrifoodXXI (NORTE-45-2020-20) is co-financed by the European Regional Development Fund (FEDER) through NORTE 2020 (Programa Operacional Regional do Norte 2014/2020). H. Fraga thanks the FCT 2022. 02317.CEECIND. We also thank the CoaClimateRisk project (COA/CAC/0030/2019) “O impacto das al-terações climáticas e medidas de adaptação para as principais culturas agrícolas na região do Vale do Côa” by the Portuguese Foundation for Science and Technology (FCT).

Conflicts of Interest

The authors declare no conflict of interest and the funders had no role in the design of the study; in the collection, analyses, or interpretation of the data; in the writing of the manuscript; or in the decision to publish the results.

References

  1. Mihailescu, E.; Bruno Soares, M. The Influence of Climate on Agricultural Decisions for Three European Crops: A Systematic Review. Front. Sustain. Food Syst. 2020, 4, 64. [Google Scholar] [CrossRef]
  2. Santos, J.A.; Fraga, H.; Malheiro, A.C.; Moutinho-Pereira, J.; Dinis, L.-T.; Correia, C.; Moriondo, M.; Leolini, L.; Dibari, C.; Costafreda-Aumedes, S.; et al. A Review of the Potential Climate Change Impacts and Adaptation Options for European Viticulture. Appl. Sci. 2020, 10, 3092. [Google Scholar] [CrossRef]
  3. Ribeiro, A.C.; De Melo-Abreu, J.P.; Snyder, R.L. Apple Orchard Frost Protection with Wind Machine Operation. Agric. Meteorol. 2006, 141, 71–81. [Google Scholar] [CrossRef]
  4. Rasera, J.B.; da Silva, R.F.; Piedade, S.; Mourão Filho, F.d.A.A.; Delbem, A.C.B.; Saraiva, A.M.; Sentelhas, P.C.; Marques, P.A.A. Do Gridded Weather Datasets Provide High-Quality Data for Agroclimatic Research in Citrus Production in Brazil? AgriEngineering 2023, 5, 924–940. [Google Scholar] [CrossRef]
  5. Zhao, C.; Liu, B.; Piao, S.; Wang, X.; Lobell, D.B.; Huang, Y.; Huang, M.; Yao, Y.; Bassu, S.; Ciais, P.; et al. Temperature Increase Reduces Global Yields of Major Crops in Four Independent Estimates. Proc. Natl. Acad. Sci. USA 2017, 114, 9326–9331. [Google Scholar] [CrossRef]
  6. Hasegawa, T.; Wakatsuki, H.; Ju, H.; Vyas, S.; Nelson, G.C.; Farrell, A.; Deryng, D.; Meza, F.; Makowski, D. A Global Dataset for the Projected Impacts of Climate Change on Four Major Crops. Sci. Data 2022, 9, 58. [Google Scholar] [CrossRef]
  7. Araghi, A.; Martinez, C.J.; Olesen, J.E.; Hoogenboom, G. Assessment of Nine Gridded Temperature Data for Modeling of Wheat Production Systems. Comput. Electron. Agric. 2022, 199, 107189. [Google Scholar] [CrossRef]
  8. Freitas, T.R.; Santos, J.A.; Silva, A.P.; Fonseca, A.; Fraga, H. Evaluation of Historical and Future Thermal Conditions for Almond Trees in North-Eastern Portugal. Clim. Chang. 2023, 176, 89. [Google Scholar] [CrossRef]
