Preliminary Analysis of Land Cover in the Sinharaja Adiviya using Planet Dove Imagery

Figure 1: Sinharaja Adiviya (or greater Sinharaja area) mapped with Planet Dove imagery. Reduced in size to fit this post.
In March this year I attended a fascinating talk entitled “Sinharaja: From a Timber Reserve to a Biological Treasure Trove. What next?” by Nimal and Savitri Gunatillike at the BMICH in Colombo. The lecture was sponsored by the WNPS in their monthly lecture series. There were several aspects of the talk reflecting back on their decades of research in Sri Lanka’s preeminent forest. Initially, Savitri did botanical studies documenting the diversity of plants in Sinharaja in the period before mechanical logging started (1960s-77). They were witness to period of commercial logging and recovery to a world renowned UNESCO-designated world heritage site. The Gunatilleke’s experiments with rainforest restoration were of particular interest to me, given the lessons that these examples hold for similar non-native plantation areas across the Western Ghats/Sri Lankan biodiversity hotspot. In the lecture, both spoke of the broader Sinharaja area of forest fragments and large patches that are connected or satellites to the core area-something they identified as the Sinharaja Adiviya.
At the same time, I was interested in mapping land cover and forest types in study areas that I take students to for field work. Up to this stage, our DP Geography studies in Sinharaja have utilized Survey Department 1:50,000 and 1:10,000 land use data. It comes as a shape file with the data that I have purchased from their map sales office. This data is satisfactory but we have found significant omissions and inaccuracies in the Kudawa area where OSC students conduct field work (much of the data is based on surveys conducted in the early 1980s).
Forest types and land cover are a key part of the Sinharaja story. Literature about the area’s successful conservation refer to primary and secondary forest as well as Pinus caribaea plantations (along the border). Yet, I couldn’t locate GIS-ready shapefiles of boundaries of these forest types! The Forest Department has files based on its 2010 forest cover map but these are, thus far, not in the public sphere. I had mapped the area using a Landsat tile from 2005 (published in my blog in 2012) but this was before I had learnt how to conduct a supervised classification of a raster image.

Pinus caribaea plantation in the Sinharaja buffer zone undergoing ecological succession as part of an ecological restoration effort. This area was once dominated by a monoculture community. The intervention of conservationists in thinning pine trees and planting appropriate native species is helping to return it to the climax lowland rainforest community. See linked articles by Professors Mark Ashton, Nimal Gunatillike and others for details of these efforts. (September 2019)

Primary/ridge forest below Moulawella Peak in the Sinharaja core zone. This area did not experience any logging in the period of commercial exploitation in the 1960s-70s. (September 2019)
A Brief Literature Review
A review of land cover analysis in Sinharaja shows that only a few studies have been published to date. The most significant, publicly available study looking at land cover change in Sinharaja was conducted by Buddhika Madurapperuma and Janak Kuruppuarchchi in 2014 (see link). Their analysis used Landsat ETM data between 1993 and 2005 in an area slightly larger than the Sinharaja boundary provided by the Forest Department. They used a Normalized Difference Vegetation Index (NDVI) and Burn Index (BI) to assess changes in three years (2001, 1993 and 2005). The study is thorough but the data that they use is coarse and it is difficult to get a sense of the land cover patterns at a large scale. They conclude with an acknowledgment that ground surveys need to be conducted to better understand the change. Thanura Madusanka Silva published a study entitled “Land Cover Changes of a Tropical Forest Buffer Zone” in 2018 that used Sri Lanka Survey Department data to assess changes in land use in the Kudawa area (see link). This study is based on secondary data and not satellite imagery and it concludes that major change has occurred in home garden areas. There may be other studies that I have missed but the field of land cover change in Sinharaja, as seen in satellite imagery, is ripe for further study.

