Satellite vs. UAVs Remote Sensing of Arid Ecosystems: A Review with in an Ecological Perspective

The management of arid ecological systems is a key component for long-term sustainability and has been recognized globally as arid regions cover 41% of the earth’s land surface and contain 38% of the human population [1,2]. Arid ecosystems are dramatically influenced by human activities resulting in desertification and soil compaction leading to losses in ecosystem services [3-5]. Assessing and monitoring arid ecosystems is difficult due to the complexity of the ecological system as several elements including vegetation, animals, soils, and climate are all connected together. However, remote sensing is one of the most powerful technologies in assessing and monitoring the ecological system due to its contribution in providing large amount of data, which helps in answering critical ecological questions [6-8]. Several studies show that remote sensing technology can provide powerful information in the field of ecology by enabling scientists and researchers with decent information to analyze dynamic changes in landscapes [6,9,10], evaluate the distribution of vegetation cover [7,11,12] and monitor landscape degradation [13-15]. Additionally, the integration of RS and GIS was successfully examined in our previous work in Kuwait [16-19], which assisted in:

a) Developing a site history for disturbed locations to determine the major source of disturbance,

b) Selecting reference/or control sites for the restored areas,

c) Monitoring the dynamic changes in vegetation in a war affected area,

d) Evaluating the potential soil loss at the restored location, and

e) Selecting optimum locations for revegetation.

Thus, some limitations were found through the use of satellite imagery when dealing with arid ecosystems [20], but modern RS technologies including low-altitude unmanned aerial vehicles (micro-UAVs) may resolve some of these limitations [21]. Therefore, this paper focuses on reviewing the main challenges and limitations of RS satellite imageries in the field of ecology in arid ecosystems, and whether UAVs can resolve such limitations.

Even though RS applications using satellite imagery showed a decent amount of useful information, several limitations and challenges were detected when dealing with arid ecosystems. These concerns include the small size and low amount of vegetation cover in arid landscape, which makes it difficult to detect the coverage and specific types of desert species due to the high reflectance of the soil background, variable mixture of green and senescent grasses (shrubs and herbs), and multiple reflectance from open canopies and bright soils [20]. Moreover, common RS vegetation indices and methods are insensitive to detect non-photosynthetic vegetation (NPV) which include dry / dead shrubs, leafless drought deciduous plants, and senescent plants [22]. NPVs are considered the major component of the total desert surface cover, and play an important role in vegetation dynamics in arid ecosystems [23]. They may also be important, especially when studying soil erosion (both wind and water), as they play a critical role in preventing the risk of soil loss [16,23].

Data acquisition for required time periods is also an issue with satellite imageries. There are some restrictions in the availability of satellite imagery data as satellites travel along geosynchronous orbits providing only a partial view of the earth with specific timing which may be an issue for time sensitive applications [24]. This is important for assessing and monitoring vegetation dynamics as arid plants are more distinct and visible in specific seasons. Spectral resolution is also important in identifying and mapping arid ecosystem features. The broad spectral bands of multispectral sensors such as Landsat Thematic Mapper (TM) and SPOT XS as well as medium spatial resolution sensors such as IKONOS greatly reduce their ability to spectrally discriminate between arid ecosystem features that have similar reflectance. However, hyperspectral sensors, with higher spectral and spatial resolutions, have the ability to identify arid ecosystem features in more details than multispectral sensors [17]. Hyperspectral data can provide reliable results for identifying arid ecosystem features as they expand and improve the spectral capability of multispectral sensors thus providing more accurate results for mapping [25]. However, hyperspectral data are not commonly used due to the large expenditures involved in acquiring such information. From previous work it can be seen that the multispectral and hyperspectral sensors showed similar results of vegetation cover depending on the classification method, and the number and types of classes within the landscape. For instance, hyperspectral sensors may differ when mapping more complex areas with multiple classes such as urban areas. However, it may not show significant differences when dealing with the mapping of an open landscape covering a few classes such as vegetation and bare soils [18].

