Estimation of crop transpiration and its scale effect based on ground and UAV thermal infrared remote sensing images

https://doi.org/10.1016/j.eja.2021.126389Get rights and content

Highlights

  • A method was proposed to monitor spatiotemporal variations of transpiration during the whole growth period of crops over heterogeneous farmland.

  • The estimated transpiration was basically consistent with the measured transpiration, with a high R2 (>0.83) and low RMSE (≤0.1 mm/h).

  • The mean transpiration of full-irrigated, medium-irrigated, and low-irrigated corn was 0.72, 0.63, and 0.59 mm/h, that of well-irrigated and no-irrigated soybean was 0.77 and 0.27 mm/h.

  • Although T became slightly underestimated as the pixel scale rose from ground to UAV, the scale effect of estimating T using ground and UAV thermal images is acceptable with small difference between T-ground and T-UAV, especially when the crop canopy is well closed (R2>0.8).

Abstract

Accurate monitoring of crop transpiration (T) is essential for saving and managing irrigation water resources. Ground and UAV (Unmanned Aerial Vehicle) thermal infrared (TIR) images have high enough resolution to meet the needs of precision agriculture. At present, they are mostly used to assess crop water stress instead of to monitor T, especially through UAV TIR images. And the pixel scale effect on T estimation based on ground (T-ground) and UAV (T-UAV) images remains less explored. In this study, the diurnal T of regulated deficit-irrigated corn and soybean was estimated using a three-temperature (3T) model paired with ground and UAV TIR remote sensing images. This method was verified with hydrogen-oxygen stable isotopes measurements, and the pixel scale effect on estimating T with it was analyzed by comparing the LST (Land Surface Temperature) and T derived from ground and UAV TIR images. The results show that the method can accurately estimate T with a high coefficient of determination (R2, 0.84–0.88) and a low root mean square error (RMSE, 0.099–0.104 mm/h) and mean absolute percentage error (MAPE, 12.33 %–12.58 %). Contrary to LST, T-ground and T-UAV decreased as the water stress increased, and their peaks appeared around 15:00. During the experimental period, the mean T of full-irrigated, medium-irrigated, and low-irrigated corn was around 0.72, 0.63, and 0.59 mm/h, respectively, and that of well-irrigated and non-irrigated soybean was 0.77 and 0.27 mm/h, respectively. As the pixel scale increased, the T became slightly underestimated, and the scale effect increased with the complexity of the farmland. This happened mainly because the larger pixel scale of UAV images leads to more mixed pixels and greater soil-vegetation interaction, resulting in overestimation of the canopy temperature. But the T-UAV was highly consistent with the T-ground, with an R2, a RMSE, and a MAPE of 0.60–0.89, 0.06–0.14 mm/h, and 7.39 %–15.51 %, respectively. Therefore, the scale effect on estimating T is acceptable, especially when the canopy is more closed. The proposed method is concluded to be feasible for use in estimating T over heterogeneous farmland.

Introduction

Evapotranspiration (ET), which links the water, energy, and carbon cycles, plays an important role in driving climate patterns (Fisher et al., 2017). In arid and semiarid areas where agriculture relies on irrigation, there exists a serious shortage in water resources, so many water-saving irrigation strategies, such as regulated deficit irrigation (RDI), are popular. Furthermore, irrigated water in arid areas is mainly consumed by ET. Over 90 % of the water absorbed by crops is consumed by transpiration (T), and ET is the foremost variable required for agricultural water management (Zang et al., 2012). Therefore, accurate calculation of ET and T is of great significance for calculating irrigation water demand and realizing precision irrigation to promote water saving in agriculture, especially from a spatial perspective.

Although many researchers have used different approaches to study ET (Anapalli et al., 2018; Delogu et al., 2018; Talsma et al., 2018; Tie et al., 2018; Wu et al., 2017; Yao et al., 2017), the ability to accurately capture its spatiotemporal dynamics from leaves to ecosystems and from seconds to years remains limited (Page et al., 2018). For example, more traditional methods, such as using eddy covariance (EC) and weighing lysimeters, can measure ET at a resolution of seconds to months in different regions (Goulden et al., 1996; Liu et al., 2002), but they cannot reflect the spatial variability of ET. Although satellite remote sensing (RS) technology overcomes the shortcomings of traditional methods because of its spatiotemporal continuity and large-span characteristics, the spatial resolution of satellite images is too coarse to provide the submeter-scale resolution data required for precision agriculture. It is worth mentioning that the pixel scales are larger than some small plots in most agricultural regions (Gowda et al., 2008), and its observation frequency is low, which may result in a particular period of crop growth being overlooked. Therefore, the low spatial and temporal resolution limits the use of satellite imagery in precision agriculture (Khanal et al., 2017).