  9. Blanco-Ward, D.; Monteiro, A.; Lopes, M.; Borrego, C.; Silveira, C.; Viceto, C.; Rocha, A.; Ribeiro, A.; Andrade, J.; Feliciano, M.; et al. Climate Change Impact on a Wine-producing Region Using a Dynamical Downscaling Approach: Climate Parameters, Bioclimatic Indices and Extreme Indices. Int. J. Climatol. 2019, 39, 5741–5760. [Google Scholar] [CrossRef]
  10. Hofstra, N.; New, M.; McSweeney, C. The Influence of Interpolation and Station Network Density on the Distributions and Trends of Climate Variables in Gridded Daily Data. Clim. Dyn. 2010, 35, 841–858. [Google Scholar] [CrossRef]
  11. Njoku, E.A.; Akpan, P.E.; Effiong, A.E.; Babatunde, I.O. The Effects of Station Density in Geostatistical Prediction of Air Temperatures in Sweden: A Comparison of Two Interpolation Techniques. Resour. Environ. Sustain. 2023, 11, 100092. [Google Scholar] [CrossRef]
  12. Mourtzinis, S.; Rattalino Edreira, J.I.; Conley, S.P.; Grassini, P. From Grid to Field: Assessing Quality of Gridded Weather Data for Agricultural Applications. Eur. J. Agron. 2017, 82, 163–172. [Google Scholar] [CrossRef]
  13. Serrano-Notivoli, R.; Beguería, S.; de Luis, M. STEAD: A High-Resolution Daily Gridded Temperature Dataset for Spain. Earth Syst. Sci. Data 2019, 11, 1171–1188. [Google Scholar] [CrossRef]
  14. Cornes, R.C.; van der Schrier, G.; van den Besselaar, E.J.M.; Jones, P.D. An Ensemble Version of the E-OBS Temperature and Precipitation Data Sets. J. Geophys. Res. Atmos. 2018, 123, 9391–9409. [Google Scholar] [CrossRef]
  15. Herrera, S.; Cardoso, R.M.; Soares, P.M.; Espírito-Santo, F.; Viterbo, P.; Gutiérrez, J.M. Iberia01: A New Gridded Dataset of Daily Precipitation and Temperatures over Iberia. Earth Syst. Sci. Data 2019, 11, 1947–1956. [Google Scholar] [CrossRef]
  16. Dash, P.; Göttsche, F.-M.; Olesen, F.-S.; Fischer, H. Land Surface Temperature and Emissivity Estimation from Passive Sensor Data: Theory and Practice-Current Trends. Int. J. Remote Sens. 2002, 23, 2563–2594. [Google Scholar] [CrossRef]
  17. Yan, Y.; Mao, K.; Shi, J.; Piao, S.; Shen, X.; Dozier, J.; Liu, Y.; Ren, H.; Bao, Q. Driving Forces of Land Surface Temperature Anomalous Changes in North America in 2002–2018. Sci. Rep. 2020, 10, 6931. [Google Scholar] [CrossRef]
  18. Liu, Z.; Ballantyne, A.P.; Cooper, L.A. Biophysical Feedback of Global Forest Fires on Surface Temperature. Nat. Commun. 2019, 10, 214. [Google Scholar] [CrossRef]
  19. Magarreiro, C.; Gouveia, C.; Barroso, C.; Trigo, I. Modelling of Wine Production Using Land Surface Temperature and FAPAR—The Case of the Douro Wine Region. Remote Sens. 2019, 11, 604. [Google Scholar] [CrossRef]
  20. Trigo, A.; Silva, P. Sustainable Development Directions for Wine Tourism in Douro Wine Region, Portugal. Sustainability 2022, 14, 3949. [Google Scholar] [CrossRef]
  21. Fraga, H.; Guimarães, N.; Freitas, T.R.; Malheiro, A.C.; Santos, J.A. Future Scenarios for Olive Tree and Grapevine Potential Yields in the World Heritage Côa Region, Portugal. Agronomy 2022, 12, 350. [Google Scholar] [CrossRef]
