Related Pages:   Subwatershed Analyses

Estimating Watershed Nitrogen Loading with Parcel Data

In 1991, the BBNEP helped develop the first parcel level watershed analysis for Buzzards Bay. The evaluation, for Buttermilk Bay, was little more than a spreadsheet listing every parcel in the watershed (identified with GIS techniques), with assumptions about occupancy, lawn size, and loading rates built into the formulas. Thus each parcel was assigned a N-load contribution, and the total watershed loading was calculated by summing the individual parcel loads.

Technology, software, and available electronic data have changed considerably in the past two decades, and there are also new assumptions about attenuation of nitrogen through watersheds, and the relative contributions of some sources, but many of the principals of conducting watershed loading remain unchanged. This page identifies some principals, common sense approaches, and GIS tips for anyone conducting a watershed nitrogen loading assessment, using the loading rates adopted by the Massachusetts DEP’s Massachusetts Estuary Project. There are many steps involved with conducting a nitrogen loading analysis, but the most important and most challenging is to evaluate GIS, assessors, and land use data to locate and quantify nitrogen loading in a watershed. Once this data is derived, then it can be added to a spreadsheet to estimate loadings based upon accepted or adopted loading coefficients.

The techniques described here, and the loading demonstration example stepped through in the next pages, require moderate proficiency with ArcMap and Spreadsheet formula construction, including the use of array formulas. The approach described is appropriate for watersheds. Loading estimates from a smaller area, like a new subdivision, may require more exact measurements of roof, road, and lawn areas directly calculated from plans. Please report errors or send comments about this page to Joe Costa.

Incorporating Land Use areas from Other datasets into a Parcel Loading Analysis

In many respects, conducting a watershed nitrogen loading analysis is little more than enumerating the number of houses with septic systems in a watershed, approximating their occupancy (and thus loading from those septic systems), accounting for any permitted point source discharges (average flow times concentration), and estimating the loadings of certain non-point sources of nitrogen, based on their areas, including agricultural land, impervious surface, water surface areas, and undeveloped land. Because municipal data sets do not typically have accurate estimates of building footprints, actual turf areas of golf courses, or production areas of parcels classified as agricultural lands, any loading analysis will need to either estimate these areas by either 1) indirect means based on certain assumptions about average cover for a given parcel size, 2) directly digitizing relevant surface areas, or 3) relying on other spatial GIS coverages that reasonably approximate these areas. Because numerous spatial data sets exist relating to land use, given that these coverages are reasonably accurate, and given the fact that many of these sources typically represent less than 10% of watershed loading, these available spatial coverages can be used to avoid the time and expense of creating new spatial coverages.

Depending upon the watershed and GIS data sets available, the best strategy for estimating septic loading will be based on the enumeration of units and buildings from the assessors data, estimating average annual occupancy from municipal water use records, and determining septic and sewer service from sewer department billing records. With respect to areal based loading estimates, impervious area can be estimated from MassGIS impervious and road coverages (unless municipal building footprints and paved road areas are digitized) , and other land use acreage such as cropland, pasture, and golf course can be approximated from the 2005 Mass. Cranberry bog production area can be estimated from the DEP wetland Conservancy maps (or a comparable updated data set), with altered and denuded areas on cranberry bog production parcels estimated from the difference of GIS land use cranberry coverage area less the production area of the bogs from the Conservancy Program data set. These altered areas on cranberry bog properties can also be calculated from the MassGIS impervious cover, which includes bare land areas in addition to paved area. In the loading analyses presented on this website, any watershed area not accounted for by impervious, agricultural, water, or other measured land use type is assumed to be “natural landscape” by the difference in watershed area and the cumulative total of the other measured types.

Three Fundamental Principles or Parcel Level Loading Analyses

1) Nitrogen loading estimates of existing inputs used for establishing TMDLs should be based on best actual of quantities multiplied times an assumed loading rate. Where possible, analysts should utilize best available information about land use, occupancy (including seasonality), impervious areas, and other parcel specific information. The goal of the analysis is to accurately quantify discreet nitrogen sources as best practicable, then applying consensus loading assumptions.

Data used include the town’s property tax assessment database (“Assessor’s data”), GIS land use data and impervious cover, water use records, and census data. Decisions about which data sets are most meaningful are area-specific and may be based on best professional judgment. Compromises on the accuracy of certain loading components for individual parcels are acceptable if they represent a minor component of loading or if small positional errors cancel out over a larger number of parcels. Loading assumptions used to link watershed inputs to existing water quality may differ from loading rates used for regulatory purposes, where more conservative assumptions are used to provide a margin of safety to account for changes in future conditions. Techniques and methods used will depend on the accuracy and completeness of available data.

