GIScience I: Introduction to Geographic Information Science Final Project

Using Unmanned Aerial Systems to Assess Water Needs and Improve Irrigation Efficiency for Northwest Berry Crops

Jesse Carroll

GEOG 560


Topic Selection

My topic will be assessing the use of remote sensing and Unmanned Ariel Systems (UAS) to develop crop water stress indices, estimate canopy cover and crop evapotranspiration, and develop crop coefficients in blueberry, blackberry, and raspberry. I will also be looking for ways to schedule irrigation with these new techniques. Below is an annotated bibliography of peer reviewed journal articles pertaining to the topics mentioned previously. References are cited according to the American Society for Horticultural Science guidelines

Annotated Bibliography

1) Acharya, S.M., S.S. Pawar, and N.B. Wable. 2018. Application of remote sensing and GIS in agriculture. Int. J. of Adv. Engin. Res. and Sci. 5(4):63-65.

In this article Acharya et al. gave a brief synopsis of the importance of remote sensing and GIS in agriculture, as well as ways to apply these technologies to improve production. The authors also explained briefly how GIS and remote sensing work, and why they are important not only in India, but across the world. While no experiments were conducted, this article provided more background on how other countries are applying GIS and remote sensing in agriculture. This article provides more support for my project because it confirms that other countries see the benefits of using remote sensing and GIS to improve agricultural production.

2) Behmann, J., J. Steinrücken, and L. Plümer. 2014. Detection of early plant stress responses in hyperspectral images. J. of Photogrammetry and Remote Sensing. 93:28-111.

Behmann, a researcher in the department of Geoinformation at the University of Bonn, Germany, proposed a new method for the detection of drought stress in higher plants using hyperspectral images. This method looks at the distribution of leaf senescence in a plant. Additionally, the study examined how using that info can detect drought stress before it is visible to the naked eye and in some cases ten days before another vegetative index can detect water stress. Spectram imagery was collected using wavebands between 430 nm and 890 nm. Machine learning, both supervised and unsupervised, were employed to analyze the spectral information gathered from these wavebands then create features and based on the label information create a final ordinal SVM model that can be used to detect crop stress. This paper provides insight on how machine learning can be used to assess hyperspectral images. For my project, this information provides another potential way to analyze the data that will be collected and possibly assess it for drought stress. Since a portion of my project focuses on developing crop water stress thresholds, and assessing plant water needs, this information is quite helpful.

3) Blackburn, G. A. 2006. Hyperspectral remote sensing of plant pigments. J. of Exp. Bot. 58(4)855-867.

In this paper, Blackburn summarizes the technology and methods that are used to quantify pigments using hyperspectral imaging. Additionally, Blackburn provides information on each pigment, its reflectance, problems facing detection with hyperspectral imaging, and what each pigment can identify when imaged. Furthermore, Blackburn touched on emerging applications of hyperspectral imaging. This paper provides a wealth of background on how hyperspectral imaging of plant pigments functions. It also provides information on methods, technology, and applications for imaging. While this information doesn’t specifically touch on a GIS program, or its application, it is background information that needs to be known in order to understand the results that are gained from the GIS analysis performed on my study.

4) Egea, G., C.M. Padilla-Díaz, J. Martinez-Guanter, J.E. Fernández, and M. Pérez-Ruiz. 2017. Assessing a crop water stress index derived from aerial thermal imaging and infrared thermometry in super-high density olive orchards. Agric. Water Manage. 187:210-221.

Egea et al. assessed the potential of deriving a crop water stress index with aerial thermal and infrared imaging for olive trees grown in super-high density orchards. The olive trees were subjected to three different irrigation treatments. Images were obtained by infrared temperature sensors that were installed 1 m above the canopies. Thermal cameras were installed on an unmanned aerial vehicle and flights were performed on five different dates throughout the season. Data from these images was extracted using R instead of a GIS software and crop water stress index values were obtained. These values showed a difference in water stress between each treatment. These results lead the authors to conclude that using thermal and infrared imaging to assess crop water stress is a potential tool. This paper supports my project and its goal of developing crop water stress indices for various berry crops using remote sensed thermal and infrared images collected by an unmanned aerial system. While this study did not use a GIS software, the basic idea of how the crop water stress index was derived is the same.