In addition, the capability of satellite remote sensing to identify and map arid ecosystem features can be limited by the accuracy of the geographic location of the target feature and radiometric factors (e.g. atmospheric scattering). The geometric distortion (e.g. platform motion, optical and scanner distortions, and target motion such as the Earth’s rotation) can limit the accuracy of the geographic locations [26]. In addition, radiometric effects, such as atmospheric scattering, can influence the amount of radiation reaching the sensors and therefore reduce the accuracy of remotely sensed satellite data. Consequently, geometric and radiometric limitations of sensors should be considered and corrected before processing any imaging analysis.

With the development of modern RS technologies, most of these challenges and issues could be resolved. In the past few years, rapid increase has been seen in the use of low-altitude unmanned aerial vehicles (micro-UAVs), and LIDAR (Light Detection and Ranging) sensors (including airborne and terrestrial LIDAR) [21]. These technologies can be used in several ecological applications, including detailed ecological studies [27] and biodiversity assessment [21]. One of the major advantages of UAVs is the flexibility of data acquisition for required time periods as well as the low cost for capturing data. This is important in studying ecological systems, and especially vegetation dynamics due to the seasonal dynamics and spectral characteristics of vegetation as desert plants are often more distinct and visible in specific seasons [24,28]. They also provide much higher spatial resolution that ranges from meters to millimeters, which is relevant in assessing several ecological processes [27]. Such methods are also helpful in monitoring and modeling ecological issues, including plant population and communities, soil condition, and plant monitoring, which includes estimations of biodiversity and biomass [21] as LIDAR sensors are able to directly measure the function of height [29].

The image resolution captured by UAVs also depends on the flight altitude and the size of the area covered [30]. The higher the altitude of the UAVs, the larger the area covered. However, covering large area with UAVs will lower the resolution of the imagery, limiting feature identification in the image frame. In addition, the technical camera/sensor specifications (i.e. camera / sensor focal length) can also affect the quality of UAV images. These specifications can validate the capability of the UAVs in vegetation identification in arid ecosystems.

In addition, UAVs require a real-time weather information. For example, the information on wind speed and direction, atmosphere visibility, and cloud coverage can all affect the quality of UAV images. Selecting the optimum flight time can be an advantage of the UAVs. However, the uses of UAVs could be limited in arid ecosystem areas with consistently bad weather conditions.

The question here focuses on whether UAVs imagery can replace satellite imagery. The answer for this question depends on the purpose and scale of the project. Satellite imagery and UAVs imagery differ according to the type of information where each has its own advantages and disadvantages (Table 1). If the project focuses on a large scale, then satellite imagery could be the best option as UAVs are unable to cover large scale landscapes. Figure 1 represents an example showing the differences between UAVs and Landsat 8 imagery (medium resolution imagery) at Al- Abdali agriculture area located in the north of Kuwait. The Landsat imagery covers north portion of the country including Abdali agriculture area in one image which is useful for large scale objectives such as assessing the changes in agriculture area, and detecting the green areas within the agriculture area which is difficult to study using UAVs. The UAV image covered only one farm in the agriculture area with a very high resolution image which was captured using Parrot Disco Pro-AG with Sequoia Multispectral camera. 300 images were mosaicked together to cover one farm within the agriculture area with a very high resolution, whereby small species such as the shrubs were easily detected from the UAVs imagery. However, the processing time for the UAVs imagery was long compared with the satellite imagery which was faster. Figure 2 shows another example that compares high resolution Pleiades satellite imagery with UAV (Phantom 4 Pro drone) imagery for Jahra pools nature reserve area, State of Kuwait. The high resolution imagery shows a high level of details (e.g. spectral, shape, context and texture) of the area over a large scale. Whereas, imagery obtained by UAV (Phantom 4 Pro drone) covers a small scale area with higher details of vegetation information (e.g. species type, size, health, etc.) than that of the satellite imagery.