  22. UNESCO Alto Douro Wine Region. Available online: https://whc.unesco.org/en/list/1046/ (accessed on 30 May 2023).
  23. Lourenço-Gomes, L.; Pinto, L.M.C.; Rebelo, J. Wine and Cultural Heritage. The Experience of the Alto Douro Wine Region. Wine Econ. Policy 2015, 4, 78–87. [Google Scholar] [CrossRef]
  24. Gonçalves, D.A. Caracterização Agro-Ecológica Do Vale Da Vilariça; Instituto Superior Politécnico de Bragança: Bragança, Portugal, 1942. [Google Scholar]
  25. Gonçalves, F.; Carlos, C.; Crespo, L.; Zina, V.; Oliveira, A.; Salvação, J.; Pereira, J.A.; Torres, L. Soil Arthropods in the Douro Demarcated Region Vineyards: General Characteristics and Ecosystem Services Provided. Sustainability 2021, 13, 7837. [Google Scholar] [CrossRef]
  26. Fraga, H.; Santos, J.A.; Malheiro, A.C.; Moutinho-Pereira, J.; Trindade, H.; Dinis, L.-T.; Freitas, T.R.; Stolarski, O.; Rodrigues, D.; Guimarães, N.; et al. CoaClimateRisk—O Impacto Das Alterações Climáticas e Medidas de Adaptação Para as Principais Culturas Agrícolas na Região do Vale Do Côa; Zenodo, Ed.; Zenodo: Genève, Switzerland, 2023. [Google Scholar]
  27. Copernicus Land Monitoring Service Corine Land Cover. Available online: https://www.eea.europa.eu/data-and-maps/data/copernicus-land-monitoring-service-corine (accessed on 30 May 2023).
  28. Copernicus Land Monitoring Service © European Union, Copernicus Land Monitoring Service 2023, European Environment Agency (EEA). European Digital Elevation Model; Copernicus DEM, 2023. [Google Scholar]
  29. Blanco-Ward, D.; Ribeiro, A.; Barreales, D.; Castro, J.; Verdial, J.; Feliciano, M.; Viceto, C.; Rocha, A.; Carlos, C.; Silveira, C.; et al. Climate Change Potential Effects on Grapevine Bioclimatic Indices: A Case Study for the Portuguese Demarcated Douro Region (Portugal). BIO Web Conf. 2019, 12, 01013. [Google Scholar] [CrossRef]
  30. Mitchell, H.; Feldman, G.; Kuring, N. Aqua Satellite and MODIS Swath. Available online: https://svs.gsfc.nasa.gov/3348/ (accessed on 17 October 2023).
  31. Tan, J.; Che, T.; Wang, J.; Liang, J.; Zhang, Y.; Ren, Z. Reconstruction of the Daily MODIS Land Surface Temperature Product Using the Two-Step Improved Similar Pixels Method. Remote Sens. 2021, 13, 1671. [Google Scholar] [CrossRef]
  32. Wan, Z.; Hook, S.; Hulley, G. MODIS/Aqua Land Surface Temperature/Emissivity 5-Min L2 Swath 1 km V061 [Data Set]. Available online: https://lpdaac.usgs.gov/products/myd11_l2v061/ (accessed on 10 August 2023).
  33. Wan, Z.; Dozier, J. A Generalized Split-Window Algorithm for Retrieving Land-Surface Temperature from Space. IEEE Trans. Geosci. Remote Sens. 1996, 34, 892–905. [Google Scholar] [CrossRef]
  34. GDAL/OGR Contributors. GDAL/OGR Geospatial Data Abstraction Software Library. 2020. Available online: https://gdal.org/ (accessed on 19 October 2023).