5) Gómez-Candón, D., J. Torres-Sanchez, S. Labbé, A. Jolivot, S. Martinez, and J.L. Regnard. 2017. Water stress assessment at tree scale: High-resolution thermal UAV imagery acquisition and processing. Acta Hortic. 1150:159-165.

Gómez-Candón et al. discussed assessment of water stress in apple trees using thermal images collected by an Unmanned Aerial Vehicle (UAV). The images were collected beginning at solar noon and four more times throughout four different dates in an apple orchard located in France. The images were then processed and analyzed using a GIS program. The camera was calibrated using ground thermal targets while the UAV was in flight. Indices were then extracted from the thermal images and evaluated for canopy temperature and crop water stress. While significant differences were found, there was a lot of variation between genotype causing the author to conclude that a more finely tuned analysis would be needed to determine responses to water stress. This paper outlines methods that could be applied for the development of crop water stress in berry crops, as well as statistical analyses that can be used for the data collected in my project.

6) Gontia, N.K., and K.N. Tiwari. 2010. Estimation of crop coefficient and evapotranspiration of wheat (Triticum aestivum) in an irrigation command using remote sensing and GIS. Water Resour. Manage. 24:1399-1414.

Gontia and Tiwari outlined a method and reported results that were obtained from an experiment looking at crop coefficient and evapotranspiration in irrigated wheat. Using images obtained from a remote sensing satellite, the authors used GIS software to obtain the Normalized Difference Vegetation Index and the Soil Adjusted Vegetation Index from a wheat field. Using this data, the relationship between these indices and the estimated crop coefficients were then developed. Using a regression equation, the authors then estimated the monthly reference crop evapotranspiration. Using the data gathered from these estimates the authors concluded that the total crop water demand for wheat was 1.21 million cubic meters. This paper provides a basis of estimating crop evapotranspiration by using GIS software. Additionally, it confirms that NDVI is accurate in estimating both crop ET and can be used in crop coefficient estimations.

7) Iatrou, G., S. Mourelatos, S. Gewehr, S. Kalaitzopoulou, M. Iatroum, and Z. Zartaloudis. 2017. Using multispectral imaging to improve berry harvest for wine making grapes. Ciência Téc. Vitiv. 32(1):33-41.

Iatrou et al. evaluated the use of multispectral imaging for the determination of berry ripeness and harvest time for wine making grapes. By using a carotenoid reflectance index (CRI) ripeness and harvest time were determined from multispectral images. Once the grapes were harvested, they were evaluated for total soluble solids, pH, and terpenes. Two CRIs were evaluated. The authors used multiple linear regression and a support vector machine to evaluate the data. It was concluded that CRI 2 was the most accurate at determining ripeness and harvest time. However, the authors stated that this was preliminary methodological work and more research would be needed. This paper provides information on the statistics that would be beneficial for my proposed study, background on different indices that can be used with multispectral imaging, as well as potential methodologies that could be helpful if berry quality wanted to be evaluated in any crops contained in this study.

8) Jones, H.G. 2014. Remote sensing of plant stresses and its use in irrigation management. Acta hortic. 1038:239-247.

Jones, a researcher from the University of Dundee at SCRI, compiled different ways in which remote imaging can be used to schedule irrigation, monitor deficit developments and plant responses, and evaluate canopy cover development in horticultural crops. This paper also discussed the benefits and drawbacks of multiple methods used in capturing the images and evaluated vegetative indices used to evaluate the data collected. Furthermore, this paper discussed the advantages and disadvantages of applying remote sensing to crop management practices and how the accuracy of these methods impacts their potential usefulness. This paper provided a basis on which remote sensing techniques, sensors, and GIS would be most beneficial to use while attempting to develop crop coefficients and deficit thresholds, schedule irrigation, and monitor canopy development in various berry crops. Additionally, it brought to light some of the potential drawbacks or problems that may be encountered in my proposed study.