Table 1: UAVs vs. landsat imagery.

Figure 1: Example of Abdali agriculture area conducting Landsat 8 and UAVs. the location of the agriculture area obtained from Landsat 8 imagery, PAFF farm which covers native desert species, the image was obtained using UAVs, and zoom in to the Arial imagery, shrubs can be easily detected through the imagery.

Figure 2: Example of Jahra pools nature reserve area conducting high resolution Pleiades satellite image and UAVs (Phantom 4 Pro Drone).

(A) Vegetation from arid ecosystem area and river side obtained from high resolution Pleiades satellite imagery.

(B) The species type and size of vegetation, and the vegetation cover near the river appear clearly in the image obtained using UAVs (Phantom 4 Pro Drone).

Assessing and monitoring land cover of arid ecosystems is complicated due the small size and low amount of vegetation coverage. RS satellite imageries showed good results in assessing land cover in arid ecosystems, but there are some limitations due to the difficulty of the RS methods and indices in detecting nonphotosynthetic vegetation (NPV). However, most of these issues could be resolved by using very high resolution imagery obtained from UAVs, which provide decent information for small scale projects due to the very high spatial resolution. Yet, the fine spatial resolution is not always beneficial, it depends on the scale and purpose of the study. For large scale projects, satellite imageries will be better in detecting the healthy vegetation coverage. UAVs are powerful in monitoring and modeling detailed ecological issues within a small scale, including plant populations and communities, soil condition, and plant monitoring, which also includes estimations of biodiversity and biomass. Nevertheless, we believe that research on the use of UAVs is limited in arid ecosystems. Therefore, more research and experiments are needed in modeling the ecological process in arid ecosystems in order to provide better information for land management decisions.

