Analysis of changes in rivers planforms using

Analysis of changes in rivers planforms using

google,earth,engine

Jianglong shen1,2 Tingyu zhang1,2

1 Key Laboratory of Degraded and Unused Land Consolidation Engineering, The Ministry of Natural Resources, Xi’an, Shaanxi, China

2 Institute of Land Engineering and Technology, Shaanxi Provincial Land Engineering Construction Group Co., Ltd., Xi’an, Shaanxi, China

Project: Internal scientific research project of Shaanxi Land Engineering Construction Group（DJNY2022-25，DJNY2022-30）

Abstract：We implement three different methods for delimiting rivers based on satellite information: the Normalized Difference Water Index (NDWI), the Modified Normalized Difference Water Index (MNDWI) and the Automated Water Extraction Index (AWEI). We apply these methods to annual mosaics generated for each source of information, and estimate rivers centerlines based on a thinning method. With the centerline and river planform delimitations, we automatically estimate three geomorphological indicators: mean width, connectivity and sinuosity. We apply our tool to analyse several reaches along eight representative Colombian rivers that exhibit different planforms and flow regimes, some of them correlated with the Southern Oscillation Index (SOI).

Key Words: Google earth, Engine, Planforms

1 Introduction

Solving many of the current needs associated with remote sensing information processing, Google implemented the Google Earth Engine (GEE) [1]. GEE facilitates access to high performance computational resources for processing satellite images, in addition to allowing the use of a large number of up to date remote sensing databases for scientific and academic purposes. GEE also facilitates the sharing of developed codes and knowledge among users, mostly in fields of geosciences [2]. In our study, we use two of such databases: Landsat and Sentinel. Landsat began its mission in 1972, and continues to release updated images with the most recent Landsat-8 mission.

2. Materials and methods

Although GEE incorporates Landsat information from 1972 to the present, only the annual information available since the mid-1990s covered the entire planet in a consistent fashion

[3]. For the purposes of our analysis, we use only images after 2000 (Landsat missions 7 and 8), to guarantee images for each year with extended coverage in any region. We also use the Sentinel-2 database with information from 2014 to the present [4]. With the satellite databases, we define a polygon of the area of interest and a period of analysis. We select the images for the defined polygon, clustered in annual sets, to generate annual mosaics. Therefore, the mosaics highlight only those changes in river planform that are significant on a yearly basis, beyond expected seasonal variations and regardless of the region of analysis.

2.1 Satellite databases

We use Landsat images from 2000 to 2018 and Sentinel-2 images from 2015 to 2018. From 2000 to 2012, we use Landsat-7 images. From 2013 on, we use Landsat-8 images. Both Landsat-7 and Landsat-8 have a granularity of 16 days, and 30 m of spatial resolution. Sentinel images have a granularity of 5 days and a spatial resolution between 10 and 20 m, depending on the band. Bands with additional information may have a lower resolution, as will be shown later when addressing the issue of cloudiness correction.

2.2 Definition of a river centerline

The delimitation of different river planforms over a territory can be a sufficient input to analyse temporal variations of each reach, associated with channel pagation processes, sediment bars formation, and width changes. The delimitation of a river form, for instance, allows the estimation of different geomorphological parameters, based on attributes such as the river centerline. A river centerline runs through the middle part of the river from each bank and is useful to describe planform changes in time.

3. Results and discussion

3.1 Annual mosaics and cloud correction

In general, uncorrected annual mosaics are highly spotted with clouds, which in some cases do not allow the distinction of land surfaces. On one hand, Landsat images contain a large presence of clouds, but the correction filters them greatly, obtaining cleaner images, from which we can appreciate the shape of the river easily. On the other hand, Sentinel generates annual mosaics less affected by clouds that improve even further with the correction.

3.2 Centerline delimitation

The method of Zhang and Suen (1984) offered very good results to delimit the centerlines. The automatic delimitation of the centerline in straight and sinuous reaches, for instance, produce optimum adjustments to the shape of the river. Centerline adjust well even in systems with geomorphological elements such as bars. As mentioned above, the delimitation has deficiencies over highly braided sections. Since the centerline depends exclusively on the river shape, the results in braided sections are not the best. However, the method of Zhang and Suen provides a valid approach to analyse the variations in the branches that manage to delimit.

4 Conclusion

The correction made on the delimitations of surface water using the GSW dataset allowed refining in very good detail the planforms. Regarding the delimitation of the centerline, the method of Zhang and Suen presented very good results in general. In braided reaches, the delimitation of the centerline presented deficiencies due to errors inherent to the delimitation of surface water. The planform delimitation of braided reaches was a major challenge, mainly because large parts of the branches that conform the braided sections are difficult to delimit within the spatial scale used (30 m). The calculation of the mean width, connectivity and sinuosity do not seek to provide an exact value to the user beyond the resolution of the images and the scale used, since values measured in the field may differ considerably.

Reference:

[1] Restrepo, J. D., and H. A. Escobar. 2018. “Sediment Load Trends in the Magdalena River Basin (1980–2010): Anthropogenic and Climate-induced Causes.” Geomorphology. 302: 76–91. doi:10.1016/j.geomorph.2016.12.013.

[2] Suarez, J. A., and M. A. Sandoval. 2017. Geomorfología y resistencia a la erosión fluvial de los suelos conformantes del cauce del Río Magdalena (Neiva-Prado). Bogotá: Universidad de la Salle. Faculta de Ingeniería.

[3]Li, C., S. Wang, X. Bai, Q. Tan, Y. Yang, Q. Li, L. Wu, J. Xiao, Q. Qian, F. Chen, and H. Li. 2019a. “New Automated Method for Extracting River Information Using Optimized Spectral Threshold Water Index.” Arabian Journal of Geosciences. 12 (1): 13. doi:10.1007/s12517-018-4124-z.

[4] Hansen, M. C., P. V. Potapov, R. Moore, M. Hancher, S. A. Turubanova, and A. Tyukavina. 2013. “Highresolution Global Maps of 21st-century Forest Cover Change.” Science. 342: 850–853. doi:10.1126/science.1244693.