Handheld thermal imaging systems can be used to estimate ET based on ground RS (Page et al., 2018). Improvements such as the reduction of the thermal infrared (TIR) imaging system’s weight and the increased availability of UAVs make it easier to apply the UAV-based thermal imaging system in agriculture (Gago et al., 2015; Sepúlveda-Reyes et al., 2016; Poblete et al., 2017; Santesteban et al., 2017). UAV RS can collect surface information with high spatiotemporal resolution (Dandois and Ellis, 2013; Faye et al., 2016) and fill the gap between ground measurement and satellite RS estimation. Ground and UAV-based thermal RS can quantitatively characterize differences at the centimeter level for pixels, and such RS meets the requirements of precision agriculture. These types to RS have also been used to estimate the leaf area index, crop water stress index, and ET under different conditions (Banerjee et al., 2018; Hoffmann et al., 2016; Ortega-Farías et al., 2016; Hou et al., 2018; Wang et al., 2017).

With the gradual improvement of ground-UAV-satellite RS systems, images with various spatial resolutions from different sources (TM/ETM+, ASTER, MODIS, AVHRR, Fluke, FLIR, and others) have been used for ET observation. However, when images with a range of resolutions are applied to the same model, the output is inconsistent, which limits the application of ET and its components, especially over heterogeneous landscapes (Wang et al., 2016). Surface heterogeneity is used to describe the complexity of the surface scene, and heterogeneity widely exists in various scales when it comes to land surface. The surface area represented by a pixel can be a mixture of many kinds of ground objects on a heterogeneous surface. In this sense, the pixel scale problem has become increasingly prominent when using images with different resolutions to estimate ET. Recently, many researchers have studied the scale effect on ET estimation that comes from using images with different spatial resolutions based on various RS models (Wu and Li, 2009). There are two main methods: one is to scale a high-resolution image up to a low-resolution image and then compare their estimated ETs, and the other is to invert ET independently using satellite images with different resolutions and then compare them directly, a method that is considered to have practical significance (Wang et al., 2016). For example, McCabe and Wood (2006) used three types of satellite images (ETM+, ASTER, and MODIS) independently to study the scale effect on ET estimation over a small watershed in central Iowa, and their results suggest that high-resolution datasets are better able to describe surface heterogeneity.

According to Ershadi et al. (2013), the accuracy of ET estimation decreases with the increase of pixel scale, and it is greatly influenced by surface properties when models contain aerodynamic resistance. In this regard, the three-temperature (3T) model without aerodynamic resistance has been widely used to estimate ET with promising results (Qiu and Ben-Asher, 2010; Luo et al., 2012; Xiong et al., 2012; Tian et al., 2014; Zhang et al., 2019). For example, Hou et al. (2019) and Tian et al. (2020) used the T measured by the LI-6400 portable photosynthesis measurement system and the stable isotope method to verify the soybean and corn T estimated by the 3T model and ground thermal images with a mean absolute percentage error (MAPE) and a coefficient of determination (R2) of 9.53 % and 0.86–0.88, respectively. In addition, Wang et al. (2016) studied the scale effect on ET estimation that uses aerial photos (3 m) and images from ETM+ (60 m), ASTER (90 m), and MODIS (1000 m) combined with the 3T model, and they verified the estimated ET against EC tower measurements with a MAPE of 12.0 %. However, the scale effect on ET estimation that comes from ground and UAV TIR RS images has not been addressed, especially in heterogeneous farmland.

Different RDIs have various effects on crop growth, such as plant height and leaf area, among other characteristics, thus resulting in different canopy structures. And the transpiration rate of crops varies greatly under different types of deficit irrigation. Since T is the central part of ET during the crop reproductive period, this study mainly estimated crop T using the 3T model and analyzed the scale effect on T estimation under RDI. The objectives are to (1) validate the reliability of T estimated using the 3T model combined with ground and UAV TIR images, (2) use the 3T model to estimate the diurnal T of corn and soybean under different levels of water stress, and (3) compare the temporal and spatial characteristics of T estimated by the 3T model based on ground and UAV TIR images to analyze the scale effect on estimating T that comes from using this method.

Section snippets

Experimental site and design

The experiment was conducted at China Agricultural University’s Shiyanghe Experimental Station during the crop reproductive stage. The experimental station is located in the Liangzhou District, Wuwei City, Gansu province, of Northwest China (37°52′20′′N, 102°50′50′′E). The site’s climate is a typical inland desert climate, with an average annual rainfall, pan evaporation, and air temperature of about 164.4 mm, 2000 mm, and 7.8 °C, respectively.

