Geospatial technology on road, terrace and planting point for oil palm cultivation in hilly terrain

One of marginal area for oil palm cultivation is on hilly terrain, which pose various agro-management problems start from planting preparation, during oil palm growth/maintenance and up to the harvesting and transporting of the fresh fruit bunches. Geospatial technology was used to help designing road system with focuses on 1) road gradient of less than 7°, 2) road density of less than 200 m/ha and 3) carry distance of about 100m along the terraces. Commercial trial in Pelalawan Riau showed that the design can be implemented resulting average road gradient of 4.7°, road density of 234 m/ha and average carry distance of 143m. Geospatial technology was also used to help designing terraces where digital terrain model was used to produce suitable rajah line location. On planting point, improved Gawthorn and Violle method was introduced incorporating geospatial analysis in detection of potential etiolation areas. The method produced stand per ha of 142 palm/ha (and 145 in commercial) with minimized potential etiolation area to 6%.


Introduction
One of marginal area for oil palm cultivation is on hilly terrain, which pose various agro-management problems.In most cases, those agro-management problems result in lower yield in comparison to flat area with typical straight-line planting.The main agro-management problems in such area that can affect oil palm growth and yield are [1]: • high risks on erosion, landslides and run-off losses of nutrients, • poor water balance due to excessive run-off, • the need to terrace implies the planting on less fertile sub-soil, which is commonly devoid of organic matter and generally firmer consistence, • difficulty in harvesting and field maintenance operations with probably results in poorer crop recovery.Fresh fruit bunch (FFB) yield from commercial field in an estate in Riau Indonesia can be used as example on how hilly terrain (without terraces) affects the yield.In average for the last 18 years, we identified 5.7 t/ha lower FFB yield in hilly terrain areas in comparison to that in flat areas and the yield gap gets larger over time (figure 1).Various factors are attributed to this lower yield and amongst of them are described below.

Land clearing and planting preparation
The steeper the terrain, it is more difficult and need more careful planning during early preparation for oil palm cultivation.Steep land development may result in environmental degradation including surface run-off and erosion.Delay in planting may also aggravate erosion problems.Moreover, the establishment cost on steep land is 45% higher than that on normal terrain which mainly due to terracing/land preparation and road construction [2].Complex terrain with variable slope is commonly found and smart decision is required to categorize the areas namely, area to be planted in straight lining pattern, platforming, areas with conservation terraces, area with actual terraces or even areas to set aside due to extreme slope limitation.Areas with slope of above 40% is difficult to be cultivated and with its extent of more than 25 ha, it should be conserved under RSPO requirement.

Oil palm growth and maintenance
Higher FFB yields and vegetative growth parameters are commonly resulted by oil palm on lowlands as compared to upland area which has less water and nutrient retention capacity [3].High risk of erosion and runoff also causing degraded soil as well as its fertility.Soil and water conservation play an important role on oil palm performance in hilly area by maintaining and improving the soil's physical and chemical properties.
Sloping soils need platforming, terracing with back slope, barriers and ridges mainly for conservation measures.Terracing generally improves soil moisture by reducing surface runoff [3] and increasing soil water capacity [4].Diminished runoff also allows better soil erosion control and the accumulation of organic layers [5].Consequently, nutrient losses due to erosion and runoff are significantly reduced in terraced slopes [6].Other conservation measures such empty fruit bunch (EFB) mulching and pruned fronds staking were commonly implemented in the industry and it could retain moisture in the top 20 cm of the soil on hilly slope under oil palm [7].The combination of these conservation measures contributes to better plant growth and, subsequently, leads to higher yield.

Harvesting, FFB collection and transport
Common management problem in hilly plantings is mainly due to poor access to palm, including during harvesting operations.This causes delayed harvesting interval, lower harvesting productivity which in turn reduce both quantity of FFB and quality of oil.Steeper roads limit the transport tonnage while FFB collection is often only in platform areas.If hilly area is not designed for mechanization, the increasing demand for labour is inevitable.Inaccessibility of planting points on individual terraces to wheeled vehicles requires increasingly scarce labour be devoted not only to harvesting but also to almost every other phase from field planting to maintenance including manuring, upkeep, etc. [2].

