The impact of (DEM) accuracy on the watersheds areas as a function of spatial data

© The Author 2021. This work is licensed under a Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/) that allows others to share and adapt the material for any purpose (even commercially), in any medium with an acknowledgement of the work's authorship and initial publication in this journal. 1118 The impact of (DEM) accuracy on the watersheds areas as a function of spatial data


Introduction
A watershed is a landform of a basin defined by ridgelines and highpoints that slope into stream valleys and lower elevations. A watershed, in other terms, is a geographical region that includes a common collection of rivers and streams that all flow into a single bigger body of water, such as a lake, a larger river or an ocean. Watershed characteristics such as size, shape, slope, land use/land cover, and vegetation are significant variables that influence different aspects of runoff [1].The quality (accuracy and resolution) of primary input GIS data which is required for delineating watersheds and configuring the hydrologic simulation model, such as the digital elevation model (DEM) and the land cover/land use (LC/LU) map, may affect the simulation results [2].A digital elevation model (DEM) is a three-dimensional representation of the surface of a terrain generated from topography elevation data, it is a digital representation of the elevation of the land surface. with respect to any given reference datum [3] . The selection of (DEM) resolution is influenced by some factors, such as cost, accessibility, simulation time with reason, and user need, among others [4]. The output when using all the DEMs in watershed delineation will be considered with regard to several topographic characteristics (boundary, elevations, area, perimeter and slope areas).

Problem statement
Many studies use satellite data (DEM) with certain spatial resolutions and accuracy to define the area of watershed and simulate it. Most of the hydrological models employ input data such as topographic, where these data become available as products of many satellites in different resolutions, accuracies, and sources. There for, it is important to understand the effect of using satellite data from different sources and accuracies on watersheds with various topographic characteristics. The variety of the source, accuracy, and spatial resolution for the satellites data raises several questions: What are the effects of using the highest available online freely open source (DEM) original resolution with different accuracies and sources on the general topographic characteristics (boundary, elevations, area, perimeter and slope areas) of different watersheds in the study area? What is the suitable available online freely open source satellite data (DEM) in the study area?

Objective of the research
The main goals of this thesis are: Analyze the effects of using different source inputs of free Digital Elevation Model (DEM) for the highest original resolution to delineate and compute topographic characteristics (boundary, elevations, area, perimeter and slope areas) with different sizes and topography watersheds. Define the best free satellite data (DEM) for selected watersheds. Find the topographic characteristics (boundary, elevations, area, perimeter and slope areas) of Watersheds in the selected study area to be prepared for the water resources Specialists for calculating the water runoff discharge with these areas by the unit hydrograph procedure.

Study area description
The study area is located in the northeastern part of Iraq, in the Mosul reservoir about 60 Km north of Mosul City between latitude (36°38'0" to 37°5'0")north and longitude (42°5'0" to 42°52'0")east, Nineveh, Iraq, which is represented by five watersheds of varied characteristics on both sides of the Mosul reservoir (Swaidy, Garlond, Khuwayr Hirah) valleys on the west side, and (Naqab, Kalak) valleys on the east side, these watersheds are different in size, topography, and LC/LU as shown in  The water in the reservoir comes from the Tigris River and ten side reservoir valleys, seven from the east side, and three from the west side of the reservoir [11]. The topography of the study area on the West side of Mosul reservoir is ranged from a flat arable area with a low slope rate for a Swaidy valley that reaches Qarachock Mountain in Syria. The Garlond valley is ranged from flat to mountainous, as for the Khuwayr Hirah Valley which is located within a mountainous area to reach Jamrok Mountain, a large proportion of unfit lands for agriculture. On the East side of the reservoir which is part of Duhok Governorate, there is a wide variation in elevation, the valleys of Naqab and Kalak begin with undulating, rugged lands near the lake, then gradually turn into arable lands and ending with mountainous lands in Bekher Mountain. Geologically, the Pilaspi Formation is the oldest exposed formation in the Mosul reservoir's neighborhood, this formation's exposures are limited to hilly terrain. Dolostone, limestone, marl, and marly limestone make up the content. Fatha Formation and Injana Formation (both Lower-Upper Miocene) are exposed in the plains [12].The Land cover Seasonal crops (wheat and barley), vegetables, and pastures cover a large portion of the study area [13].
Climate of the study area shows significant seasonal variations in temperature and precipitation, which are characterized by dry in summer months and wet in winter months, the climate of the study area is close to that of the Mediterranean, with certain variations owing to the nature of the mountainous zone in Turkey. Summer seasons are hot and dry, while winter seasons are cold and rainy, with snowfall occurring on occasion in the mountainous areas. The rainy season begins in October month and continues until May month [14].

1
Material and methods

DEM sources and processing
The delineation and topographic characteristics of selected watersheds in study area are compared using three free available online different sources, and accuracy input DEMs with same spatial resolutions as shown in Table 1.

Methodology
The overall research can be divided into several phases, as seen in Fig. 2. Figure 2. Flowchart of the research.

Watersheds delineation
The watersheds delineation was performed using (Copernicus DEM (30) m, SRTM DEM (30) m, and GDEM (30) m) data for various steps which were needed to transform the elevation data into flow direction and accumulation data, and finally into Watersheds. The specific tools needed for the watersheds delineation are found in the (ArcGIS Desktop 10.8) Hydrology tools within the Spatial Analyst toolbox, finally, the watersheds are delineated using the watershed tool in (ArcGIS Desktop 10.8) software, this operation takes into account the pour points, flow direction, and flow accumulation as shown in Fig. 3 to Fig. 5.