9) Jones H.G., R. Serraj, B.R. Loveys, L. Xiong, A. Wheaton, and A.H Price. 2009. Thermal Infrared imaging of crop canopies for the remote diagnosis and quantification of plant responses to water stress in the field. Functional Plant Bio. 36:978-989.

In this article, Jones et al. examined the use of thermal imaging on crop canopies for the identification of plant responses to water stress in the field. Jones et al. began with an overview of how thermal sensing works and discussed difficulties and benefits of the method proposed in this paper. Additionally, the article examines other uses such as phenotyping, leaf temperature variation, canopy structure, and more. The experiments were performed in a vineyard for Cabernet-sauvignon grapes. The authors found that thermal imaging does indeed provide a way to examine crop canopies for plant responses to water stress. While this paper does not specifically mention GIS software, the principles of thermal imaging can be used in such software to obtain the same information. Additionally, this paper provided background information on how thermal imaging is able to detect water stress in plants, and various applications that can be developed from these methods. This information helps to support the ideas proposed in my project.

10) Liu, J. 2009. A GIS-based tool for modelling large-scale crop-water relations. Environ. Modelling and Software. 24:411-422.

Liu, a scientist at the Swiss Federal Institute of Aquatic Science and Technology, proposed a tool for modelling crop-water relations on a large scale. To accomplish this, a GEPIC model was created and data was analyzed in a GIS program. Crop water potentials for wheat, maize, and rice were them simulated using data including crop evapotranspiration, soil evaporation, and crop growth (Leaf Area Index). These simulations showed that the GEPIC model provided a flexible tool to study crop water relations over large-scale areas. Liu cautioned though, that the accuracy of this model depends on the quality and accuracy of the data that is provided for the model. Additionally, Liu outlines a few more potential drawbacks of large scale estimation. This article provided a method for estimating crop water potential on a larger scale which may be helpful on my proposed project if it can be adapted to a smaller scale. Either way, the background information was informative and many of the pitfalls, drawbacks, and challenges that my project may face are discussed.

11) Madonsela, S., M.A. Cho, A. Ramoelo, O. Mutanga, and L. Naidoo. 2018. Estimating tree species diversity in the savannah using NDVI and woody canopy cover. Int. J. Appl. Earth Obs. Geoinformation. 66:106-115.

Mandonsela et al. assessed the potential of using NDVI to assess woody canopy cover and species diversity in the African Savannah. The idea behind using woody canopy cover came from the attempt to remove influences in NDVI from grass. Grass causes an overestimation of species diversity in research that uses NDVI of canopy cover because NDVI takes into account all photosynthetically active vegetation. Using NDVI and woody canopy cover during senescence provides a way to circumvent this previously mentioned issue. Data was collected from Landsat-8 Operational Land Imager satellite images captured on four different dates. This data was compared with data collected in the field to determine the accuracy of species diversity estimated by evaluating woody canopy cover. Linear regression models were fit to determine the best model. The authors concluded that combining NDVI with woody canopy cover is the best model for estimating species diversity during senescence. This paper provided information on how to use linear regression models to predict outcomes using NDVI data. This will be helpful in the proposed study to use NDVI for canopy cover and growth stage estimation.

12) Mangus, D.L., A. Sharda, and N. Zhang. 2016. Development and evaluation of thermal infrared imaging system for high spatial and temporal resolution crop water stress monitoring of corn within a greenhouse. Computers and Electronics in Agric. 121:149-159.

In this paper, Mangus et al. discuss the use of thermal imaging to assess crop water stress monitoring in corn. Using a thermal infrared imaging system, images were captured of corn canopy temperature under various treatments of water stress. Data from these images was then extracted using imaging processing software. Additionally, a live stream video was also collected. After the data was analyzed the authors found that real time crop water stress could be applied to each image. Additionally, the progression of the water stress could be observed over time. The results obtained were accurate enough for the authors to state that this method provides an unprecedented opportunity to study advanced relationships of plants in spatial and temporal resolution. While this study did not use a GIS program, the idea behind constantly monitoring crop water stress could be modified and scaled up. Additionally, it provides more support for using remote sensed thermal infrared images to estimate crop water stress. This provides background and support for my proposed project.