Figure 2: Supervised classification of land cover based on Planet Dove imagery. Because the images were not collected on the same flight, there are some unavoidable gaps and seams, that are visible on close inspection.
Planet Dove Methodology and Results
Two years ago I became familiar with Planet Dove imagery and saw that it might provide a solution in my attempts to classify land cover in the Sinharaja area. Planet Dove’s constellation of 120+ satellites, which revisit the same areas every day, offers a new opportunity to visualize and analyze any area of the earth. At the beginning of 2019 I successfully applied to Planet’s Education and Research program and was able to download a host of tiles of study areas. I found a series of cloud-free scenes from December 2018 and downloaded them. These are from around December 18 but some were collected at slightly different dates. Using ArcGIS, I mosaiced these various tiles so that I had most of the area Sinharaja Adiviya covered. Because the images were not collected on the same flight there is some unavoidable gaps and visible seams, that are visible on close inspection. The improved spatial resolution of 3-5m means that it is easier to distinguish between different land cover types (lowland rainforest, vs, Pinus caribaea plantation, for example). Initially I worked on a map using the non-visible near infrared (NIR 780-860 nm) layer to highlight vegetation (see figure 1).
In the second part of my efforts I conducted a supervised classification using tools in ArcGIS’s Spatial Analyst extension toolbar. For land cover type, I collected between 5 and 10 training samples and merged each of them into their own distinct land cover type. The classified image (Figure 2) clearly highlights the dense lowland rainforest pockets in a landscape dominated by home garden and tea agriculture. The effort to categorize the Pinus caribaea plantation was partially successful. However, there are errors with some of the classification. For example, plantation in the midst of dense (primary) forest near Moulawella peak.
Conclusion/Future
In the next attempt I plan to collect more training samples in the hopes of getting a more accurate picture of the land cover patterns. A focused study on the pine forest in the buffer area near Kudawa deserves attention. Some of these areas are being successfully restored to their original lowland rainforest vegetation type and time a change study would be illuminating. There are areas of the landscape that I am very familiar with (the Kudawa tourist and village zone) while I have far less personal experience in other areas like western Sinharaja and the various forest fragments. Further studies of landcover need to be verified with ground truthing in the field.

Figure 3: The Kudawa area of Sinharaja with classification of land cover based on Planet Dove imagery. In this image I have highlighted the popular tourist area around the settlement of Kudawa. The lower part of the map experienced mechanical logging 40+ years ago (some of it is encircled in red). The stream running down to the Sinharaja ticket gate is not depicted on this map and is missing from Survey Department 1:10,000 sheets that were used for the hydro/stream layer.
SELECTED REFERENCES (Land Cover Focus)
Ashton, Mark et al. “Restoration pathways for rain forest in southwest Sri Lanka: A review of concepts and models.” Forest Ecology and Management 154(3):409-430. December 2001. Web.
Ashton, Mark et al. “Restoration of rain forest beneath pine plantations: A relay floristic model with special application to tropical South Asia.” Forest Ecology and Management 329:351–359. October 2014. Web.
Gunatilleke, Nimal, C.V.S. Gunatilleke and M.A.A. Dilhan. “Plant Biogeography and Conservation of the South Western Hill Forests of Sri Lanka.” The Raffles Bulletin of Zoology, 2005. No. 12 9-22. Web.
Gunatilleke, C.V.S, et al. Ecology of Sinharaja Rain Forest and the Forest Dynamics Plot in Sri Lanka’s Natural World Heritage Site. Colombo: WHT Publications, 2004. Print.
Lockwood, Ian. “Into the Wet: Field Notes From Sri Lanka’s Wet Zone.” Sanctuary Asia. August/September 2007. 3-11. Print. PDF.
Madurapperuma Buddhika and Kuruppuarachchi Janaka.“Detecting Land-cover Change using Mappable Vegetation Related Indices: A Case Study from the Sinharaja Man and the Biosphere Reserve.” Journal of Tropical Forestry and Environment Vol.4, No 01 (2014) 50-58. May 2014. Web.
Madusanka, Thanura. “Land Cover Changes of Tropical Forest Buffer zone A case study of Kudawa Village, Sinharaja forest buffer zone; Sri Lanka.” International Journal of Scientific and Research Publications. October 2018. Web.
Ministry of Mahaweli Development and Environment. Proceedings of the Stakeholder Workshop on Landscape Planning & Management. 29 September 2017. Web. See page 10 for Sinharaja map.
Planet Team. Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. 2017. Web.
Sri Lanka Survey Department. District Land Use Maps. 1983. Print/Web.
UN-REDD Programme. Sri Lanka’s Forest Reference Level submission to the UNFCCC. January 2017. Web.
Wijesooriya W. A. D. A. and, C. V. S. Gunatilleke. “Buffer Zone of the Sinharaja Biosphere Reserve in Sri Lanka and Its Management Strategies.” Journal of the National Science Foundation of Sri Lanka. 31(1–2), 57. June 2003. Web.
Written by ianlockwood
2019-09-16 at 7:41 pm
Posted in Around Sri Lanka, Ecological Restoration, GIS related, Sinharaja
Tagged with Adiviya, ecological restoration, land cover, map, Sinharaja
Leave a Reply