Use of MODIS and OLI/TIRS to estimate TVDI and Surface Moisture in Agricultural Monitoring Programs

MODIS e OLI/TIRS Para Estimar TVDI e a Umidade da Superfície em Programas de Monitoramento Agrícola

Abstract

One of the major challenges for effective agricultural activity monitoring systems is defining robust indicators of spatial and temporal variability for the main risk factors associated with crop production. In this context, this study aimed to analyze the potential of the Temperature-Vegetation Dryness Index (TVDI), obtained by terrestrial and orbital sensors from soybean production areas in southern Brazil, in generating spatial and temporal patterns of the main risk factor, surface moisture, to be incorporated in operational agricultural monitoring systems. For this purpose, MODIS Terra and Landsat-8 OLI/TIRS sensor images were used, as well as data from surface positioned sensors to serve as a reference. The study area encompassed one soybean crop area, soybean mapped crop areas near the experimental area, and the municipality of Carazinho-RS. The experimental area was analyzed during the soybean growing season. As the TVDI data estimated by OLI/TIRS and MODIS sensors were coherent and robust, both sensors can be used in conjunction for agricultural risk monitoring. Its main features are continuous monitoring of large production regions by TVDI_MODIS and spatial distribution detailing by TVDI_OLI/TIRS in critical periods to water deficit.

Keywords
agriculture; water deficit; remote sensing; surface temperature; NDVI

Resumo

Um dos importantes desafios para construção de sistemas eficazes de monitoramento agrícola é a definição de indicadores que representem a variabilidade espacial e temporal dos principais fatores de risco associados à produção. Neste contexto, se analisou a potencialidade de uso do TVDI (Temperature-Vegetation Dryness Index), obtido a partir dos sensores terrestres e orbitais em áreas de produção de soja no sul do Brasil, na geração dos padrões espaciais e temporais do principal fator de risco, a umidade da superfície, visando sua incorporação em sistemas operacionais de monitoramento agrícola. Utilizou-se imagens dos sensores MODIS Terra e OLI/TIRS Landsat 8, e dados de sensores posicionados na superfície que serviram como referência. A área de estudo abrangeu lavoura de soja, onde foi conduzido o experimento, lavouras de soja próximas e o município de Carazinho/RS. O período de análise foi o período de cultivo da soja. Dada a coerência e robustez do TVDI estimado pelos sensores orbitais, verificou-se que é possível o uso conjugado destes sensores em sistema de monitoramento de risco agrícola. A principal característica é o acompanhamento contínuo da umidade em amplas regiões agrícolas usando TVDI_MODIS e detalhamento da distribuição espacial da umidade da superfície com TVDI_OLI/TIRS nos períodos críticos à deficiência hídrica.

Palavras-chave
agricultura; déficit hídrico; sensoriamento remoto; temperatura de superfície; NDVI

1. Introduction

One of the major challenges for effective agricultural activity monitoring systems is defining robust indicators of spatial and temporal variability for the main risk factors associated with crop production (Fraisse et al., 2016; Radin & Matzenauer, 2016). In the case of soybeans grown in southern Brazil, the most important risk factor is water deficit (Sentelhas et al., 2015, Zanon et al., 2016, Matzenauer et al., 2020).

Surface moisture is therefore a key variable in this scenario and can be either measured by direct methods or estimated by indirect methods. Indirect estimation of surface moisture can be effective and convenient, as it reduces labor force, time, and expenses, especially for large agricultural areas (Uniyal et al., 2017). One of the ways to estimate it indirectly and that has been gaining increased space in the agricultural context is through remote sensing techniques. Satellite images are considered one of the best technologies for systematic data collection in monitoring agricultural activities, from which data can be obtained using portable sensors, drones, or orbital sensors (Tsukahara et al., 2016). The use of data from different sensors in a complementary way helps in monitoring; however, there is still a lack of specific studies for different proposed indicators.

Sandholt et al. (2002) proposed to estimate surface moisture using an empirical approach that relates, in a two-dimensional space, the Normalized Difference Vegetation Index (NDVI) and Surface Temperature (Ts), resulting in the TVDI (Temperature-Vegetation Dryness Index). Studies have shown that TVDI is a robust indicator of surface moisture in both dry (Sandholt et al., 2002) and wet (Holzman et al., 2014, Uniyal et al., 2017, Schirmbeck et al., 2018) climates. This is because this index uses information on vegetation status and temperature changes, both determined by the moisture factor (Sandholt et al., 2002, Chen et al., 2015, Wang et al., 2020). Schirmbeck et al. (2022a,b) demonstrated that specifically for soybean production regions in the Rio Grande do Sul State, TDVI is a robust index and has a significant correlation with several variables (storage, deficit, ETr, ETr/ET0, and moisture) associated with the water conditions in the soil-water-plant system.