2) Loading spreadsheets and other calculations should preserve loading information to the parcel level (where applicable), or enable analysts to re-aggregate information to the parcel level (where applicable).

This approach allows maximum flexibility in adjusting loads and data at the parcel. Combining or merging data by assessor use code or some other attribute destroys this flexibility. It is essential that all databases (GIS dbf files and spreadsheets preserve a common-join file (map-lot or parcel id, and when split, a new parcel id field), to enable easy joining of GIS coverages with loading spreadsheets. If watersheds cross municipal boundaries, parcel coverages from all towns should be combined in a single data set to enable maximum analytical flexibility and data-processing to reduce work load.

3) Watershed loading estimates are based on a series of assumptions. All assumptions and formulas used should be made explicit and transparent, and original data sets on which the data are derived should be identified or made available, so that loading estimates can be independently validated and calculations corroborated.

Given that it may cost municipalities hundreds of millions of dollars to meet TMDLs, it is important that municipalities are confident that the source loading estimates on which a TMDL is based are accurate.

Key guidelines for handling GIS data for parcel level loading analysis

1) All coverages should be seamless, and cover the entire watershed.

All town parcels must be combined in a common database. Differing field names among town databases must be made consistent. In practical terms, this means that not only must slivers and gaps between parcels must be integrated, but roads and ponds often need to be added to coverages so that there are no gaps in land area. (Tip: Use the ArcMap Integrate function to eliminate smaller slivers.). Where towns do not have certain areas covered, such as roads and coastal waters, these should be created through unions with MassGIS Town Survey or Town Bounds coverages.

2) Join only what is required into the parcel shapefiles.

The MEP loading methodology requires estimates of wastewater flow, lawn or crop area, and impervious area (roof and road). GIS fields that that can help estimate any of these should be preserved or joined in later into the data set. Because many loading calculations are best handled in spreadsheets, it is important that spreadsheets contain the common join fields (e.g., Loc_ID or Map_ID). Then you can save your spreadsheet tables as dbf files and later join them to incorporate loading calculations into your ArcView projects.

Information Transparency and Municipal and Public Review

It is essential for key loading assumptions, and summaries of loading sources, to be presented clearly and succinctly in any loading analysis. Such data should also be presented in a way so that it can corroborated easily by municipal staff and environmental managers. In practical terms, this means that for any given watershed, the numbers of buildings, the number of units served by septic systems or sewer, and the acreage of key land use types (cranberry bogs, road area, roof area, etc.) should by provided. Such data can easily be independently corroborated. Maps should also be provided that show properties sewered or served by onsite systems, the number of units per parcel, and the potential buildout units on each parcel. Such maps can be prepared easily with GIS software using the data developed for a loading analysis, and the veracity and reasonableness of the data can be easily assessed by local officials in this format. This transparency of information can also build public confidence in the validity of a loading analysis used to develop a TMDL, especially if difficult management measures are required to meet TMDLs. We have employed these principals in our subwatershed loading estimates.

Special Problems

Parcels spanning more than 1 subwatershed

Because parcels may span one or more subwatershed boundaries, many different solutions are employed to assign loading from that parcel into loading estimates. Given that a wealth of data is available on land surface coverage, and given the powerful quantitative capabilities of GIS software, the BBNEP recommends that loading sources that can be derived by surface features (like extent of agricultural land, golf course area, and impervious surface area) be based on existing MassGIS coverages. The location of “point sources” like septic systems must be assigned to just one of the split parcels.

Tip: If the location of septic systems has not been mapped, before splitting parcels by watershed, create a geographic center point coverage (centroid), which will be the basis of a presumed septic location. This point coverage can also be used to join to water use records. The exact location of the auto generated point is unimportant, except in the case of larger parcels on a watershed divide, where the point can be dragged to a presumed septic location based on the location of a structure visible in the aerial photograph base map. Because lawn area cannot be auto calculated by existing GIS coverages, and since they generally represent a small percentage of watershed loading, the point coverage established for septic systems can be used to assign the location of lawns in the loading calculations (conveniently septic system leach fields are typically under lawns).

Go to the Wareham River Estuary Loading Example page to see a step-by-step example of how to do parcel level loading in a watershed.