The measurements focused on two typical crops, corn

Verification with T-ES

The findings indicate that the changes in T-3T and T-ES were consistent, T-3T-UAV was closer to T-ES, and T-3T-Ground was higher than them (Fig. 4 (a)). This may be related to the fact that the EC system underestimates ET due to its unclosed energy (Li et al., 2018). The scatter plot shows that the slope and R2 of the best-fit lines between T-3T-UAV and T-ES were 1.023 and 0.838, and those of the T-3T-Ground and T-ES were 1.114 and 0.883, respectively (Fig. 4 (b)). The error analysis of the

Discussion

TIR RS based on ground and UAV approaches not only overcomes the problem of spatial changes being difficult to spot in traditional ground point observations but also makes up for the shortage of satellite RS with low spatial and temporal resolution. It can better meet the needs of precision agriculture and detect water consumption in farmland. In this study, the daytime variations of T of RDI corn and soybean were estimated using the 3T model combined with ground and UAV TIR images. The 3T

Conclusions

In this study, the transpiration of corn and soybean under RDI was estimated using a method that combines the 3T model with ground and UAV TIR images. The main advantage of this method is that it requires fewer input parameters and simplifies the application of RS. The estimated transpiration was evaluated against the measurements based on the EC system and the stable isotope technique, and the scale effect of estimating T based on ground and UAV TIR images was analyzed. The conclusions are as

Author contributions

Conceptualization, Fei Tian; Methodology, Mengjie Hou, Tong Zhang and Fei Tian; Software, Mengjie Hou; Writing-original draft preparation, Mengjie Hou; Writing-review & editing, Fei Tian, S. Ortega-Farias, C. Riveros-Burgos and Aiwen Lin; Project administration, Fei Tian; Funding acquisition, Fei Tian.

Funding

This work was supported by the International and regional cooperation and exchange projects of the National Natural Science Foundation of China (51961125205) and National Agency for Research and Development (ANID)/PCI (NSFC190013), the financial support provided by the Major Program of the National Natural Science Foundation of China (51790534).

Declaration of Competing Interest

The authors report no declarations of interest.

References (46)

  • M.F. McCabe et al.

    Scale influences on the remote estimation of evapotranspiration using multiple satellite sensors

    Remote Sens. Environ.

    (2006)
  • G.F.M. Page et al.

    Spatiotemporal dynamics of leaf transpiration quantified with time-series thermal imaging

    Agr. Forest Meteorol.

    (2018)
  • G.Y. Qiu et al.

    An improved methodology to measure evaporation from bare soil based on comparison of surface temperature with a dry soil surface

    J. Hydrol.

    (1998)
  • G.Y. Qiu et al.

    Theoretical analysis of a remotely measurable soil evaporation transfer coefficient

    Remote Sens. Environ.

    (2006)
  • L.G. Santesteban et al.

    High-resolution UAV-based thermal imaging to estimate the instantaneous and seasonal variability of plant water status within a vineyard

    Agr. Water Manage.

    (2017)
  • C.J. Talsma et al.

    Partitioning of evapotranspiration in remote sensing-based models

    Agr. Forest Meteorol.

    (2018)
  • F. Tian et al.

    Estimation of evapotranspiration and its partition based on an extended three-temperature model and MODIS products

    J. Hydrol.

    (2013)
  • F. Tian et al.

    Use of high-resolution thermal infrared remote sensing and "three-temperature model" for transpiration monitoring in arid inland river catchment

    J. Hydrol.

    (2014)
  • F. Tian et al.

    Salinity stress effects on transpiration and plant growth under different salinity soil levels based on thermal infrared remote (TIR) technique

    Geoderma

    (2020)
  • Q. Tie et al.

    Comparing different methods for determining forest evapotranspiration and its components at multiple temporal scales

    Sci. Total Environ.

    (2018)
  • Y.Q. Wang et al.

    Is scale really a challenge in evapotranspiration estimation? A multi-scale study in the Heihe oasis using thermal remote sensing and the three-temperature model

    Agr. Forest Meteorol.

    (2016)
  • T. Zhang et al.

    Estimation of transpiration and canopy cover of winter wheat under different fertilization levels using thermal infrared and visible imagery

    Comput. Electron. Agr.

    (2019)
  • J. Bai et al.

    Characterizing the footprint of eddy covariance system and large aperture scintillometer measurements to validate satellite-based surface fluxes

    IEEE Geosci. Remote. Sens. Lett.

    (2015)
  • Cited by (13)

    • Estimation of sugar content in sugar beet root based on UAV multi-sensor data

      2022, Computers and Electronics in Agriculture
      Citation Excerpt :

      UAV (Unmanned aerial vehicle) has become one of the most popular methods to obtain crop canopy features due to its speed, flexibility, low cost and high temporal and spatial resolution (Maes and Steppe, 2019; Feng et al., 2021). These canopy features obtained by UAV include spectral, structural, and thermal features and have been used in the estimation of various crop physical (Yan et al., 2019; Che et al., 2020), physiological (Zhu et al., 2020; Yang et al., 2021), and biochemical metrics (Osco et al., 2019; Hou et al., 2021). Spectral features of the canopy obtained by UAV equipped with multispectral and hyperspectral camera are widely used to estimate various physiological and biochemical metrics of crops.

    View all citing articles on Scopus
    View full text