13) Nandy, S., S.P.S. Kushwaha, V.K. Dadhwal. 2011. Forest degradation assessment in the upper catchment of the river Tons using remote sensing and GIS. Ecological Indicators. 11:509-513.

In this paper, the authors investigate the use of remote sensing and GIS to assess forest degradation in the upper catchment of the river Tons. Using satellite points, topography, and GPS information a map layout of the study area was created. The slope was generated using a Shuttle Radar Topography Mission digital elevation model. Using this map, forest cover was interpreted and categorized into four classes. This study found that this method was 89.5% accurate in detecting forest degradation. Additionally, it was able to map the other environments as well, including grasslands, agriculture areas, rock outcrops, snow, and the river area. Using this information, the authors were able to make a case of instituting methods to prevent erosion in areas with degraded forests. This study, while not directly involved with agriculture, shows that GIS can be used to map agriculture lands. Additionally, forest degradation was identified using canopy cover data, which is a key part of my proposed project.

14) Ray, S.S. and V.K. Dadhwal. 2001. Estimation of crop evapotranspiration of irrigation command area using remote sensing and GIS. Agric. Water. Manage. 49:239-249.

In this paper, Ray and Dadhwal outline the use of remote sensed data processed in a GIS program to estimate crop evapotranspiration in irrigated fields. Using a reference crop evapotranspiration map and crop coefficients for various crops, a seasonal crop evapotranspiration map was created for both wheat and tobacco. The study thoroughly explained the procedure they developed, as well as the benefits, drawbacks, and challenges that were faced. This paper, while older, provides a good starting point for my project. It clearly outlines the use of a GIS program to estimate crop evapotranspiration, which is necessary when developing more accurate crop coefficients.

15) Shafian, S., N. Rajan, R. Schnell, M. Bagavathiannan, J. Valasek, Y. Shi, and J. Olsenholler. 2018. Unmanned aerial systems-based remote sensing for monitoring sorghum growth and development. PLOS ONE. 13(5):1-15.

Shafian studied the feasibility of using unmanned aerial system remote imaging for assessing crop growth (leaf area index, fractional vegetation cover, and yield) in Sorghum bicolor. This study used a fixed-wing unmanned aerial vehicle (UAV) with a multispectral sensor for image acquisition. Flights were done three times during the growing season and the images were processed after for analysis. Additionally, in-field samples were taken for leaf area index, fractional vegetation cover, and yield which were used to validate information collected from the UAV. This study found that images collected by UAV can be used effectively for data collection on agricultural fields. This study validates the relationship between NDVI and fractional vegetation cover and that this data could be used to estimate leaf area index and fractional vegetation cover by use of images obtained from an UAV. Additionally, a high correlation was found between NDVI with fractional vegetation cover, leaf area index, and yield leading to possible application of yield estimation in the agriculture field. This study provided information on using multispectral data to create canopy cover maps of entire fields through a GIS program. My project requires mapping entire fields and estimating canopy cover, water stress, and etc, so the methods are quite applicable. Additionally, this study provided more information on the relationships of NDVI, fractional vegetation cover, and yield.

16) Trout, T., L.F. Johnson, and J. Gartung. 2008. Remote sensing of canopy cover in horticultural crops. HortScience. 43(2):333-337.

