CAPITULO XX Aspectos ambientales
CAPITULO XXIII Garantías mineras
As already outlined, two approaches to combining sub-model outputs are demonstrated here: an ordered weighted average of temperature and land suitability results, and use of specified limits for temperature and water availability to define an area within which land suitability can be considered.
117 In each case Nile tilapia (Oreochromis niloticus) were considered as a model species with a wide distribution in aquaculture terms.
Figure 3-28 shows outputs from the Combination of land suitability and temperature using an OWA under late 20th century conditions and under an average 2°C global temperature increase . For display purposes every second month is shown. Figure 3-29 highlights the difference between the images shown in Figure 3-28 under the two climate scenarios. In this example there are regions that stand to both gain and lose in terms of suitability depending on the time of year. The Bangladesh area provides a good example of this with improved suitability score being seen for the colder months (December and February) while looking at the outputs for August there is a slight decrease in suitability in association with temperature above those specified as optimum. A similar situation is seen in Thailand and some other areas of south East Asia. Southern China is also notable as somewhere where there could be potentially positive impact in terms of an improved growing season.
The results provided in the current example are obviously highly dependent on the reclassification of the temperature data. In the case of Nile tilapia it would potentially have been reasonable to extend the temperature ranges used to represent conditions that would more represent possible rather than optimal ranges while at the same time excepting a greater degree or risk in relation to unusually hot or cold periods. Such an approach, especially in the case of colder temperatures, could be used to indicate areas that would currently be considered marginal in terms of fish survival during colder months that may become more usable under warming conditions.
Figures 3-30 and 3-31 show examples of outputs obtained by using specified limits for mean, minimum, and maximum temperature along with water availability to specify areas which can then be considered in relation to land suitability. Table 3-13 shows areas (km2) for Figures 3-30 and 3-31 as well as for where ponds are expected to contain water for 9 and 12 consecutive months per year. Results are shown based on both the GlobCover based land suitability sub-model output and that produced using MODIS land cover data. The areas defined by temperature and water availability are notably larger under the warming scenario which can be accounted for by the increased temperatures moving new areas into the defined acceptable range and again highlights the potential for increased production in many areas in relation to warmer conditions.
118 This method contrasts to the previous one in that there is no 'fuzziness' in relation to temperature suitability; areas are either considered useable or excluded. The advantage of such an approach is that by asking a number of specific questions of the database a potentially more focused answer is provided. In this case it allows for the land suitability sub model results to be considered without being 'diluted' through combination with other data. In theory this exercise could be repeated using a number of different temperature specifications to allow for a more detailed picture of suitability to be built up over different regions. Another potential advantage to using a decision tree approach to ask questions of the data is that it can avoid the need to standardise data to a common scoring system (Drobne and Lisec, 2009) which if not done with care may lead to potentially questionable results when factors are then traded against each other using weighted averaging methods. Finally, there is the potential to eliminate the risk of an area being indicated as suitable after combination of a range factors using a weighted average when according to one of the factors it is not.
A key disadvantage of a decision tree approach is that it forces those developing it to make choices that can have a very definitive effect on the final outcome in a way that can be conceptually much more complex. It also potentially forces strict decision-making in situations where that may be very difficult due to levels of uncertainties about how a given variable relates to suitability, especially in the case of variables that may influence suitability but which are not considered critical. Another consideration is that the reality of working with spatial data sets often involves some degree of uncertainty about data quality and again in situations where such data relates to non-critical attributes the ability to allow the data to have some degree of influence on the final outcome without it having to result in a definitive answer can be potentially very useful.
In summary, spatial data can be combined in a range of diverse ways to assist decision-making. A substantial advantage of a GIS approach is that once a spatial database is established changes can be made to decision support models relatively easily in relation to varying requirements and decision questions. It is suggested here that the use of multi-criteria evaluation approaches such as WLC and OWA would seem especially useful for the broad-scale assessment of aquaculture site suitability by allowing for the incorporation of variables with different levels of significance and degrees of uncertainty. The use of Boolean intersection, or constraints, is also valuable in allowing the exclusion of highly unsuitable areas as well as providing a means of investigating changes in potentially suitable area under changing conditions in relation to a specific set of requirements.
119 February (late 20th century conditions) February (2°C average global warming)
April (late 20th century conditions) April (2°C average global warming)
June (late 20th century conditions) June (2°C average global warming)
120 August (late 20th century conditions) August (2°C average global warming)
October (late 20th century conditions) October (2°C average global warming)
December (late 20th century conditions) December (2°C average global warming)
121
February April
June August
October December
122
Figure 3-30: Areas meeting specified temperature requirements and with water predicted in rain fed ponds for at least 6 consecutive months based on the low seepage rates and precipitation over an area representing 150% pond surface area. All climate data based on late 20th century conditions. Defined areas are overlayed with outputs from the land suitability sub-model based on MODIS land cover data (See: Combination of sub-model outputs section for full details).
Figure 3-31: Shows the same output as Figure 3-30 but based on the 2°C average global warming scenario rather than late 20th century conditions.
123
3-13: Areas (km2) obtained by using specified limits for mean, minimum, and maximum temperature along with water availability using MODIS or GlobCover (see Figures 3-30 and 3-31).
Suitability
Percentage change between late 20th century and 2°C average global warming scenarios
Late 20th century conditions 2°C average global warming
Number of consecutive months where model ponds have at least a 90 percent change of being over half full.
Number of consecutive months where model ponds have at least a 90 percent change of being over half full.
Number of consecutive months where model ponds have at least a 90 percent change of being over half full. 6 months 9 months 12 months 6 months 9 months
12
months 6 months 9 months 12 months MODIS Constraint 5,152,099 4,599,448 3,396,707 5,846,756 5,331,893 4,698,242 113.48 115.92 138.32 1 363,725 254,716 156,999 521,385 395,587 296,310 143.35 155.30 188.73 2 1,559,071 1,185,971 823,249 2,094,198 1,730,445 1,372,258 134.32 145.91 166.69 3 3,066,259 2,411,789 1,733,814 3,498,741 3,013,081 2,344,133 114.10 124.93 135.20 4 752,919 497,152 319,684 823,134 629,220 404,900 109.33 126.56 126.66 5 493,003 229,670 121,356 380,156 274,944 127,201 77.11 119.71 104.82 GlobCover Constraint 5,176,209 4,621,917 3,416,816 5,872,550 5,354,701 4,717,648 113.45 115.85 138.07 1 271,755 200,509 126,666 391,728 308,048 232,841 144.15 153.63 183.82 2 1,472,278 1,125,999 781,113 1,970,943 1,617,660 1,287,193 133.87 143.66 164.79 3 2,821,406 2,201,061 1,576,090 3,299,626 2,834,253 2,222,976 116.95 128.77 141.04 4 966,827 684,602 450,558 1,064,434 843,396 576,812 110.10 123.20 128.02 5 667,674 336,144 195,071 553,725 407,489 200,580 82.93 121.22 102.82
124