Objectives of the study
Agro-management approach to improve oil palm cultivation in hilly terrain has been studied by many researchers.Common solution suggested includes terracing and planting point lining in which Gawthorn [8] and Violle [9] are amongst common reference used.Road and drainage design is actually another important exercise that need to be focused prior to any operations.Geospatial technology has been developed very rapidly and various analysis can now be carried out including for oil palm hilly planting.This study focuses on application of latest geospatial technologies for road and drainage design, improved terrace design as well as in planting point lining method.

Goals of road design
Accessibility in oil palm estate is one of key important factor in sustainability and profitability of an oil palm plantation.Proper access to every oil palm tree is required to not only bring inputs to the palms ie.fertilizers, upkeeps, treatments, etc. but also harvest FFB and transport it to mill.Road design, particularly in hilly terrace planting, is therefore crucial and it should be properly designed.Some important aspects in road design that need to be addressed carefully are 1) road density, 2) road gradient and 3) carry distance.
Density of road should be optimized so that it is not too much that is expensive but also not too less that is insufficient.Patrick and Goh [10] mentioned that density of road in oil palm plantation should be 50-80 m/ha and 150-200 m/ha for gentle and hilly to steep terrain respectively.Road density in hilly to steep terrain is generally higher due to complex terrain condition that require winding roads following the hills and ridges.Straight line road is not possible in most cases due to gradient limitation.Gentle road gradient will not only improve tonnage limitation for transport vehicle but also reduce road damages due to erosion.Patrick and Goh [10] suggest limit of road gradient at 7% (4.0 o ) for rolling to hilly terrain and 12.5% (7.1 o ) for hilly to steep terrain.
Carry distance is basically the distance where harvester should carry or move FFB to nearby road.Reasonable carry distance for manual operation is about 100 m or a distance of 11-12 palms but it can be more if mechanization is implemented.For the moment, mechanization is commonly implemented only in flat to undulating areas while for terrace areas, due to limitation in width of terraces, manual operation using wheel barrow is generally used.For optimal terrace planting, ideal carry distance is used to determine terrace length.Terrace length refers to road to road distance of a horizontal terrace and ideal terrace length should be set twice of ideal carry distance (or about 200m).This will allow FFB to be transported out from either one of the closest 100m distance from road.

Road design techniques
Prerequisite of this road design technique is 1) accurate digital terrain model (DTM) and 2) geographic information system (GIS) package.Free online terrain data have been evaluated in this study earlier but they are not suggested due to poor vertical accuracy.Ideal terrain model can be obtained by LiDAR data acquisition but better and affordable terrain model is suggested by aerial photo.Data acquisition should be carried out after land clearing (new opening from jungle) or after palm felling (replanting areas).Condition of zero vegetation above ground will give higher accuracy of DTM that is important for this road design exercise.
In this technique, roads are divided into 4 types based on its features and functionalities (figure 2a).
• Hill road.It connects one hill/high ground to another hill/high ground through and along higher ground.• Ravine road.It connects and cover low-lying areas, can be foot hill or middle road adjacent to ravine's drain.• Contour road.It connects hill road and ravine road in controlled gradient and density.It has function as the main access to all terraces.• Main road or highway.It is the main access to the block having the shortest distance in/out of the block and it gives good access coverage of the block.This road is selected from the above three roads (hill, ravine and contour road) that meet the requirements.For replanting field with existing road network, analysis should be carried out to select existing road that meet the requirement of the above 4 types of road.This will allow savings in road construction but yet, still adopt improvements of new road design.
Using GIS package, 1m interval contour line can be generated from normalized DTM.The package can also easily identify high grounds and hills.Hill roads can be digitized by connecting hills or high grounds through and/or along higher ground.Attention should be put in road gradient that is explained in detail for contour road establishment.Similarly, ravine and low-lying areas can also be identified using GIS package.Foot hill road is one of ravine road type that can be digitized based on contour line along the low-lying areas.It can be slightly varying up and down but it should be around 1-2 m contour line higher than the lowest point/line in the DTM.
Contour road is the one that will cover sloping areas and its gradient is the focus of interest.Contour road connects hill road in the upper side of terrain with ravine road in the lower side.The technique starts by identifying areas suitable for junction, either in hill road or ravine road.Ideal junction points should be chosen in relatively gentler area so that vehicle can turn without much difficulties.Digitizing contour road starts from the junction point to upper direction (if point is in ravine road) or lower direction (if the point is in hill road) and connecting it to the first adjacent 1m contour line.In order to get desirable road gradient of 7 o or lower (figure 2b inset), the length of connection should be set to 8m or above (shown in figure 2b as the distance from B1 to B2).It can be to toward any desirable direction (left or right) but it should be consistent throughout the hills.The next step is to continue connecting from the first 1m contour line to the second 1m contour line with the same approach until the it connects to the opposite hill road or ravine road.
In order to get carry distance of about 200m, parallel contour road is introduced in this technique.Once the first contour road is established, point around the middle part of the road or hill should be identified (shown as point A in figure 2b).A horizontal distance of 200m using contour line is to be measured and one should set this point as one of point in next contour road (shown as blue dotted line in figure 2b).Same approach of digitizing the first contour road shall be repeated but it is now started from the new point at the end of 200m distance.Digitizing is to upper direction toward hill road and lower direction toward ravine road and same direction with the first contour road shall be adopted.The upper part generally has shorter distance of contour line (which is subsequently become terrace) while the lower part will have longer distance.It will however have average length of about 200m.