  35. Fonseca, A.R.; Santos, J.A. High-Resolution Temperature Datasets in Portugal from a Geostatistical Approach: Variability and Extremes. J. Appl. Meteorol. Clim. 2018, 57, 627–644. [Google Scholar] [CrossRef]
  36. Snow, R. Continental Climate and Continentality. In Encyclopedia of World Climatology; Oliver, J.E., Ed.; Springer: Dordrecht, The Netherlands, 2005; pp. 303–305. ISBN 978-1-4020-3266-0. [Google Scholar]
  37. Scheitlin, K. The Maritime Influence on Diurnal Temperature Range in the Chesapeake Bay Area. Earth Interact. 2013, 17, 1–14. [Google Scholar] [CrossRef]
  38. Scheitlin, K.N.; Dixon, P.G. Diurnal Temperature Range Variability Due to Land Cover and Airmass Types in the Southeast. J. Appl. Meteorol. Clim. 2010, 49, 879–888. [Google Scholar] [CrossRef]
  39. Hereher, M.E.; El Kenawy, A. Extrapolation of Daily Air Temperatures of Egypt from MODIS LST Data. Geocarto Int. 2022, 37, 214–230. [Google Scholar] [CrossRef]
  40. Neteler, M. Estimating Daily Land Surface Temperatures in Mountainous Environments by Reconstructed MODIS LST Data. Remote Sens. 2010, 2, 333–351. [Google Scholar] [CrossRef]
  41. van Niekerk, A.; Sandu, I.; Vosper, S.B. The Circulation Response to Resolved Versus Parametrized Orographic Drag Over Complex Mountain Terrains. J. Adv. Model. Earth Syst. 2018, 10, 2527–2547. [Google Scholar] [CrossRef]
  42. Houze, R.A. Orographic Effects on Precipitating Clouds. Rev. Geophys. 2012, 50, RG1001. [Google Scholar] [CrossRef]
  43. Duckson, D.W. Continentality. In Climatology; Kluwer Academic Publishers: Dordrecht, The Netherlands; pp. 365–367.
  44. Stone, P.H.; Carlson, J.H. Atmospheric Lapse Rate Regimes and Their Parameterization. J. Atmos. Sci. 1979, 36, 415–423. [Google Scholar] [CrossRef]
  45. Stern, M.A.; Flint, L.E.; Flint, A.L.; Boynton, R.M.; Stewart, J.A.E.; Wright, J.W.; Thorne, J.H. Selecting the Optimal Fine-Scale Historical Climate Data for Assessing Current and Future Hydrological Conditions. J. Hydrometeorol. 2022, 23, 293–308. [Google Scholar] [CrossRef]
  46. Bonfils, C.; Santer, B.D.; Pierce, D.W.; Hidalgo, H.G.; Bala, G.; Das, T.; Barnett, T.P.; Cayan, D.R.; Doutriaux, C.; Wood, A.W.; et al. Detection and Attribution of Temperature Changes in the Mountainous Western United States. J. Clim. 2008, 21, 6404–6424. [Google Scholar] [CrossRef]
  47. Bales, R.C.; Molotch, N.P.; Painter, T.H.; Dettinger, M.D.; Rice, R.; Dozier, J. Mountain Hydrology of the Western United States. Water Resour. Res. 2006, 42, 13. [Google Scholar] [CrossRef]
  48. Hamlet, A.F.; Lettenmaier, D.P. Production of Temporally Consistent Gridded Precipitation and Temperature Fields for the Continental United States*. J. Hydrometeorol. 2005, 6, 330–336. [Google Scholar] [CrossRef]
  49. Largeron, Y.; Staquet, C. Persistent Inversion Dynamics and Wintertime PM10 Air Pollution in Alpine Valleys. Atmos. Env. 2016, 135, 92–108. [Google Scholar] [CrossRef]
  50. Joly, D.; Richard, Y. Frequency, Intensity, and Duration of Thermal Inversions in the Jura Mountains of France. Theor. Appl. Clim. 2019, 138, 639–655. [Google Scholar] [CrossRef]
  51. Sabatier, T.; Paci, A.; Canut, G.; Largeron, Y.; Dabas, A.; Donier, J.-M.; Douffet, T. Wintertime Local Wind Dynamics from Scanning Doppler Lidar and Air Quality in the Arve River Valley. Atmosphere 2018, 9, 118. [Google Scholar] [CrossRef]
  52. Price, J.D.; Vosper, S.; Brown, A.; Ross, A.; Clark, P.; Davies, F.; Horlacher, V.; Claxton, B.; McGregor, J.R.; Hoare, J.S.; et al. COLPEX: Field and Numerical Studies over a Region of Small Hills. Bull. Am. Meteorol. Soc. 2011, 92, 1636–1650. [Google Scholar] [CrossRef]