TVDI can be calculated using data from different platforms if they have sensors capable of providing NDVI and Ts data, generating information at different spatial and time scales. The most common are studies with Landsat TM- and OLI-derived TVDI (Gao et al., 2011, Chen et al., 2015, Li et al., 2016, Sayago et al., 2017, Wang et al., 2020) and MODIS-derived TVDI, from Terra (Chen et al., 2011, Son et al., 2012, Garcia et al., 2014, Sun et al., 2017). Despite this, the use of TVDI in operational systems for agricultural monitoring still lacks research characterizing similarities and differences between indexes arising from different sensors and their feasibility when coming from different platforms in a joint and complementary way, as well as establishing protocols for use.

Thus, this study aimed to analyze how the spatial and temporal patterns of TVDI, obtained in soybean production areas in southern Brazil, from terrestrial and orbital sensors (OLI/TIRS and MODIS), can be used efficiently in operational agricultural monitoring systems.

2. Material and Methods

The study was carried out at three scales: a soybean crop where the experiment was conducted, soybean crops mapped in areas surrounding the experiment, and the entire municipality of Carazinho-RS, located in the Pampa Biome, in southern Brazil (Fig. 1).

Figure 1
Location of the study areas in northwestern mesoregion of the Rio Grande do Sul State and municipality of Carazinho/RS. Landsat-8 image from November 20, 2017, orbit/point (222/80).

Carazinho is in the northwestern mesoregion of Rio Grande do Sul State, which is responsible for more than 60% of the soybean production in the state (IBGE/SIDRA; May 12, 2021). Of the total 46,200 ha of cultivated area in the municipality, 40,400 ha are soybeans, mostly rainfed. According to Alvares et al. (2013), the local climate is Cfa type, which stands for humid subtropical, with hot summers and rainfall with regular distribution throughout the year, but with high interannual variability.

The experiment was carried out in an on-farm format, in a partnership among Embrapa Trigo from Passo Fundo, Granja Capão Grande, and the Faculty of Agronomy at UFRGS. The total area of the farm is 553.7 ha under intensive soybean production, with the experiment covering 27.4 ha. The experiment was analyzed from November 2017 to April 2018, with soybeans being sown on November 13, 2017, and harvested on April 3, 2018. Grain yield achieved 4,629 kg ha^-1. More details on farming practices are described in Schirmbeck et al. (2019).

2.1. Obtaining TVDI

As proposed by Sandholt et al. (2002), input data for calculating TVDI were T_S and NDVI (Eq. (1)), which were obtained from different sensors. T_S is the radiative temperature of the pixel (K); T_Smin is the minimum surface temperature (K) corresponding to the wet limit in the evaporative triangle dispersion; NDVI is the Normalized Difference Vegetation Index; “a” and “b” are the linear and angular coefficients of the line representing the dry limit obtained from the NDVI and T_S scatter plot, and are used for TVDI model standardization.

(1) $$TVDI=\frac{\left(T_{S\mathrm{}}-T_{Smin\mathrm{}}\right)}{(a+b\mathrm{ }NDVI\mathrm{}-T_{Smin\mathrm{}})}$$

The evaporative triangle from two-dimensional space dispersion between Ts and NDVI defines different soil cover types and their respective moisture content. For points closer to the dry limit, represented by the slope of the T_S/NDVI line, TVDI equals 1 and indicates a water deficit. When TVDI equals zero, points are within the wet limit, which is given by the average minimum surface temperature over the analyzed period (Sandholt et al., 2002).

Wet and dry limit delimitation and triangle-shaped dispersion are called index parameterization. This study covered the entire soybean cycle, as proposed by Schirmbeck et al. (2018). As TVDI is an index that provides moisture estimation relative to wet and dry limits, a parameterization per season, using all images available during the period of interest, is required to assess TVDI variations at different dates throughout the season. Aiming for integrated and complementary use as surface moisture indicators in crop monitoring systems, TVDI was determined at different spatial and time scales for data acquisition.

2.2. Surface-mounted sensors (reference)

A radiometric station was installed in the on-farm experiment to measure NDVI and T_S data on the surface. The station has technology like that of sensors aboard orbital platforms and thus served as a reference for analyzing orbital platform TVDI data. NDVI was measured by SRS - Meter Group sensors, while surface temperature (T_S) was measured by a model SI 421-Apogee sensor. The data were stored in a datalogger EM50 - Meter Group, with records every 15 minutes.

NDVI sensors were used to measure incident and reflected radiation flux within the red (0.6 to 0.7 µm) and near-infrared (0.805 to 0.815 µm) spectrum. A sky-facing hemispheric sensor was installed 1 m above the soybean canopy to measure incident electromagnetic radiation. The surface-facing directional sensor was used to measure reflected surface radiation (soybean/soil) and was restricted to a 20° field of view. T_S sensor measured surface-emitted radiation within the thermal infrared spectrum (8 to 14 µm) at a half-angle field of view of 18°. T_S and directional NDVI sensors were installed in pairs, with a 90° angle pointing to the same area.