Field application
Digital design should be put and marked in the ground as the guide for heavy machineries to work.Ground pegging using high accuracy equipment such RTK GNSS or total station is ideal but since the road is wide enough ie.ranging from 4 to 6m width, the use of regular GPS equipment will not give significant differences.
Road should be constructed first before the terraces to avoid miss connection between road and terraces.Based on field experience, the use of excavator is required at first to cut the hill and form the shape of road.Subsequently, bulldozer or motor grader should be used to finalize including the angle construction.For contour road, it is suggested with angle of 8-10 o toward the hill.Inner road side drain and discharge points (rorak) is required to ensure water is not stagnated and causing pothole.At last, road need to be gravelled and compacted.

Semi commercial trial
Semi commercial trial was conducted in our principal estate in Pelalawan Riau covering an area of 246 ha.The topography is rolling to hilly with average slope of 11 o and maximum slope can reach 35 o .Total length of road using the approach is 57,522 m and thus, road density is about 234 m/ha.It was slightly more than the ideal range of 150 to 200 m/ha.Road elevation at representative sampling points were measured using Titan TR7 GNSS RTK.Average slope was obtained at 4.7 o that is lower than the maximum threshold of 7 o .For carry distance, we measured length of all terraces and we recorded a range from minimum of 39m to maximum of 255m with average terrace length of 143m.This was lower than the objective of about 200m but it could be the optimum figure as we noted the variance of up to 255m.

Drainage design
The goal of drainage design is to remove excess water safely, reduce erosion and land slide, and also to determine locations and sizes of bridges and culverts.In GIS package, watershed analysis is a powerful tool to produce reasonable size of catchment areas, water flow network as well as flow direction pattern.Using RGB aerial photo, drain network can be digitized by synchronizing water flow data from watershed analysis with actual ground data such as the presence of natural streams, ponds, etc.If gradient allows, the drain should be straightened as much as possible to ensure smooth flow of water.By overlaying drain network and road design, location of culvert or bridges can be identified (figure 2c).Catchment area from upper part of each crossing can be used to determine the size of culvert or bridges.Simple rational method may be used to determine water peak discharge rate and estimate the size of drain as well as the culverts or bridges.
For ravine areas, design of field drain is suggested in the pattern of 60-degree or herring bone design.In such areas, gradient of terrain is generally clear to one direction (around the middle of ravine) and thus, 60-degree and herring bone design will optimize the water flow.The 60-degree pattern also has advantages in term of planting point as the space are utilized better.

Goals of terrace design
Terrace should be designed not only for conservation measures but also to improve accessibility to individual palms as well as to maximize spacing for optimal planting density.As discussed earlier, terracing will reduce erosion, run-off and fertilizer wash.It also improves accessibility to individual palms, whereas explained earlier, proper access is required to give the best inputs to the palms as well as for harvesting operations.In term of palm density, rolling to hilly terrain with complex sloping is normally resulted in difficulties to set up optimum planting point.Density will be generally either lower or higher than the standard in which both are disadvantages to estate.Lower density is obviously a loss as land productivity become lower.Higher density planting causes many areas having etiolation problem due to close planting.Etiolated palms will produce less FFB as it has shading effect from neighbouring palms and has lower portion of solar radiation for photosynthesis activities.