  53. Whiteman, C.D.; Bian, X.; Zhong, S. Wintertime Evolution of the Temperature Inversion in the Colorado Plateau Basin. J. Appl. Meteorol. 1999, 38, 1103–1117. [Google Scholar] [CrossRef]
  54. AL-Anbari, R.H.; Jasim, O.Z.; Mohammed, Z.T. Estimation High Resolution Air Temperature Based on Landsat8 Images and Climate Datasets. IOP Conf. Ser. Mater. Sci. Eng. 2019, 518, 022033. [Google Scholar] [CrossRef]
  55. Jang, J.-D.; Viau, A.A.; Anctil, F. Neural Network Estimation of Air Temperatures from AVHRR Data. Int. J. Remote Sens. 2004, 25, 4541–4554. [Google Scholar] [CrossRef]
  56. Cristóbal, J.; Ninyerola, M.; Pons, X. Modeling Air Temperature through a Combination of Remote Sensing and GIS Data. J. Geophys. Res. 2008, 113, D13106. [Google Scholar] [CrossRef]
  57. Liu, Y.; Ortega-Farías, S.; Tian, F.; Wang, S.; Li, S. Estimation of Surface and Near-Surface Air Temperatures in Arid Northwest China Using Landsat Satellite Images. Front. Environ. Sci. 2021, 9, 791336. [Google Scholar] [CrossRef]
  58. Benali, A.; Carvalho, A.C.; Nunes, J.P.; Carvalhais, N.; Santos, A. Estimating Air Surface Temperature in Portugal Using MODIS LST Data. Remote Sens. Environ. 2012, 124, 108–121. [Google Scholar] [CrossRef]
Remotesensing 15 05373 g001
Figure 1. The Douro Demarcated Region in Portugal. Letters A and B locate where thermal inversion was investigated in the Vilariça Valley area. Blue lines represent the existing rivers.
Figure 1. The Douro Demarcated Region in Portugal. Letters A and B locate where thermal inversion was investigated in the Vilariça Valley area. Blue lines represent the existing rivers.
Remotesensing 15 05373 g001
Remotesensing 15 05373 g002
Figure 2. Feature-scaled mean temperature differences (LST-2mT) in (a) winter daytime, (b) winter nighttime, (c) summer daytime, and (d) summer nighttime at the Douro Demarcated Region in the period between 2002 and 2020.
Figure 2. Feature-scaled mean temperature differences (LST-2mT) in (a) winter daytime, (b) winter nighttime, (c) summer daytime, and (d) summer nighttime at the Douro Demarcated Region in the period between 2002 and 2020.
Remotesensing 15 05373 g002
Remotesensing 15 05373 g003
Figure 3. LST versus 2mT scatterplots and respective linear regression models in (a) winter and (b) summer.
Figure 3. LST versus 2mT scatterplots and respective linear regression models in (a) winter and (b) summer.
Remotesensing 15 05373 g003
Remotesensing 15 05373 g004
Figure 4. Temperature differences (LST–2mT) versus elevation scatterplots and respective linear regression models in (a) winter and (b) summer.
Figure 4. Temperature differences (LST–2mT) versus elevation scatterplots and respective linear regression models in (a) winter and (b) summer.
Remotesensing 15 05373 g004
Remotesensing 15 05373 g005
Figure 5. LST and 2mT versus elevation scatterplots and respective linear regression models in (a) winter and (b) summer.
Figure 5. LST and 2mT versus elevation scatterplots and respective linear regression models in (a) winter and (b) summer.
Remotesensing 15 05373 g005
Remotesensing 15 05373 g006
Figure 6. Winter MODIS land surface temperature and PTHRES air temperature at points A and B in the Vilariça Valley during (a) day and (b) night times. Values correspond to the centered moving mean with a span of 11 days, which in turn corresponded to the mean of values available for each day in the period between 2002 and 2020.
Figure 6. Winter MODIS land surface temperature and PTHRES air temperature at points A and B in the Vilariça Valley during (a) day and (b) night times. Values correspond to the centered moving mean with a span of 11 days, which in turn corresponded to the mean of values available for each day in the period between 2002 and 2020.