2.3. Satellite images used

TVDI was determined from images using data from two orbital sensors with distinct characteristics, OLI/TIRS from Landsat-8 and MODIS from Terra.

For OLI/TIRS sensor, TVDI parameterization made use of 4 images referring to 11/20/2017, and 01/7, 02/8, and 02/24/2018, orbit/point 222/80. To do so, we used daily NDVI product from USGS - ESPA database (United States Geological Survey - Center Science Processing Architecture), with 30-m resolution within the red (band 4) and near-infrared (band 5) bands, and thermal bands 10 and 11, with 100-m resolution. T_S was estimated using the split-window algorithm proposed by Jiménez-Muñoz et al. (2014), which were previously tested by Schirmbeck et al. (2017, 2019). To calculate TVDI, an evaporative triangle was obtained from the 739k pixels in the municipality of Carazinho. Quality data were used to remove clouds and shadows from the OLI/TIRS images, as well as apply a mask to the urban area.

For MODIS sensor images, TVDI parameterization comprised images from 11/01/2017 to 04/23/2018 (12 images) from the LP DAAC database (Land Processes database Distributed Active Archive Center). The images were from the h13v11 and h13v12 quadrants and encompassed the entire Rio Grande do Sul State, which is required to cover different land covers and surface moisture levels for images with 1-km spatial resolution. NDVI images were obtained from the MOD13A2 product as 16-day maximum value images. For T_S, MOD11A2 product was used, with an image composition corresponding to the first 8 days obtained in the day and night, with no clouds, using the split-window method (Schirmbeck et al., 2018, 2019). For TVDI calculation, an evaporative triangle was obtained using 309k pixels covering the entire area of Rio Grande do Sul State, due to the smaller spatial detail of the sensor.

2.4. TVDI data analysis

The analyses were organized in three stages to cover the main aspects of TVDI use in operational monitoring systems. In the first stage, evaporative triangle adjustment parameters were obtained using OLI/TIRS and MODIS sensors and compared, discussing procedures for their adjustment from three sensor types.

The second step evaluated the coherence of TVDI data from the different sensors. TVDI, NDVI, and T_S were estimated for the data collection point from the sensors: the plot where the experiment is being implemented, the surrounding soybean areas, and the entire municipality of Carazinho. Statistics were obtained in a sampling window of 3 x 3 pixels, for both satellites, covering 90 x 90 m for OLI/TIRS and 3000 x 3000 m for MODIS. These values were analyzed together with the data obtained on the surface at the radiometric station. Means and variability were calculated, and frequency distribution histograms were also made.

In the last stage, a proposal for TVDI use in an operational monitoring system was explored, integrating OLI/TIRS and MODIS data. For water condition characterization at a regional scale (to entire RS State), images and a plot of TVDI mean values throughout the soybean growing season, both from the MODIS sensor, were presented. Spatial detailing (municipality and producing crops) was carried out with the joint use of MODIS and OLI/TIRS images, which were temporally defined according to the critical period to water restriction for soybeans when there is the greatest risk of losses.

3. Results and Discussion

3.1. Evaporative triangle adjustment

OLI/TIRS and MODIS inherent characteristics show that evaporative triangles for parameterization during soybean growing season have a similar pattern. Therefore, both sensors were coherent but with differences in parameters to define dry and wet limits. If compared to OLI/TIRS, the MODIS triangle (Fig. 2a) had a lower T_Smin (293.6 K) and higher negative slope of the dry boundary line (-28.8), with the latter associated with surface evapotranspiration (Chen et al., 2015; Silva-Fuzzo & Rocha, 2016). OLI/TIRS triangle had a higher T_Smin (296.7 K) and lower negative straight slope (-20.5). Dry and wet limits were different because they include water conditions at different dates and spatial and time resolutions between both orbital sensors evaluated. For MODIS, a larger number of images (November and April) and spatial coverage (RS State) were used in parameterization. As a result, MODIS had a greater negative slope, therefore, surface moisture conditions are more contrasting (Schirmbeck et al., 2019).

Figure 2
Evaporative triangle obtained from: a) MODIS sensor - Earth for the Rio Grande do Sul State and b) Landsat-8 OLI/TIRS sensor for the municipality of Carazinho, 2017/18 soybean growing season.

When comparing surface moisture with the development cycles of annual crops, the evaporative triangle must be adjusted to cover the entire crop season (Schirmbeck et al., 2018). Thus, when building a triangle, extreme water conditions for the entire period under analysis will be represented, making dry and wet limits fixed. This is critical for comparisons over time since TVDI is related to these limits. If adjustment is made on each date separately, the same TVDI value may represent different water conditions on different dates.

