Herefordshire Ancient Woodland Inventory (AWI) revision David Lovelace July 2014
1. Introduction
1.1 The statutory definition of ‘ancient woodland’ is that which has been continuously wooded since at least 1600. Although archive sources for this period or earlier give valuable insights into how the county’s woods and trees were managed and used, details of the extent and boundaries of a particular wood can rarely be deduced. Large scale maps prior to 1800 were commissioned for private estates and are usually of high quality but those known to exist cover but a small proportion of the county and only some of these are accessible for research (see Brian Smith ‘Herefordshire Maps 1577 to 1800’ Logaston 2004 and its supplement Woolhope Naturalists’ Field Club 2012) . The first county maps to depict the county’s woodlands date from the early 19th century county maps and include those by Henry Price 1817, Andrew Bryant 1834 and the Ordnance Surveys Drawings (OSD) on 9 sheets covering Herefordshire 1813 – 1816 at 2 inches to the mile. These maps are however unreliable for woodland boundaries and locations and often do not depict small woods. Despite these limitations they can be a useful additional source of information in some cases if used with care as discussed in more detail below.
1.2 The current AWI for Herefordshire (1984 with some subsequent partial revisions) is derived mainly from the first edition 1 inch to the mile Ordnance Survey (OS) maps of the county c1835, itself based upon the OSDs, with the assumption, based upon general land use history, that woods so depicted would have been present a couple of centuries earlier unless there is evidence otherwise.
1.3 This present revision has used the parish tithe maps drawn up as a result of the 1836 Tithe Commutation Act as a baseline for determining historical status. Not all land was ‘titheable’, examples being land subject to enclosure awards and land historically exempt for ecclesiastical reasons but at 94% Herefordshire has an unusually high proportion of land area coverage of tithe maps. These are accurate large scale maps (between 14 and 26 inches to the mile) of each parish surveyed using triangulation. Even very small land parcels are depicted but importantly details of each parcel including land use are listed in the associated ‘apportionment’. Tithe maps are approximately contemporaneous with the first 1 inch to the mile OS maps used for the existing AWIs. Historic woodland cover is determined by comparing current woodland boundaries using aerial photography and OS MasterMap polygons with the same woodland as it appears on the tithe map. Continuity of that woodland land use is then checked against the subsequent large scale first edition OS 25 inch to the mile map of c1885, the 1953 census of woods and where necessary with the early 19th county maps and, where they exist and are available, pre 19th century estate maps.
2. Methods
2.1 This AWI revision revises the current inventory (AWI version 2.5 on MAGIC web site) by analysing the continuity of woodland cover of land parcels in Herefordshire as indicated by a time sequence of the following geo-rectified digitised maps and aerial photography:
· Parish tithe maps 1838–1844.
· First edition 25 inch to the mile OS c1885
· Forestry Commission census of woods 1953 compartment boundaries.
· Digital boundaries of the current AWI.
· Current digital colour air photos (UK perspectives).
· OS MasterMap digital boundaries.
This revision has also used, but not relied upon, the following maps in digital form:
· The 2 inch to the mile Ordnance Survey Drawings (OSD) 1813-1816.
· Price’s map of Herefordshire 1819
· Bryant’s map of Herefordshire 1832-1835
· Estate maps where accessible
2.2 Woods in whole or part which are missing from the current inventory but are depicted as woodland on the above maps and aerial photographs are considered to have sufficient historical continuity to be put forward as ‘candidate ancient woodlands’. The majority of these are small woodlands, typically < 2 hectares, which where omitted from the original AWI.
The polygons of the existing AWI have been similarly reviewed and edited accordingly by deletion or addition in whole or part. In those cases where the evidence is equivocal but strong enough to suspect the likelihood of historical woodland or wood pasture continuity, that woodland is ‘flagged’ pending further evidence especially from the field.
The mapping of candidate additions and removals is based upon current OS MasterMap woodland polygons in combination with recent aerial photographs. Where necessary these polygons are edited to remove any areas failing the continuity criteria.