Gawthorn [8] technique
Gawthorn [8] has established a guidance in terrace planting for oil palm.The author stated that the minimum acceptable distance between palms for optimal growth and yield is 26 ft.(or 7.9m and equivalent to density of 184 palm/ha).That palm to palm distance results in palm row distance of 22.5 ft.(or 6.9m) and this is the minimum terrace width that should be set.Any terrain slope that make terrace width of less than 22.5 ft.should be cut off.In the other hand, terrain slope having terrace width of above 45 ft.(or 13.7m and twice of the minimum width) should have additional planting rows or terraces (figure 3a).

Terrace design technique
With accurate DTM and GIS package, area having potential terrace width of between 22.5 ft. and 45 ft.can be identified.At first, steepness of complex hill should be differentiated and in this case is through terrace vertical distances (CT).CT refers to vertical distance from terrace to terrace.The technique suggests 4 categories of CT comprises of 1m (CT1), 2m (CT2), 3m (CT3) and 4m (CT4).Terrace with vertical distance of above 4m will have hill slope of more than 30 o and it is not suitable for oil palm cultivation.Under RSPO requirement, CT3 is in fact also not suitable as it has slope of more than 23 o .For this study however, the limit was decided up to CT4.Terrace horizontal width is set to 5m in view of future requirement to implement mechanization along terraces.Using trigonometry, slope range that have specific CT and horizontal terrace width of between 22.5 and 45 ft.can be calculated.CT1 has slope range of between 6 o to 8 o and normally straight planting is still implemented in such terrain condition or probably conservation terrace is ideal.CT2 has slope range of 9 o to 16 o and terrain is varied from rolling to hilly.Terracing with 2m vertical distance is suitable for such condition.CT3 has slope range of 17 o to 23 o while CT4 has slope range of 24 o to 30 o (figure 3b).Using GIS package, areas having specific slope range ie.6-8, 9-16, 17-23 and 23-30 can be produced and similarly with contour line with interval to the corresponding CT.CT2 should be then overlaid with 2m interval contour line and similarly to CT3 and CT4.Combining all the CT's areas into one layer will allow identification of boundaries of each CT (figure 3c).The boundary areas are the places where operations such as cutting of a terrace or connecting to upper/lower terrace as well as the starting of additional row of planting exist.With additional processes, the above operation can be digitized manually to get optimal results.The digital map produced is the final result and it can be used as field guidance.

Field application
Similar with road design, digital terrace design should be put and marked in the ground as the guide for heavy machineries to work.Ground pegging using high accuracy equipment such RTK GNSS is ideal but we encountered difficulties to get it done.It takes relatively long time to get high accuracy points marked in the ground.Unlike for road design that allow some horizontal errors up to 1-2m, terrace design needs higher accuracy ie.below 50cm to get the terrace line properly constructed.Such condition is not practical for commercial plantings and thus, we keep ground pegging exercise to further study until applications or equipment are available to support the approach.
The map produced is however can be used to generate terrace rajah line.Terrace rajah line is imaginary line across the proposed terraces (from top of hill down to lowest point in ravine) and it should be located at representative sloping degree of a target hill.Terrace points are determined along the rajah line horizontally according to palm row length of target palm density.From each point, terrace line (at horizontal angle) can be determined using waterpass (autolevel/dumpy level).Using the final map, dominant CT area within hill and in each hill can be identified.Rajah line should be drawn in these dominant areas as the representative sloping degree of particular sloping area.The number of rajah line in an area/hill can vary depend on the complexity of the hill (figure 3d).