3.2. Coherence of TVDI from different sensors

Figure 3 shows that TVDI had a similar time variation pattern between sensors, with coherence in the magnitude of values on dates in which the acquisition coincided, despite known differences in sampled area size for each set of collected data. However, Fig. 3 also demonstrates some important differences between sensors, changing the usefulness of the index if incorporated in agricultural monitoring systems.

Figure 3
Time profile of the surface moisture index (TVDI) calculated using sensors installed on the surface (TVDI_sup) for the pixels composing a 3 x 3 window obtained by the MODIS sensor (TVDI_MODIS), and by the OLI/TIRS sensor (TVDI_OLI/TIRS), and daily rainfall data throughout the 2017/18 soybean growing season. Carazinho-RS.

A TVDI dataset from a radiometric station (TVDI_sup) installed in a crop is analyzed to verify daily variability. This is because NDVI values are associated with green biomass (Monteiro et al., 2012; Prabhakara et al., 2015) and are quite stable from one day to another. High TVDI variability over days is due to its high sensitivity to surface temperature change (Schirmbeck et al., 2017), which, in turn, may be associated with rainfall events. TVDI_sup was determined using data from a sensor installation point in the field; however, in this study, it only served as a reference for orbital data analysis. This data has a positive point of high frequency over time, but with spatial representation restricted to the sensor view area (e.g., at 1 m high, it has 64 cm in diameter). In a precision agriculture context, these sensors, or similar ones, could be onboard terrestrial platforms, or even on drones, and generate TVDI maps with a detailed spatial distribution. TVDI obtained from OLI/TIRS sensor (TVDI_OLI/TIRS) also showed superior values to those from MODIS (TVDI_MODIS). Such difference can be attributed mainly to the sampled area. While OLI/TIRS sensor has a 90 x 90 m window, the MODIS sensor shows a wider context (3000 x 3000 m window). High values occurred at the beginning of the soybean growing season and, as expected (Silva-Fuzzo & Rocha, 2016; Uniyal et al., 2017; Holzman et al., 2018), they are associated with lower surface moisture. In this period, TVDI_OLI/TIRS had higher values (0.77 on 11/20) than did TVDI_MODIS (0.65 on 11/17). The following images, 01/07 and 01/19 for OLI/TIRS and MODIS respectively, showed a TVDI reduction as surface moisture increased by occasion of rainfall events (12/23 and 01/01, respectively), totaling 60 mm. In the following images 02/02 (MODIS) and 02/08 (OLI/TIRS), the index increases again due to a sequence of dry days before image capture. There was another index drop in the next pair of images [02/18 (MODIS) and 02/24 (OLI/TIRS)]. This can be associated with 4 significant rain events in the 8 days before image capture. Thus, the pattern obtained with data from both orbital sensors was coherent with surface moisture changes throughout the cycle. Therefore, TVDI datasets from both sensors can be used (Xu et al., 2018).

By using TVDI_OLI/TIRS, variability within the experimental crop could be seen with a detail level like the spatial resolution of a temperature sensor (i.e., 90 m). It is useful for detailing large crops or even large production regions, but with a limited number of images due to the revisit and cloudiness, which in the evaluated soybean growing season was 4 images throughout the cycle. Yet, for TVDI_MODIS, as it uses a 16-day composition (different from instantaneous data), it has as positive a continuous representation throughout the season, identifying temporal patterns throughout the season; in the soybean cycle under analysis, there were 12 images. In this case, the scale for obtaining information is 1000 m, which is suitable for monitoring large areas, and at a regional scale.

By expanding the study scale to the entire municipality of Carazinho, we confirmed coherence between TVDI values from OLI/TIRS and MODIS orbital sensors throughout the soybean growing season (Fig. 4). Just as seen in the experimental area (Fig. 3), TVDI_OLI/TIRS values were slightly above those from TVDI_MODIS. Even so, the trend of the TVDI_MODIS is also observed in the TVDI_OLI/TIRS data. It is important to note that TVDI_OLI/TIRS describes moisture conditions on a specific day, while TVDI_MODIS reflects a 16-day condition from the NDVI product; thus, it carries somewhat distinct information, which may explain part of the magnitude differences in the data. Also, the mean TVDI on each date, from both sensors, for selected crops around the experimental area and crops throughout the municipality of Carazinho were similar. This is because soybean cultivation occupies about 60% of the municipality area (IBGE, SIDRA 2021), and the fact that farmers in the region may have adopted similar crop management practices. This ends up impacting NDVI and Ts in a similar way, as well as TVDI both in crop areas and in the entire municipality. These characteristics facilitate monitoring and integrating information from different sensors in this region.

Figure 4
Time profile of the surface moisture index (TVDI) from OLI/TIRS (TVDI_OLI/TIRS) and MODIS (TVDI_MODIS) sensors for selected crops (SCrop) and the entire municipality (Mun) of Carazinho.