2.3 All the data of this AWI revision including the digitised maps above exists in a Geographic Information System (GIS) the components of which are online www.r5r.eu/awir.html which includes links to a number of online sample map and air photo sequences illustrative of the methods used and the data issues that have arisen. These same links can be made directly from this document in the discussions below. Also on that web page is the presentation of the Herefordshire AWI revision from the national seminar at Peterborough 12th February 2014.
Descriptions of the data fields and abbreviations used are detailed in section 7 below.
3 Using the parish tithe maps
3.1 Dating back more than a millennium, the tithe was a religious tax consisting of a tenth part of the produce of the land, and over time became a regular source of legal dispute. The 1836 Tithe Commutation Act attempted to make this church tax uniform, predictable and fair across the country by commuting the complex and varied customary payments to a standard annual charge levied on each parishioner according to the extent and productivity of their land. The Act therefore required a large scale and detailed mapping of each parish to determine an individual’s tithe dues. The resulting ‘tithe maps’ along with their ‘apportionment’ of land use, tenure and ownership remain a unique and detailed record of even the smallest parcels of land.
The ‘apportionment’ of each parish tithe map includes the ‘state of cultivation’ which is typically ‘arable’, ‘pasture’, ‘meadow’, ‘orchard, or ‘woodland’. Woodland categories include ‘woodland’, ‘coppice’, ‘ash bed’, ‘plantation’, ‘orles’ (alder trees), ‘osier’, ‘grove’ and ‘brake’ (clump of bushes). Care has be taken when interpreting ‘pasture’ or ‘rough pasture’ as this may refer to a woodland open to grazing, the grazing income being the titheable but not the trees that may be present. In these cases it is necessary to cross check against the later 25” to the mile OS map which depicts tree symbols, the original tithe map itself, earlier county and estate maps.
3.2 An example is the companion to Pregg Wood in Newton St. Margaret parish subject of a recent planning application, see map sequence www.r5r.eu/fof.html. The tithe map records the land use as pasture and the name “Cae Main” [= ‘Main field’] is in contrast to the adjacent parcel which is recorded as woodland and called “Preggs Wood”. However the subsequent 25 inch to the mile map (c1885) depicts 2/3 of Cae Main as treed and the two county maps (Bryant and Price) along with the OSD show Cae Main as at least partly woodland. As discussed below attempts to geo-reference county maps and the OSD have generally proved unsatisfactory in defining the locations of woodlands and their boundaries many of which are located away from roads travelled by the surveyors at that time for gathering data for their maps. In this particular case the county maps have proved more consistent than other cases so geo-referencing to the modern grid is more accurate probably due to the close proximity of the wood to the road.
This sequence also shows (a) the AWI revision data fields for the two woodland polygons featured (b) tithe map colour coding used in this revision and (c) compares the two recent aerial photographic surveys: UK Prespectives 2002 and Bing maps 2012.
3.3 Being the responsibility of each parish, the quality and style of each tithe map varies considerably. Only about a quarter of all Herefordshire tithe maps are classified as ‘high quality’ that is depicting tree symbols and other features. These ‘high quality’ tithe maps depict the distribution of trees within fields and parks so allowing the continuity of woodland and tree cover to be inferred even when the land use is recorded as ‘pasture’ on the ‘apportionment’.
3.4 The historian and calligrapher Geoff Gwatkins has transcribed all the Herefordshire tithe maps onto a standard 6” scale and entered the land use, parcel number and name onto each parcel. These are more convenient to use than the original tithe maps which generally only have the parcel number. We have with permission scanned and geo-referenced printed copies of all of Geoff’s copyrighted tithe maps (referred to as TMGs here) and have then digitally colour coded each parcel for its apportionment land use. As a background layer within GIS this tithe map colour coding allows for a quick assessment of woodland and other land uses at this time.