Goals of planting point design and threshold of canopy overlap
Once terrace is established, the next step of laying out planting point is crucial.The objective of this exercise is to ensure that planting points are optimal to achieve density namely from 138 to 148 palm/ha and at the same time well-spaced to avoid etiolation problems in future.
Etiolation problems can be associated with too much canopy overlap due to close planting.Using GIS package, canopy overlap of optimal density at 138 and 148 palms/ha can be calculated.Using average canopy radius of mature palm at 5m, we noted canopy overlap of 12.69 m 2 (16.27% of canopy size) and 20.89m 2 (26.74%) from palm density of 138 and 148 palm/ha respectively.Using similar approach, we calculated canopy overlap of Gawthorn [8] method in its palm limit distance at 45.69 m 2 (58.84% of canopy size).This value is set as threshold of maximum canopy overlap and larger value will results in close planting that cause etiolation problems.[8] and Violle [9] Gawthorn [8] method used 26 ft.minimal palm distance as base line.On target density of 128, 138 or 148 palms/ha, the distance of planting points along terrace is determined by the horizontal distance of terrace to terrace.The wider the terrace, the narrower the planting distance along terraces but up to certain limit.Using string, the distance of planting point along terrace (A) and the distance of terrace to terrace (B) is fixed to certain value based on target density.For palm density at 128 palm/ha, total A and B is fixed at 58 ft.while it is set at 56 ft. and 54 ft. for 138 and 148 palm/ha respectively.The B value is taken from somewhere in the middle of 2 planting point either in the upper or lower terrace.

Simulation of lining according to Gawthorn
Violle [9] method was established earlier and it was initially used for rubber.It is rather similar with Gawthorn [8] method but it adopts constant surface area equivalent to that which occupies on flat land.The surface area refers to area between palms along terrace (A) and between terrace to terrace in that particular length (B).For target density of 138 palm/ha, the constant surface area (AxB) is set to 72.48 m 2 while for 148 palm/ha, it is at 67.58 m 2 .In technical for field application, it uses string with colour code.B is ranged from 7m to 12m in 10 categories.The A value is determined based on B value so that the surface area meet the chosen constant value.Subsequent B value is taken from the last measurement point and it doesn't take into consideration the presence or distance of planting points in the upper or lower terrace.
Using terrace design discussed earlier, we simulated lining method from both Gawthorn [8] and Violle [9] in GIS package.In gently sloping land, Gawthorn [8] method works quite well and may produce desired palm density.In irregular terrain with sharp angles of ravines however, it was noted that the technique produces relatively lower palm density.Our simulation in mixed terrain condition produced palm density of 126 palm/ha using target density of 148 palm/ha (or only 85% of target).Since density was low, the method produced minimal (2%) canopy overlap beyond the threshold (figure 4a).Violle [9] method produced better palm density at 143 palm/ha (96% of target) but also resulted more palms (11%) with potential etiolation problem (figure 4b).We found that continuous B at high value results in continuous A at low value.This causes several palms along terrace having close planting that produce higher canopy overlap above the threshold.

Violle AAR method
We modified the Violle [9] method based on its limitation.Once A values meet condition in lower side (<6m) consecutively ie. 2 times in a row, the A value is then fixed to longer distance of 8m.After resetting this, the following A will follow back the original method until it meets other close distances again.With this exercise, there will be no close planting along the terrace of more than 2 palms and thus, close plantings are minimized.Simulation of this method resulted in palm density of 142 palms/ha (96% of target) but with potential etiolation problem reduced to 6% (figure 4c).
The method has been implemented in our commercial trial covering 246 ha.For field application, we developed an android application so that calculation of A and B is correct including the adjustment made when close plantings are identified.Every single points are also recorded in both the basic data and also GPS coordinates.From trial area, we obtained average palms density of 145 palms/ha that is even better than the simulation.Overall, our modification to Violle method produces palm density that is better than Gawthorn [8] method but having potential etiolation problem lower than Violle [9] method.

Concluding remarks
Advance development of geospatial technologies allows improvement in many aspects of businesses including in oil palm industry, particularly in this study is to help improve plantings in hilly terrain.New approach in road design was introduced by putting much attention to road density, road gradient and

Figure 1 .
Figure 1.Hilly terrain without terraces shows significant lower yield

Figure 2 .
Figure 2. Proposed road and drainage design technique: suggested different road type (A), gradient control for road in sloping terrain (B), drainage network based on DTM and catchment analysis (C)

Figure 3 .
Figure 3. Proposed terrace design technique: Gawthorn [8] technique as base line (A), differentiation of complex terrain using CT or terrace vertical distance (B), differentiation results in map (C), rajah line generation (D)