Variability in mean values (Fig. 5 and Table 1) could be assessed by analyzing the dataset from all orbital-image pixels. The region showed deficit at the beginning of the crop cycle until mid-February. Afterwards, water condition was adequate, with a reduction in TVDI in virtually the entire region. In the pairs of images Nov 17 and Nov 20, as well as Feb 02 and Feb 08, higher mean TVDI was observed, greater variability with the occurrence of values close to 1, which indicates that surface moisture condition can be worrying in the period. On the other hand, in the pair of images of Feb 18 and Feb 24, lower mean values and variability were observed, farther from 1, indicating a satisfactory moisture situation (Sandholt et al., 2002, Holzman et al., 2018).

Among the analyzed histograms, November images were the ones with the greatest differences between both sensors (histograms, Fig. 5 and mean values, Table 1). This can be attributed to a greater spectral mixture between soil and vegetation at the beginning of the crop cycle, which is detected distinctly by MODIS and OLI/TIRS sensors. Furthermore, the histograms of MODIS images showed peaks and depressions in frequencies, which is attributed to the spatial resolution of these images.

Figure 5
Histograms of the frequency of TVDI occurrence for soybean crops in the municipality of Carazinho, obtained by the OLI/TIRS sensor (TVDI_OLI/TIRS) in orange, and by the MODIS sensor (TVDI_MODIS) in gray, between November/2017 and February/2018.

Table 1
TVDI mean values and standard deviation for the municipality of Carazinho, obtained by the OLI/TIRS sensor (TVDI_OLI/TIRS) of the Landsat-8 satellite and by the MODIS sensor (TVDI_MODIS) of the Terra satellite.

TVDI	Nov 20	Nov 17	Feb 08	Feb 02	Feb 24	Feb 18
	TVDI OLI/TIRS	TVDI MODIS	TVDI OLI/TIRS	TVDI MODIS	TVDI OLI/TIRS	TVDI MODIS
Mean	0.732	0.655	0.660	0.644	0.414	0.458
Standard deviation	0.194	0.093	0.172	0.072	0.156	0.072

3.3. Proposal for TVDI use in monitoring systems

Coherence between the results of both sensors, whether in restricted areas (cropland) or wider ones, supports the combined use of MODIS and OLI/TIRS sensors for agricultural monitoring purposes. After defining the index temporal profile throughout the growing season by TVDI_MODIS, regional water conditions can be characterized for a given period. This result is important since robust indices that allow a clear view of the occurrence and variability in some conditions are essential in monitoring systems (Fraisse et al., 2016; Ravelo et al., 2016). In subsequent analysis, TVDI_OLI/TIRS index can be used to detail spatial distribution in risk areas, improving understanding of impacts that variability in moisture conditions can cause when it occurs in critical phenological phases of crops, thus helping in decision-making and minimizing risks (Fraisse et al., 2016).

In this conception, Fig. 6 shows TVDI use as a surface water indicator in a monitoring system, especially on detailing and mapping of risk areas. These are fundamental especially during the critical period of the crop to water restriction. This information is useful for decision making, increasing resource use efficiency, and ensuring agricultural production stability (Fraisse et al., 2016; Romani et al., 2016).

Figure 6
Operating scheme of a monitoring system: TVDI_MODIS images for the RS State, the phenological scale of the crop and TVDI_MODIS profile for the municipality of Carazinho-Mun. Image detail TVDI_MODIS-Mun for Feb 02 and TVDI_OLI/TIRS-Mun for Feb 08.

A first monitoring component could be generating 16-day images for the entire RS State using large-scale TVDI_MODIS products continuously throughout the whole soybean cycle. From these images, the average TVDI values for different cities could be extracted, generating a profile with the time course of the index. For the crop under analysis, the index varied a lot over time and, on some dates, between regions within the RS State. TVDI_MODIS images with a greater presence of red areas occurred at the beginning of the crop cycle, especially on December 3rd. On this data, after extracting data from the municipality of Carazinho, the highest TVDI in the profile was observed. Afterwards (images from Dec 19 to Feb 2) yellow and green tones predominated, consistent with high profile values. At the end of the crop cycle, from Feb 18 to Apr 23, colors in shades of blue were prevalent and coincided with the low values of TVDI_MODIS in the profile.

A second component can be inputting phenological data to be able to infer the effects of moisture restriction on grain yield (Romani et al., 2016). When water deficit coincides with a critical soybean period, the largest losses will be observed since water shortage directly impacts the formation of yield components (Matzenauer et al., 2020). In this study, phenology was observed during the experiment, and the crop was at flowering and grain filling stages in January and February. For this component, the data collected by EMATER-RS, GeoR Package could also be used and made available in an Economic Newsletter, in which crop evolution is monitored every 8 days in different regions within RS so that the critical period could be identified. By opting for orbital data only, NDVI can be used to estimate phenology (Fontana et al., 2015) and define critical periods.

