Methodology and Data Sources

The intended outcome of these investigations is to discover points of intervention that will mitigate the degradation of the mangrove forest. The approach to the problem in this phase is to examine the history of mangrove degradation in the region. Correlating the history of mangrove conversion to other variables, such as land use, land ownership, crop production, public works, community attitudes and government policy will provide insights into the factors that have led to the degradation of this resource. A better understanding of the processes at work will lead to improved mitigation interventions.

The initial data source for the investigation was the INEGI 1:50,000 topographic maps because they were the most easily available source of information about mangrove distributions in the area. I digitized the mangrove distributions for the entire wetland complex using the ArcView Geographic Information System (GIS) software. The obvious concept was to take these data into the field and look for areas classified as mangroves to see if they were still occupied by this land cover thirty years later. This procedure produced some quite striking results when it was applied to an easily accessible sample area just to the north of the San Blas city limits. The sample area had 49 square kilometers classified as mangrove on the topographic maps. The field investigation revealed that approximately 44% of that area (21 square kilometers) currently falls in other classifications, as follows: 14 square kilometers in agriculture, 4 square kilometers in Shrimp Ponds, and 3 square kilometers in deforested salt flats. This result was truly alarming as it pointed to the loss of nearly half of the mangrove ecosystem in a period of 35 years.

Further investigation, however, revealed that these figures were extremely misleading. I acquired the 1970 air photos from which the topographic maps had been compiled by INEGI and it quickly became obvious that the mangrove distributions on the maps had been highly generalized in the interpretation process. The inescapable conclusion is that the data on the maps are completely useless as a base line from for measuring change in the mangrove ecosystem. The other conclusion is that the figures reported above for the extent of the mangroves in the entire wetland complex, 1,117 square kilometers, is likewise highly suspect, though it is approximately the number that is most widely reported in the existing literature.

This conclusion meant that I was back at square one in my attempt to map mangrove distribution. My next thought was to use satellite imagery. I investigated the acquisition of SPOT and Landsat TM imagery, but the cost of both were beyond the reach of my financial resources. I was able to acquire Landsat Multi-spectral Scanner (MSS) imagery from the EROS Data Center National Land Classification Data (NLCD) set. This program provides three MSS images of a single scene for the years 1970, 1980, and 1990 and a digital terrain model for a price of US$15.00. Though the price was right, the resolution was not. The resampled pixel size is 60 X 60 meters, which proved to be far too coarse to identify much of the mangrove in the study area.

The only remaining data source within our means was aerial photography. INEGI was able to supply orthophoto images derived from 1997 air photos. Five images were sufficient to cover almost the entire study area. The location of these images are shown in the map below.

Location of Orthophotos

As the location map shows, there is one small area of about 1,500 ha. on the northwest corner of the study area to which the orthophoto coverage does not quite stretch. We felt that it was more cost effective to simply make a field trip to this area. The spatial resolution of these images is 2 X 2 meters, which is more than adequate to identify mangrove distributions. The cost of these images was $1,500 Mexican pesos (about US$150) each. These data provide an excellent source for reasonably current information on mangrove distribution.

The orthophoto images were overlaid on the base map data in the Arcview GIS using the roads on the digitized topographic maps to match the position of the roads on the orthophoto images. It quickly became apparent that the rectification of the orthophotos was not perfect. When the edges of the images were mozaicked to a best fit, the road systems did not match. This artifact forced a choice between minimizing discontinuities at the edges of the images or minimizing the positional error at the center of the images.

Given the small size of the study area I chose the first option. The images were arranged into a mosaic that most closely matched at the edges, then the entire mosaic was registered to the road system at the center of the study area, just to the north of the Port of San Blas. The nature and magnitude of the error that this decision introduces into the interpretation is illustrated in the figure below.

Location of Orthophotos

The enlargement on the lower left shows the image registration at the road intersection that was used to register the image, and the enlargement on the upper right shows the resultant error near the northern edge of the study area. The dotted red lines are the road positions on the topographic maps. The registration displacement in this second enlargement is approximately 150 meters in both the x and y axis directions. I have now changed my opinion, and believe that it would have been better to register each image independently and accept the discontinuity at the image boundaries. The procedure that was used will not affect the mangrove area measurements that were interpreted from the images, but the errors will affect the correlations with other data base layers. Since the error is cumulative it will increase when the study area is enlarged to include the rest of Nayarit State and, eventually, the entire wetland complex. My intention is to reposition the images and the derived interpretations before expanding the study.

Once registered to the maps, the images were manually interpreted by drawing polygons around areas identified as mangrove. Most boundaries were very distinct and easily drawn. These include the mangrove-water, mangrove-bare soil, mangrove-agriculture interfaces and areas where the mangroves are bounded by steep slope inflections.

Other boundaries were more difficult to discern. This was especially the case where the mangrove had been heavily modified by cattle grazing. The transition between mangrove and monte seco (dry brushland) was also difficult to detect. These two different vegetation types have very similar tones and textures on the images, and only a few centimeters of elevation separate them. Field verification was the only means of detecting this difference. This was possible around the edges of the forests, but suggests that there probably are "islands" of monte seco in the interior of the mangroves that have been misidentified.

Thirty one areas proved to be difficult to interpret, and were deemed to be candidates for field verification. The location of these areas is shown outlined in yellow on the map below.

Field Verification Sites

Field verification involved going to the site in question and observing the boundaries that were established by the image interpretation to establish their validity. Trips were taken on foot, by bicycle, by boat and in automobiles or a combination of all of these, depending on the location of the area in question. This procedure was very time consuming, and wherever possible local inhabitants were interviewed to obtain their knowledge of the location of the mangroves. Individuals engaged in cutting mangrove for poles proved to be particularly useful in this regard.

Enlargements of the areas of interest on the images were used as the basis for the field verifications. The enlargements were overlain with a UTM grid calibrated at 500 meter intervals, which, in combination with a GPS receiver, was used to navigate to the area of interest. Corrections to the interpretations were noted directly on the enlargement and subsequently entered into the GIS data base. A