Remotesensing 15 05373 g006
Remotesensing 15 05373 g007
Figure 7. Summer MODIS land surface temperature and PTHRES air temperature at points A and B in the Vilariça Valley during (a) day and (b) night times. Values correspond to the centered moving mean with a span of 11 days, which in turn corresponded to the mean of values available for each day in the period between 2002 and 2020.
Figure 7. Summer MODIS land surface temperature and PTHRES air temperature at points A and B in the Vilariça Valley during (a) day and (b) night times. Values correspond to the centered moving mean with a span of 11 days, which in turn corresponded to the mean of values available for each day in the period between 2002 and 2020.
Remotesensing 15 05373 g007
Table 1. Minimum, mean, and maximum values of the mean MODIS land surface temperature in the Douro Demarcated Region.
Table 1. Minimum, mean, and maximum values of the mean MODIS land surface temperature in the Douro Demarcated Region.
MODIS Land Surface Temperature (°C)
DaytimeNighttime
MinimumMeanMaximumMinimumMeanMaximum
Winter6.9414.1518.41−2.000.923.66
Summer29.7843.5550.4813.0716.7220.23
Table 2. Minimum, mean, and maximum values of the mean PTHRES air temperature in the Douro Demarcated Region.
Table 2. Minimum, mean, and maximum values of the mean PTHRES air temperature in the Douro Demarcated Region.
PTHRES Air Temperature (°C)
DaytimeNighttime
MinimumMeanMaximumMinimumMeanMaximum
Winter10.4713.1215.42−0.681.302.59
Summer29.3133.0635.7012.3716.9618.95
Table 3. Spearman correlation coefficients for the winter MODIS land surface temperature and PTHRES air temperature at points A and B in the Vilariça Valley. p-values for each of the coefficients were below 0.05.
Table 3. Spearman correlation coefficients for the winter MODIS land surface temperature and PTHRES air temperature at points A and B in the Vilariça Valley. p-values for each of the coefficients were below 0.05.
Daytime
LST ALST B2mT A2mT B
LST A1.000.990.990.99
LST B 1.000.990.99
2mT A 1.001.00
2mT B 1.00
Nighttime
LST ALST B2mT A2mT B
LST A1.000.660.820.84
LST B 1.000.790.78
2mT A 1.000.99
2mT B 1.00
Table 4. Spearman correlation coefficients for the summer MODIS land surface temperature and PTHRES air temperature at points A and B in the Vilariça Valley. p-values for each of the coefficients were below 0.05.
Table 4. Spearman correlation coefficients for the summer MODIS land surface temperature and PTHRES air temperature at points A and B in the Vilariça Valley. p-values for each of the coefficients were below 0.05.
Daytime
LST ALST B2mT A2mT B
LST A1.000.960.730.74
LST B 1.000.670.69
2mT A 1.000.99
2mT B 1.00
Nighttime
LST ALST B2mT A2mT B
LST A1.000.940.900.93
LST B 1.000.960.97
2mT A 1.000.99
2mT B 1.00
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Adão, F.; Fraga, H.; Fonseca, A.; Malheiro, A.C.; Santos, J.A. The Relationship between Land Surface Temperature and Air Temperature in the Douro Demarcated Region, Portugal. Remote Sens. 2023, 15, 5373. https://doi.org/10.3390/rs15225373

AMA Style

Adão F, Fraga H, Fonseca A, Malheiro AC, Santos JA. The Relationship between Land Surface Temperature and Air Temperature in the Douro Demarcated Region, Portugal. Remote Sensing. 2023; 15(22):5373. https://doi.org/10.3390/rs15225373

Chicago/Turabian Style

Adão, Filipe, Helder Fraga, André Fonseca, Aureliano C. Malheiro, and João A. Santos. 2023. "The Relationship between Land Surface Temperature and Air Temperature in the Douro Demarcated Region, Portugal" Remote Sensing 15, no. 22: 5373. https://doi.org/10.3390/rs15225373

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop