Ante tales circunstancias a criterio de quienes resuelven el presente medio de impugnación se estima pertinente traer a colación la siguiente normatividad
LEY ORGÁNICA DE LA DMINISTRACIÓN PÚBLICA DEL DISTRITO FEDERAL
In 2013, radar engineers from the Council of Scientific and Industrial Research (CSIR) embarked on a radar clutter measurement campaign conducted from an airborne platform using an L-band radar [4]. This campaign was funded by the Armaments Corporation of South Africa (ARMSCOR). The campaign was conducted in the Western Cape region of South Africa. Measurements were taken of five terrain types, namely farmlands, fynbos, commercial urban, residential urban and informal urban areas. Permission was obtained from CSIR and ARMSCOR to be granted access to this measurement trial data to use for the purpose of this research study. The data collected from these measurements were
processed to determine the backscatter coefficient σ◦ in dBs as a function of grazing angleδ for the five terrain types. The range of measured angles are in the plateau and high grazing region. This is shown in Figure 4.2.
Figure 4.2: Average backscatter coefficientσ◦F4for various terrain types.
From Figure 4.2 it can be seen that there is general dependence ofσ◦on grazing angle in the plateau and high grazing region that agrees with literature. It also shows that urban terrains produce σ◦ values that are of an order of magnitude higher than farmlands or shrub like plantation terrain types. This is in agreement with the the list of given land clutter models, as shown in Figures 3.1, 3.2, 3.3, 3.4, 3.6 and 3.5.
Figures 4.3, and 4.4 show a comparison of the measuredσ◦ values for different terrain types withσ◦values simulated using the various land clutter models described in chapter
3 for plateau and high grazing angles. The measured data is compared to land clutter models terrain types that are the same or similar. Measured farmland and fynbos data is compared with land clutter model, Constant Gamma with terrain type ‘farmland’ and ‘flatland’, Morchin with terrain type ‘farmland’ and ‘desert’, Kulemin with terrain type ‘arable land’, GTRI with terrain type ‘tall grass crops’ and ‘soil, sand and rocks’, and Ulaby and Dobson or Generating Function with terrain type ‘shrubs’. Similarly measured commercial urban, residential urban, and informal urban data is compared with land clut- ter model, Constant Gamma with terrain type ‘metropolitan’, Kulemin with terrain type ‘urban territories’, GTRI with terrain type ‘urban’ and ‘soil’ and Ulaby and Dobson or Generating Function with terrain type ‘urban’.
The root mean squared error (RMSE) between the measured and simulated average backs- catter (σ◦F4) data for plateau and high grazing angle regions is shown in Figure 4.5.
Figure 4.3: Average backscatter coefficientσ◦F4 for terrain types farmland and fynbos compared to land clutter models.
From Figures 4.3 it can be seen that the measured data for farmlands is for most grazing angles larger than the simulated data from land clutter models. It is believed that this is due to discrete scatterers such as tree trunks interspersed in the homogeneous terrain. As expected, the land clutter model that closely agrees with the measured farmland data in the plateau grazing angle region is the Ulaby and Dobson and Generating Function land clutter models with terrain type (shrubs), with an associated RMSE of 1.4 dBs, and 2 dBs respectively. The backscatter from the measured fynbos data is significantly larger than the measured farmland data for the plateau grazing angle region and decreases in difference as the grazing angle moves towards the high grazing angle region. As with the farmland data, the land clutter model that is closest to the the measured fynbos data is the Ulaby and Dobson and Generating Function land clutter models with terrain type (shrubs). Even though fynbos is considered shrub-like terrain type, it produces a RMSE that is as much as 9.9 dBs, and 10.5 dBs for the simulated Ulaby and Dobson and Generating Function land clutter models respectively. This is again believed to be as a result of discrete scatterers. The land clutter model that produced the largest RMSE (larger than 34 dBs) relative to the measured data was for all cases the Kulemin land clutter model. This agrees with the assumption that Kulemin is a considered a weak and low validity
Figure 4.4: Average backscatter coefficientσ◦F4for all urban terrain types compared to land clutter models.
land clutter model. From literature it was believed that the GTRI land clutter model would be a more valid model to use for farmlands than the Constant Gamma land clutter model. However from these figures it is seen that the Constant Gamma land clutter model produces less of an RMSE (11.5 dBs against measured farmlands, and 20.3 dBs against measured fynbos) relative to the measured data than the GTRI land clutter model (RMSE of 15.8 dBs against measured farmlands, and a RMSE of 24.6 dBs against measured fynbos) for grazing angles in the plateau region. For high grazing angles, the Ulaby and Dobson and Generating Function land clutter models produces values that are closest to the measured data (RMSE of 7.8 dBs and 8.1 dBs against measured farmlands, and a RMSE of 10.4 dBs and 10.6 dBs against measured fynbos), followed by the Morchin land clutter model (RMSE of 8.9 dBs against measured farmlands, and a RMSE of 10.5 dBs against measured fynbos). However, for the limited grazing angle range between 70◦ and 90◦, the Morchin land clutter model has the lowest RMSE (8.9 dBs against measured farmlands, and 10.5 dBs against measured fynbos) as compared to the Ulaby and Dobson and Generating Function land clutter models (RMSE of 11.1 dBs and 12.1 dBs against measured farmlands, and a RMSE of 11.8 dBs and 12.9 dBs against measured fynbos). From Figure 4.4 it is seen that commercial urban terrains produces larger backscatter coefficient values than that of residential and informal urban terrains. This is due to the the large scattering sources such as tall buildings in commercial urban areas. The measured
(a) (b)
(c) (d)
Figure 4.5: RMSE of the measured average backscatter σ◦F4 for and simulated data from land clutter models : (a) RMSE error between measured and simulated land clutter for plateau grazing angle region , (b) RMSE error between measured and simulated land clutter for high grazing angle region , (c) RMSE error between measured urban terrains and simulated land clutter for plateau grazing angle region , (c) RMSE error between measured urban terrains and simulated land clutter for high grazing angle region.
urban data is as much as 10 dBs larger than that of the closest land clutter model. The land clutter model that produces values that are closest to the measured Urban terrain data is the Constant Gamma land clutter model for grazing angles in the plateau region, with an associated RMSE of 10 dBs, 5.5 dBs, and 5.3 dBs for commercial, residential and informal urban respectively). This is followed by GTRI, Ulaby and Dobson and Generating function, and Kulemin (limited grazing angle range for Kulemin model) land clutter models respectively. As with the measured farmland and fynbos data, the Constant Gamma land clutter model performs better than the GTRI land clutter model. The land clutter model that produced the largest RMSE relative to the measured urban data was the Ulaby and Dobson and Generating function land clutter model for grazing angles in the plateau region. This is due to the lack of sufficient data available for urban terrain in the Ulaby and Dobson and Generating Function land clutter models. Only the Ulaby and Dobson and Generating Function is valid for the high grazing angle region for urban
terrain. Care should be taken when modelling for urban terrain in the high grazing angle region as the RMSE is large (15.5 dBs, 10.4 dBs and 11.4 dBs for commercial, residential and informal urban respectively).