3.5 Because TMGs do not have the ancillary pictorial information such as tree symbols, we have digitised and geo-referenced many of the original high quality tithe maps using the ‘deposit copies’ at the National Archives. For example, the sequence http://r5r.eu/walterstone.html is of a 800m x 700m sample of the parish of Walterstone demonstrating the detailed accuracy of some quality tithe maps and the limits of early 19th century county maps when it comes to showing historical continuity for small woods. Referring to that sequence, the current triangular wood is a surviving part of the larger Cae Gwynne Wood present on all maps except Price (1817). The small ‘flagged’ wood called ‘grove’ on the tithe map is depicted as an unfenced group of trees on the tithe map and on the 25 inch OS (c1885). None of the county maps pick up on this detail and the Price map shows the whole valley devoid of woodland. Price’s map is inconsistent since he shows two small woods in another parish while the others omit it (see 5 below). See the comparison between Geoff Gwatkin’s transcription with the original tithe map. Note the modern expansion of secondary woodland along the dingle and the loss of most of the historic woods.
4 Using early 19th century county maps
4.1 Problems in interpreting tithe maps as described above can sometimes be resolved by comparing them with the geo-referenced digital copies of the early 19th century county maps. This cross check is much less reliable for small woods which are often not depicted at all. The other problem with the OSD and county maps is that, unlike the tithe maps, they were not surveyed to an accuracy and consistency that allows them to be geo-rectified with any accuracy.
4.2 The Haywood sequence http://r5r.eu/haywood.html
This 4km x 3km swathe of countryside just SW of Hereford is subject to a proposed bypass along with related development pressure. Characterising its woodland heritage illustrates the difficulty of using early 19th century county maps and the OSDs. Three locations, chosen for being unchanged for well over two centuries, are used as the geo-referencing coordinates for a linear ‘affine’ transform of the relevant section of digitised county maps: Price’s (1817), Bryant’s (1834) and the OSD (1816). An affine transform is a ‘linear transform’ in that it only changes the location, scale and rotation of the target image. The first three images of the sequence - air photo, 25” to mile OS 1st edition and the tithe map - are all accurately located with reference to the OS MasterMap polygons of the AWI woodlands.
However when these same woodland boundaries are overlaid upon the geo-referenced county maps, (a) the locations are displaced, sometimes so far as to make identification of an individual wood doubtful, (b) the boundary shapes are approximate and (c) many small woods are not depicted. The larger woods are generally recognisable in shape and location but the smaller woods are either omitted or greatly displaced. Any historic map (or air photo) can be ‘forced’ to align with any number of control points in the base map using non-linear image warping algorithms and indeed this AWI revision has relied on these methods to ensure that the tithe maps and the 1953 census compartments are accurately aligned to the modern grid through the county using a high density of control points, typically up to 10 per 1km x 1km.
This method only works if the maps are reasonably accurate and rectilinear in the first place, so that the non-linear deviations from affine linearity are small, typically less than a few 10s of meters. For maps where locations are displaced from their actual positions by 100s of meters and furthermore the size and directions of such displacements are not consistent, then non-linear techniques do not produce meaningful results. This is especially so for features some distance away from obvious surveyor land marks such as cross road, churches and stream crossings as is the case for many woodlands. This is not to say that early county maps cannot be used for woodland history analysis, far from it, as they are an important additional source of historical evidence. However their use requires judgement and their efficacy reduces in proportion to the size of woods, as this sequence and others linked to this document, demonstrate. An example of their use is in the above link to Pregg Wood and its companion whose its proximity to a cross roads ensures a much better than usual alignment with the rest of the times series.
4.3 The Norton sequence http://r5r.eu/norton.html
This 2km x 2km area between Bromyard and Tedstone Delamere focuses on the historically complex wood known as Vinschurch Plantation. Even though one of the control points on a cross road is less than 200m from the well defined NE tip of the wood (which is also the old township boundary of Norton) Bryant (1834) displaces the wood about 150m SE from its true location. Its location in both Price (1817) and OSD (1815) is better, but the boundary details are less accurate than Bryant. All three show the woodland occupying less area than now except Bryant which shows the southern part as rough or scrub. The first edition 25 inch OS (c1885) depicts the whole wood with scattered trees implying at least wood pasture. Price and OSD omit all the small woods present on the tithe map but Bryant maps more of them including the existing ancient woodland inventory wood Tedstone Orles which the other two map miss out completely.