A third component is spatial detailing during the critical period, using TVDI_OLI/TIRS product. In the crop under review, the critical date was February 2, when high TVDI values, R5.2 stage (critical), and OLI/TIRS image availability occurred simultaneously. On this date, by expanding the municipality of Carazinho, we could understand the importance of using the sensors together. In TVDI_MODIS, spatial resolution enabled us to detect a homogenization of values (predominance of green tones and some yellow dots). Yet in TVDI_OLI/TIRS, TVDI spatial distribution could be seen with a greater distinction between plots, thus identifying areas with greater water deficits (in red).

The original proposal for a monitoring system still lacks studies and analyses to define TVDI thresholds able to identify critical water conditions and moments for soybeans and thus define a methodology for locating and classifying risk areas. Another aspect to be considered is that the evaporative triangle in these operating systems could be adjusted using different growing season, with extremely good or bad water conditions. Thus, an evaporative triangle characteristic of the production region could be obtained, which could be suitable for any growing season and in real-time. Such a feature is fundamental in operational agricultural monitoring since, as highlighted by Massignam et al. (2016) and Sivakumar (2016), when specific and appropriate to the characterization of limiting factors for a given crop, the products of these systems generate information both for agrometeorological planning and real-time monitoring, which is the case of TVDI for soybeans in the RS State.

In addition to using TVDI for monitoring, its continuous series of several years could also allow analyzing risk related to moisture factor statistically. Although other indexes have been used for this purpose, for example, the Water Need Satisfaction Index (ISNA in Portuguese) (Cunha et al., 2001), TVDI has the advantage of generating risk data with high spatial detail, which is impracticable using data from weather stations.

4. Conclusions

Evaporative triangles adjusted using OLI/TIRS and MODIS sensor data show similarities, but with differences in magnitude for parameters defining the dry and wet limits, which are associated with radiometric, spectral, spatial, and temporal resolution.

TVDI index data obtained from surface and orbital sensors were coherent and complementary between them for spatial and time information.

As TVDI proved to be robust and there was coherence between data from the different sensors tested, the combined use of these sensors enables a proposal for an agricultural monitoring system. The main feature of the system is continuous monitoring of moisture in large production regions, using TVDI_MODIS and detailing of surface moisture spatial distribution using TVDI_OLI/TIRS for soybean phenological phases critical to water deficit.

Acknowledgments

This study was carried out with the support of the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brazil (CAPES) - Financing Code n° 001 and by the Embrapa Trigo, through the SEG project. 02.15.07.003.00.00. The team also thanks to the employees of the Embrapa Trigo, Elisson S.S. Pauletti and Cristian M. Plentz, who helped in conducting the experiment and data collection, and the Granja Capão Alto for the partnership in the work.