Trout, a research scientist at the USDA-ARS Water Management Research Unit evaluated the use of remote sensing for estimation of canopy cover and growth stages using Normalized Vegetative Difference Index (NDVI) in horticultural crops. To accomplish this images were collected using a camera suspended above 11 different horticultural crops. The NDVI from these images were extracted and compared with data obtained from Landsat 5 satellite imagery. Trout concluded that remote sensing could be used to monitor growth stage and canopy cover development in horticultural crops. This paper supports the efforts of my study which attempts to assess canopy cover development and growth stages of berry crops by using images gathered by unmanned aerial systems and processed with a GIS program. Additionally, this paper mentions the possibility of assessing irrigation water demand with remote sensing in crops as well. The methods and data analysis used in this paper provides additional information that will be helpful for this project.

17) Tsanis, I., S. Naoum, and S.J. Boyle. 2009. A GIS interface method based on reference evapotranspiration and crop coefficients for the determination of irrigation requirements. Water International. 27(2):233-242.

In this paper, Tasnis et al. used GIS software to apply a reference evapotranspiration calculation procedures to data in order to derive irrigation requirements. Meteorological data was collected from the years of 1963, 1971, 1981, 1991, and 1992. This data was then used to create a feature in the respected layer. The authors then outlined the method potential users would use to analyze the data. Results were only calculated on the data from 1991. These results were accurate enough for the authors to feel that they could estimate potential water use in the year 2020. This estimation came out to 8,350Mm3 of average irrigation requirements. The authors did state that the irrigation efficiency could improve by as much as 20% by the year 2020, and provided a corrected estimate if that ended up being the case. This paper, while old, supports my project and effort to estimate crop evapotranspiration and irrigation requirements using remote sensing and GIS software. While the methods are slightly outdated it provides a depth of information on how well these methods age with time.

18) Zhou, M.C., H. Ishidaira, H.P. Hapuarachchi, J. Magome, A.S. Kiem, K. Takeuchi. 2006. Estimating potential evapotranspiration using Shutleworth-Wallace model and NOAA-AVHRR NDVI data to feed a distributed hydrological model over the Mekong River basin. J. of Hydrology. 327:151-173.

In this paper, Zhou et al. used Arc/info software to create a model of estimated potential evapotranspiration obtained through NDVI data over river basins in southeast Asia. This paper also wanted to evaluate long term potential evapotranspiration (PET). Using publicly available data, highly accurate estimates of PET were obtained using the model and methods proposed in this paper. The method used to estimate PET was chosen because the river basin has a wide range of vegetation cover. While the method of estimating PET in this paper would not apply to my project because I will primarily be looking at developed canopies in a monoculture, the ideas of estimating evapotranspiration are still the same. Additionally, the use of a GIS system to map basin PET can be modified to map crop evapotranspiration, which is a necessary data point needed in irrigation scheduling of berry crops. Furthermore, the pitfalls and obstacles noted in this paper provide information on potential problems I might encounter as well.

19) Zhou, J., L.R. Khot, R.A. Boydston, P.N. Miklas, and L. Porter. 2018.  Low altitude remote sensing technologies for crop stress monitoring: A case study on spatial and temporal monitoring of irrigated pinto beans. Precision Agric. 19:555-569.

In this study, the authors assessed the feasibility of using multispectral and infrared thermal imaging sensors attached to a small unmanned aerial system (UAS) to assess monitoring drought stress in pinto beans (Phaseolus vulgaris L.). To assess this method, pinto bean plots were irrigated with differing rates. After the treatments were applied, flights with the UAS were conducted at three different growth stages. Images were collected and data was extracted. This data included green normalized vegetation index (GNDVI), canopy cover, and canopy temperature. Data collected on the ground included leaf area index and crop yield. This field data was correlated with the UAS collected data. This study found that GNDVI, canopy cover and canopy temperature were accurately assessed by the UAS. Additionally, these indicators were able to accurately detect crops that had been water stressed. Furthermore, water stressed plants identified by the UAS were strongly correlated with reduced yield. This study supports the methods that I plan to use for identifying crop water stress in berry crops. It supports my hypothesis that crop water stress can be monitored with an UAS and that thresholds can be established for each plant that will provide information on potential yield loss at different water stress thresholds.

Author Information

Jesse Carroll | Graduate Student | Oregon State University | Horticulture

Questions? Send me an Email!