Optimum DEM for the selected study area
For checking and selecting the best accurate DEM for the study area, (80) accurate elevation points were used, ((20) control points from Iraqi ministry of water resources (MOWR) fieldwork which gathered in (September/2020) at Mosul dam using (Topcon GPT-7501 total station, and Topcon Gr5 Gnss Receiver), and (60) point extracted from LIDAR DEM (1m horizontal resolution, 0.2m vertical accuracy with heights from ellipsoid WGS48) covering parts of the study area). Selecting the best accurate DEM for the study area is done by comparing the elevations of 50 checkpoints with elevations interpolated from DEMs used in watersheds delineation (Copernicus DEM, SRTM, and Aster GDEM) at the same E/N coordinates after unified all heights to (EGM96). Relying on the elevation differences between checkpoints and used DEMs, the RMSE (root mean square error) and RMSE at 95% confidence level are calculated for differences as shown in Table 2. Depending on the analysis of the results, the Copernicus DEM have less RMSE and RMSE at 95% confidence level which means more vertical accuracy than the rest of DEMs in the study area (Landsat SRTM and Aster GDEM) respectively. Therefore the Copernicus DEM is optimum DEM for Watershed delineation and topographic characteristics Calculations for the selected study area.

Watersheds topographic characteristics analysis
Topography is an important land-surface characteristic that affects most aspects of the water balance in the watersheds [6]. The effect of different accuracies of used DEMs on different Watersheds Topographic characteristics such as (watersheds boundary, elevations, area, perimeter and slope areas) in the study area is analyzed.

Watersheds Boundaries
By comparing each delineated watershed from Copernicus DEM with the same watershed produced from other DEMs (SRTM and ASTER) through Intersect tool in (ArcGIS 10.8) software, and the Intersected (Coincided) areas are shown in Table 3. and Fig. 6.

Watershed elevations comparison
The Minimum, Maximum, Mean, and Standard Deviation Elevations (St. Dev.) of each delineated watershed based on different used DEMs is calculated by generation of a report of statistics using (ArcGIS) software as shown in Table 4.  Table 5.  (4) while (2) in (Copernicus -ASTER), it is noticeable that the elevations accuracy of DEM affected by the vertical accuracy of used DEM.

Watersheds area and perimeter calculation
The area and perimeter of each delineated Watershed based on different used DEMs is calculated using ArcGIS software as shown in Table 6. From this table, except the Khuwayrhirah watershed, it is noticeable that the areas are directly proportional to DEM accuracy in the relatively flat watersheds (Swaidy and Garlond) and inversely proportional to DEM accuracy in the wide variation elevations watersheds (Naqab and Kalak), while the perimeters are inversely proportional to DEM accuracy in all watersheds except for Khuwayrhirah watershed , by comparing the watersheds areas and perimeters based on optimum DEM (Copernicus GLO-30) to the areas and perimeters based on SRTM and ASTER DEMs, the differences are calculated as shown in Table 7. From this Table, it seemed that the major accuracy is for (Copernicus -SRTM) DEM differences with lowest RMSE and differences of all watersheds areas and perimeters except the Khuwayrhirah watershed because of the difference in watersheds delineation and borders in the southeast part of this watershed. These results can be revealed that the areas and perimeters of watersheds are affected by the vertical accuracy of used DEM.

Watersheds land slope area calculations
For each watershed, the percentage of the land slope area is calculated as percent rise by using slope from spatial analyst tools in (ArcGIS Desktop 10.8) software, by reclassify tool all the results are reclassified to fit the wanted slope percentage. Then the zonal statistics as table tool used to summarize and extract the area with area percentage of every slope interval for each Watershed based on used DEM as shown in Table 8.  Based on the analysis of the results, it can be revealed that the considerable accuracy is for (Copernicus -STRM) DEM differences with the lowest RMSE of watersheds slope areas Percentages differences, which means these results are affected by the vertical accuracy of used DEM. The map of land slope for each watershed based on Copernicus DEM are shown in Fig. 9. Figure 9. Land slope maps for watersheds based on Copernicus DEM.

Conclusions
From the results analyses of the experimental works, GIS techniques, and remote sensing processing works in the present study, the most important conclusions that indicated are given in the following:  The Copernicus (GLO-30) DEM is the most accurate (30) m resolution with No charge DEM for the selected study area. Where, the RMSE results for the elevations differences between check points (MOWR elevations and Lidar DEM) and the used DEMs for the Copernicus (GLO-30) is (1.3521 m) and with 95% confidence level is (2.6502 m). Therefore, it can be considered as the optimum DEM for watershed delineation and topographic characteristics calculations and comes after the Copernicus DEM in accuracy both the SRTM and ASTER DEMs, respectively.  All the selected watersheds topographic characteristics (boundaries, area, perimeter, elevations and land slope area percentages) in the study area have least RMSE (more accuracy) in (Copernicus-SRTM) than in (Copernicus-ASTER GDEM), and all differences are least in (Copernicus-SRTM) for all watersheds except Khuwayrhirah watershed, so the topographic characteristics affected by the vertical accuracy of the used DEMs. Therefore, all the results can be came in proportion to the accuracy of used DEMs (Copernicus, SRTM, and ASTER) in all watersheds, while the reason for the large difference in delineation and other topographic characteristics in Khuwayrhirah watershed which is located within a rough mountainous area, and a large proportion is unfit for agriculture is the variation in the topographic representation between used DEMs in the southeast part of this watershed.