References

• ALVARES, C.A.; STAPE, J.L.; SENTELHAS, P.C.; GONçALVES, J.L.M.; SPAROVEK, G. Köppen's climate classification map for Brazil. Meteorologische Zeitschrift, v. 22, n. 6, p. 711-728. 2013.
• CHEN, J.; WANG, C.; JIANG, H.; MAO, L.; YU, Z. Estimating soil moisture using Temperature-Vegetation Dryness Index (TVDI) in the Huang-huai-hai (HHH) plain, International Journal of Remote Sensing, v. 32, n. 4, 1p. 165-1177. 2011. doi
  » https://doi.org/10.1080/01431160903527421
• CHEN, S., WEN, Z., JIANG, H., ZHAO, Q., ZHANG, X., CHEN, Y. Temperature Vegetation Dryness Index Estimation of Soil Moisture under Different Tree Species. Sustainability, v. 7, p. 11401-11417. 2015. doi
  » https://doi.org/10.3390/su70911401
• CUNHA, G.R.; BARNI, N.A.; HAAS, J.C.; MALUF, J.R.T.; MATZENAUER, R.; et al. Zoneamento agrícola e época de semeadura para soja no Rio Grande do Sul. Revista Brasileira de Agrometeorologia, v. 9, n. 3, p. 446-459. 2001.
• EMATER-RS. Ascar Informativo Conjuntural. Porto Alegre, Março de 2021. Available in http://www.emater.tche.br/site/info-agro/informativo_conjuntural.php#.YFXvLK9Kg2w
  » http://www.emater.tche.br/site/info-agro/informativo_conjuntural.php#.YFXvLK9Kg2w
• FRAISSE, C.; ANDREIS, J.; BORBA, T.; CERBARO, V.; GELCER, E.; et al. AgroClimate - Tools for managing climate risk agriculture. Agrometeoros, v. 24, n. 1, p. 121-129, 2016.
• FONTANA, D.C.; PINTO, D.G.; JUNGES, A.H., BREMM, C. Inferências sobre o calendário agrícola a partir de perfis temporais de NDVI/MODIS. Bragantia, v. 74, n. 3, p. 350-358. 2015. doi
  » https://doi.org/10.1590/1678-4499.0439
• GAO, Z.; GAO, W.; CHANG, N. Integrating temperature vegetation dryness index (TVDI) and regional water stress index (RWSI) for drought assessment with the aid of LANDSAT TM/ETM+ images. International Journal of Applied Earth Observation and Geoinformation, v. 13, p. 495-503. 2011.
• GARCIA, M.; FERNáNDEZ N.; VILLAGARCíA, L.; DOMINGO, F.; PUIGDEFáBREGAS, J., SANDHOLT, I. Accuracy of the Temperature-Vegetation Dryness Index using MODIS under water-limited vs. energy-limited evapotranspiration conditions. Remote Sensing of Environment, v. 149, p. 100-117, 2014.
• HOLZMAN M.E.; RIVAS, R.E.; PICCOLO, M.C. Estimating soil moisture and the relationship with crop yield using surface temperature and vegetation index. International Journal of Applied Earth Observation and Geoinformation, v. 28, p. 181-192. 2014.
• HOLZMAN M.E.; CARMONA F.; RIVAS R.; NICLòS, R. Early assessment of crop yield from remotely sensed water stress and solar radiation data. ISPRS Journal of Photogrammetry and Remote Sensing, v. 145, p. 297-308. 2018.
• IBGE/SIDRA. Produção Agrícola Disponível em https://sidra.ibge.gov.br/Tabela/1612, acesso em 12/05/2020.
  » https://sidra.ibge.gov.br/Tabela/1612
• JIMéNEZ-MUñOZ, J.C.; SOBRINO, J.A.; SKOKOVIC, D.; MATTAR, C.; CRISTóBAL, J. Land surface temperature retrieval methods from Landsat-8 thermal infrared sensor data. IEEE Geoscience and Remote Sensing Letters, v. 11, p. 1840-1843. 2014.
• LI, B.; CHAOPU, T.; YONGQIANG, Z.; XIAOYUAN, Y. Estimating soil moisture with Landsat Data and its application in extracting the spatial distribution of winter flooded paddies. Remote Sensing, v. 8, n. 38. 2016. doi
  » https://doi.org/10.3390/rs8010038
• MASSIGNAM, A.M.; PANDOLFO, C.; RICCE, W.S.; VIEIRA, H.J.; BRAGA, H.J. A agrometeorologia operacional em Santa Catarina. Agrometeoros, v. 24, n. 1, p. 55-63. 2016.
• MATZENAUER, R.; MALUF, J.R.T., RADIN, B. Regime de Chuvas e Produção de Grãos no Rio Grande do Sul: Impacto das Estiagens e Relação com o Fenômeno El Niño Oscilação Sul Porto Alegre: Emater/RS-Ascar, 133 p., 2020.
• MONTEIRO, P.F.C.; FILHO, R.A.; XAVIER, A.C., MONTEIRO, R.O.C. Assessing biophysical variable parameters of bean crop with hyperspectral. Scientia Agricola, v. 69, n. 2, p. 87-94. 2012. doi
  » https://doi.org/10.1590/S0103-90162012000200001
• PRABHAKARA, K.; HIVELY, W.D.; MCCARTY, G.W. Evaluating the relationship between biomass, percent groundcover and remote sensing indices across six winter cover crop fields in Maryland, United States. International Journal of Applied Earth Observation and Geoinformation, v. 39, p. 88-102. 2015. doi
  » https://doi.org/10.1016/j.jag.2015.03.002
• RADIN, B.; MATZENAUER, R. Uso das informações meteorológicas na agricultura do Rio Grande do Sul. Agrometeoros, v. 24, n. 1, p. 41-54. 2016.
• RAVELO, A.C.; PLANCHUELO, A.M.; ZANVETTOR, R.E.; BOLETTA, P.E.C. Drought monitoring and assessment system for Argentina. Agrometeoros, v. 24, n.1, p.113-120, 2016.
