A Multi-Layer Based Assessment of Wetland Changes in the Southern Iraqi Marshlands Using Remotely Sensed Data

Marshlands are important ecosystem for living organisms. The Southern Iraqi Marshland is the central habitat for freshwater fish, provides habitat for important populations of wildlife and serves as a source of income for native economies through reed harvesting. Studies have shown that variability in climate and human-induced factors affects the spatial dynamics of marsh ecosystems. This study assessed wetland changes in the Southern Iraqi Marshlands using Remotely Sensed Data from 1986 to 2019 using Landsat satellite imagery for four epochs: 1986, 2000, 2010 and 2019. To achieve this, thirty (30) pixels were obtained in selected land cover theme and their signatures were merged into one class. Furthermore, the selected pixels were re-coded and merged into ten (10) land cover classes. The multi-layer classes created were shallow water, deep water, dense marsh, medium marsh, sparse marsh, dense vegetation, medium vegetation, sparse vegetation, dry soil and wet soil. The areal extents of the land cover types were calculated for 1986, 2000, 2010 and 2019. Normalized Difference Vegetation Index (NDVI) differencing was also carried in order to highlight trends in vegetation from 1986 to 2019. The study correlated historical trends of human activities as a central factor in the degradation of marshland (by 16.25%) from 1986 to 2000. However, by the year 2000 to 2010, there was an 11.36% increase in the total marshland area, which remained almost unchanged between 2010 and 2019. In 1986, NDVI was relatively stable at 0.73 in the Al-Hammar and Al-Hwaizeh Marsh. However, by 2000, the areas of dense vegetation cover reduced drastically by over 90%. In 2010, the NDVI index indicated trends of increasing water body and an outward cluster of healthier vegetation continuing to 2019.


Introduction
Marshlands of the world are seen as an important ecosystem for living organisms and the environment, serving as a source of income for native economies through reed harvesting and fishing [1]. The Iraqi Marshland is considered to be one of the most important marshlands in the world: the largest ecosystem in Eastern and Western Asia. It is the eleventh non-marine marshland with Endemic Bird area status [2] and central habitat for Gulf's freshwater fish [3].
The marsh land is home to 500,000 native Marsh Arabs [1]; and of the world's oldest western civilization point. In 1985, levees were built and part of the marshes was drained by the Iraq government to develop oil fields discovered within the marshlands [4].
The Iraqi marshlands-originally twice the size of the Everglades in Florida-has been the major wetland area in the Middle East, as of 1990, very little had changed in the region. However, within a decade, 90 percent of the ecosystem area was turned into barren dry lands by a large drainage scheme. Additional important roles provided by marshes include detention of floodwaters, protection of shorelines from erosion, aesthetics, and improvement of water quality by trapping sediments and assimilating nutrients.
In recent times, remotely sensed imageries have been used to monitor and map wetlands changes on a global scale [5]; especially in areas with inadequate ground-based observations [6]. Due to the vast size and remoteness of the Iraqi marshlands, remote sensing is needed to monitor the dynamics and levels of change of its sensitive ecosystem.

Study Area
The southern Mesopotamian marshes lie in the southern part of Iraq, cutting across three of Iraqi's Provinces. It is along latitudes 29˚55' to 32˚55'N and longitudes 45˚25' to 48˚30'E (See Figure 1

Data Analysis
Landsat imageries from https://earthexplorer.usgs.gov were acquired for four epochs (1986,2000,2010 and 2019), layer stacked and mosaic. Radiometric and geometric errors were reduced by colour correction and feathering respectively. The Landsat imageries were first corrected for geometric (through georeferencing), and radiometric (through filtering) errors to correct for locational misplacement based on [7]. All the data sets were collected between January and May each year (Table 1).  The supervised classification technique using Maximum Likelihood Algorithm (MLC) was used. This allowed for a multivariate normal distribution to be used to represent an individual spectral class [8]. Training data were collected using the polygon method. Thirty (30) pixels were obtained in each land cover theme and the selected signatures of the same pixel were merged into one class. The 30 multi-layer classes were re-coded and merged into ten (10)

Results
The land covers distribution and changes that took place on the marshland are shown in Table 2.  Table 2 and as illustrated in Figure 2, the total area of marshland in 1986 was 7008.89 km 2 (Dense Marsh 1716.55 km 2 , Mdedium Marsh 4258.39 km 2 and Sparse Marsh 1033.95 km 2 ).
By the year 2000 (Figure 3), medium marshland areas reduced by about 9.56% (Table 3) for a decreased total marshland area of 1458.49 km 2 (Dense + Medium + Sparse) Shallow and deep-water bodies reduced by 1.3% and 0.77% from 1986 to 2000.   Hence, most of the marshes degraded to sparse vegetation and increased dry soil area. This disappearance can be attributed to anthropogenic activities [9], as well as unforeseen conditions, including droughts and high temperatures [10].
In the year 2010 (Figure 4), there was a 1.52% increase in dense marshland for a total increased marshland area of 5338.83 km 2 (Dense + Medium + Sparse).
This corresponds to about 11.35% increase in the total marshland area from the

Trends in NDVI
In the year 1986 NDVI was relatively stable at 0.73 in the Al-Hammar marsh.   By the year 2010, the NDVI index varied from −1 to 1; indicative of increasing water body areas and an outward cluster growth of healthier vegetation (Al-Hammar, central and Al-Hwaizeh marsh areas). This is evidenced in the 2000 to 2010 NDVI change detection in Figure 7 and extends into the 2010 to 2019 NDVI change detection in Figure 8. This vegetative cover improvement is a result of the collective effort by Iraq's government and international agencies to support the restoration of the marshes [12].

Conclusion
This study has demonstrated the applicability of remote sensing technique in monitoring changes on sensitive ecosystems. It also highlights how historical footprints of human activities affected the land cover and NDVI trends in the study area. This study, therefore, suggests that efforts to restore the southern marshlands of Iraq should be sustained and further studies are required to ascertain the levels of variability in climate and its effect on the dynamics of the Iraq marshes.