1. Aronson J, Alexander S (2013) Ecosystem restoration is now a global priority: time to roll up our sleeves. Restoration Ecology 21(3): 293-296.
2. Reynolds JF, Smith DM, Lambin EF, Turner BL, Mortimore M, et al. (2007) Global desertification: building a science for dryland development. Science 316(5826): 847-851.
3. Perrow MR, Davy AJ (2002) Handbook of ecological restoration: Restoration in practice. Cambridge University Press, England, Vol. 2.
4. Costantini EA (2016) Soil indicators to assess the effectiveness of restoration strategies in dryland ecosystems. Solid Earth 7(2): 397.
5. Grau J, Anton JM, Tarquis AM, Colombo F, de los R´ıos L, et al. (2010) An application of mathematical models to select the optimal alternative for an integral plan to desertification and erosion control (Chaco Area-Salta Province-Argentina). Biogeosciences 7(11): 3421-3433.
6. Groom G, Mücher CA, Margareta I, Thomas W (2006) Remote sensing in landscape ecology: experiences and perspectives in a European context. Landscape Ecology 21(3): 391-408.
7. Harris A, Carr AS, Dash J (2014) Remote sensing of vegetation cover dynamics and resilience across southern Africa. International Journal of Applied Earth Observation & Geoinformation 28: 131-139.
8. Klemas V (2013) Using remote sensing to select and monitor wetland restoration sites: An overview. Journal of Coastal Research 29(4): 958- 970.
9. Hadeel A, Jabbar MT, Chen X (2010) Environmental change monitoring in the arid and semi-arid regions: a case study Al-Basrah Province, Iraq. Environmental Monitoring & Assessment 167(1-4): 371-385.
10. Herold M, Scepan J, Clarke KC (2002) The use of remote sensing and landscape metrics to describe structures and changes in urban land uses. Environment and Planning A: Economy and Space 34(8): 1443- 1458.
11. Muthumanickam D, Kannan P, Kumaraperumal R, Natarajan S, Sivasamy R, et al. (2011) Drought assessment and monitoring through remote sensing and GIS in western tracts of Tamil Nadu, India. International Journal of Remote Sensing 32(18): 5157-5176.
12. Jungho IM, John RJ, Ryan RJ, John G, Jody W, et al. (2012) Vegetation cover analysis of hazardous waste sites in Utah and Arizona using hyperspectral remote sensing. Remote Sensing 4(2): 327-353.
13. Washington-Allen R, Ramsey R, West N (2004) Spatiotemporal mapping of the dry season vegetation response of sagebrush steppe. Community Ecology 5(1): 69-79.
14. Washington-Allen RA, Neil EW, Douglas RR, Rebecca AE (2006) A protocol for retrospective remote sensing-based ecological monitoring of rangelands. Rangeland Ecology & Management 59(1): 19-29.
15. Chen W, Tetsuro S, Kazuyuki M, Lina K, Chunxiang C (2013) Estimation of vegetation coverage in semi-arid sandy land based on multivariate statistical modeling using remote sensing data. Environmental Modeling & Assessment 18(5): 547-558.
16. Abdullah M, Feagin R, Musawi L (2017) The use of spatial empirical models to estimate soil erosion in arid ecosystems. Environmental monitoring & assessment 189(2): 78.
17. Abdullah M, Feagin RA, Musawi L, Whisenant S, Popescu S (2016) The use of remote sensing to develop a site history for restoration planning in an arid landscape. Restoration Ecology 24(1): 91-99.
18. Abdulla MM, Addae LW, Tena‐Gonzalez GA (2017) Assessing native desert vegetation recovery in a war‐affected area using multispectral and hyperspectral imagery: a case study of the Sabah Al‐Ahmad Nature Reserve, Kuwait. Restoration Ecology 25(6): 982-993.
19. Abdullah MM, Rusty AF, Mansour TA, Al-Musawi L (2017) Will autogenic succession be sufficient to recover from vegetation cover loss or will soil condition need to be addressed in the arid lands of Kuwait? Arabian Journal of Geosciences 10(5): 111.
20. Laliberte AS, Fredrickson EL, Rango A (2007) Combining decision trees with hierarchical object-oriented image analysis for mapping arid rangelands. Photogrammetric engineering & Remote sensing 73(2): 197-207.
21. Gallacher D (2013) Ecological monitoring of arid rangelands using micro-UAVs (drones). in Final Conference Proceedings.
22. Cui X (2013) Using remote sensing to quantify vegetation change and ecological resilience in a semi-arid system. Land 2(2): 108-130.
23. Okin GS, Roberts DA (2004) Remote sensing in arid regions: challenges and opportunities.
24. Müllerová J, Brůna J, Bartaloš T, Dvořák P, Vítková M, et al. (2017) Timing is important: unmanned aircraft vs. satellite Imagery in plant invasion monitoring. Frontiers in plant science 8: 887.
25. Govender M, Chetty K, Bulcock H (2007) A review of hyperspectral remote sensing and its application in vegetation and water resource studies. Water SA 33(2): 145-151.
26. Toutin T (2004) Geometric processing of remote sensing images: models, algorithms and methods. International journal of remote sensing 25(10): 1893-1924.
27. Anderson K, Gaston KJ (2013) Lightweight unmanned aerial vehicles will revolutionize spatial ecology. Frontiers in Ecology and the Environment 11(3): 138-146.
28. Bryson M, Alistair R, Calvin H, Fabio TR, Salah S (2014) Cost-effective mapping using unmanned aerial vehicles in ecology monitoring applications. Experimental Robotics, pp. 509-523.
29. Chen X, Vierlinga L, Rowell E, Thomas DF (2004) Using lidar and effective LAI data to evaluate IKONOS and Landsat 7 ETM+ vegetation cover estimates in a ponderosa pine forest. Remote Sensing of Environment 91(1): 14-26.
30. Peña JM, Torres-Sánchez J, Serrano-Pérez A, de Castro A, López-Granados F (2015) Quantifying efficacy and limits of unmanned aerial vehicle (UAV) technology for weed seedling detection as affected by sensor resolution. Sensors 15(3): 5609-5626.