• ROMANI, L.A.S.; BANBINI, M.D.; COLTRI, P.P.; LUCHIARI JR., A.; KOENIGKAN, L.V.; et al. Sistema de Monitoramento Agrometeorológico - Agritempo: inovação em rede apoiando políticas públicas e a tomada de decisão agrícola. Agrometeoros, v. 24, n. 1, p. 29-40. 2016.
• SANDHOLT, I.; RASMUSSEN, K.; ANDERSEN, J. A simple interpretation of the surface temperature/vegetation index space for assessment of surface moisture status. Remote Sensing of Enviromenmt, v. 79, p. 213-224. 2002.
• SAYAGO, S.; OVANDO, G.; BOCCO, M. Landsat images and crop model for evaluating water stress of rainfed soybean. Remote Sensing of Environment, v. 198, p. 30-39. 2017.
• SCHIRMBECK, L.W.; FONTANA, D.C.; SCHIRMBECK, J. MENGUE, V.P. Understanding TVDI as an index that expresses soil moisture. Journal of Hyperspectral Remote Sensing, v.7, n. 2, p. 82-90. 2017.
• SCHIRMBECK, L.W.; FONTANA, D.C.; SCHIRMBECK, J. Two approaches to calculate the TVDI in the humid subtropical climate of southern Brazil. Scientia Agricola, v. 75, n. 2, p. 111-120. 2018. doi
  » https://doi.org/10.1590/1678-992X-2016-0315
• SCHIRMBECK, L.W.; FONTANA, D.C.; SCHIRMBECK, J.; BREMM, C. TVDI Obtido de Imagens OLI/TIRS e MODIS. Revista Brasileira de Meteorologia, v. 34, p. 573-583. 2019. doi
  » https://doi.org/10.1590/0102-7786344070
• SCHIRMBECK, L.W.; FONTANA, D.C.; SCHIRMBECK, J.; DALMAGO, G.A.; FERNANDES, J.M.C. Surface moisture index by radiometric measurements and orbital data. Engenharia Agrícola, v. 42, e20210043. 2022a. doi
  » https://doi.org/10.1590/1809-4430-Eng.Agric.v42n2e20210043/2022
• SCHIRMBECK, L.W.; FONTANA, D.C.; SCHIRMBECK, J.; DALMAGO, G.A.; FERNANDES, J.M.C. Water monitoring of soybean crops using the TVDI obtained from surface radiometric sensors. Pesquisa Agropecuária Brasileira, v. 57, p. 1-11. 2022b. doi
  » https://doi.org/10.1590/S1678-3921.pab2022.v57.02581
• SENTELHAS, P.C.; BATTISTI, R.; CâMARA, G.M.S.; FARIAS, J.R.B.; HAMPF, A.C.; NENDEL, C. The soybean yield gap in Brazil: Magnitude, causes and possible solutions for sustainable production. Journal of Agricultural Science, v. 153, p. 1-18, 2015.
• SILVA-FUZZO, D.F.; ROCHA, J.V. Simplified triangle method for estimating evaporative fraction over soybean crops. Journal of Applied Remote Sensing, v. 10, n. 4, 046027. 2016. doi
  » https://doi.org/10.1117/1.JRS.10.046027
• SIVAKUMAR, M.V.K. Agrometeorological strategies for reducing impacts of natural disasters in agriculture. Agrometeoros, v. 24, n. 1, p. 1-13. 2016.
• SON, N.T.; CHEN, C.F.; CHEN, C.R.; CHANG, L.Y.; MINH, V.Q. Monitoring agricultural drought in the Lower Mekong Basin using MODIS NDVI and land surface temperature data. International Journal of Applied Earth Observation and Geoinformation, v. 18, p. 417-427. 2012.
• SUN, L.; CHEN, Z.; GAO, F.; ANDERSON, M.; SONG, L.; et al. Reconstructing daily clear-sky land surface temperature for cloudy regions from MODIS data. Computers & Geosciences, v. 105, p. 10-20. 2017.
• TSUKAHARA, R.Y.; OLIVEIRA, A.N.; OLIVEIRA JúNIOR, J.I.; KOCHINSKI, E.G.; PRESTES NETO, J.; et al. A.C. Pesquisa e divulgação técnica de informações agrometeorológicas aos associados das Cooperativas ABC. Agrometeoros, v. 24, n. 1, p. 71-85. 2016.
• UNIYAL, B.; DIETRICH, J.; VASILAKOS, C.; TZORAKI, O. Evaluation of SWAT simulated soil moisture at catchment scale by field measurements and Landsat derived indices. Agricultural Water Management, v. 193, p. 55-70, 2017.
• WANG, H.; HE, N.; ZHAO, R.; MA, X. Soil water content monitoring using joint application of PDI and TVDI drought indices. Remote Sensing Letters, 11, n.5, 455-464, 2020. doi
  » https://doi.org/10.1080/2150704X.2020.1730469
• XU, C.; QU, J.J.; HAO, X.; COSH, M.H.; PRUEGER, J.H.; et al. Downscaling of surface soil moisture retrieval by combining MODIS/Landsat and in situ measurements. Remote Sensing, v. 10, n. 210, 2018. doi
  » https://doi.org/10.3390/rs10020210
• ZANON, A.J.; STRECK, N.A.; GRASSINI, P. Climate and management factors influence soybean yield potential in a subtropical environment. Agronomy Journal, v. 108, p. 1447-1454. 2016. doi
  » https://doi.org/10.2134/agronj2015.0535

Internet Resources

• GeoR Package, http://www.emater.tche.br/site/info-agro/informativo_conjuntural.php#.YFXvLK9Kg2w
  » http://www.emater.tche.br/site/info-agro/informativo_conjuntural.php#.YFXvLK9Kg2w

Publication Dates

• Publication in this collection
  08 Jan 2024
• Date of issue
  2023

History

• Received
  05 June 2023
• revised
  22 Sept 2023
• Accepted
  19 Oct 2023