THE MAPPING OF THE PLANT FORMATIONS FROM THE LEFT SIDE OF IZVORU MUNTELUI-BICAZ RESERVOIR OANA ZAMFIRESCU *, ŞT. ZAMFIRESCU * Introduction

Transcription

1 Analele ştiinţifice ale Universităţii Al. I. Cuza Iaşi Tomul L, s. II a. Biologie vegetală, 2004 THE MAPPING OF THE PLANT FORMATIONS FROM THE LEFT SIDE OF IZVORU MUNTELUI-BICAZ RESERVOIR OANA ZAMFIRESCU *, ŞT. ZAMFIRESCU * Abstract: Land formation mapping is possible, among other methods, through the interpretation of satellite images of a certain area. The studied area (Izvoru Muntelui-Bicaz reservoir left side) was photographed by the Landsat TM5 system on The row images were used to create a true colour composite image and a false colour composite images that supplied complementary information. The pixels of the false colour composition image were reclassified according to the reflectance of several standard areas that were selected during a field phytosociological research. The result was a map of the main land cover formation. Further processing of this map allowed the computation of proportions of the land cover formation areas. The best-represented formation is the deciduous forest (32,45%), which is followed in descending order by grassland (20,16%), partly eroded area (17%), water (16,83%), coniferous forest (5,66%), rocky area (4,67%) and buildings (3,24%). The high percentage of the partly eroded areas indicates an important human impact, especially over grasslands. This fact would endorse at least a land usage monitoring. Indirectly, the location of the eroded areas in grasslands outlines the role of the forests in the land protection against erosion. On a long-term scale, this phenomenon may have a harmful effect over life and energetic use of the reservoir through the increase of sediment content in water and, consequently, through the slow decrease. Teledetection allows the localisation of land formations and the assessment of their status and quality Keywords: land formations mapping, teledetection, satellite imagery processing Introduction The mapping of the vegetation from a certain area can be done by several methods. This paper takes into consideration the teledetection method that involves satellite imagery. The studied area is on the left slopes of Izvoru Muntelui-Bicaz reservoir from the southwestern part of the Stânişoara Mountains from the Oriental Carpathians (Neamţ district, Romania). This area was the subject of several phytosociological studies [2, 3, 4]. The most recent one was carried out between 1998 and 2003 [5, 8, 9, 10, 11, 12]. The mapping through teledetection gave us the opportunity to expand the results from the analysis of certain sample areas to the entire studied area, in order to supply a synoptic picture of its status. * Al. I. Cuza University Iaşi, Faculty of Biology 131

2 Material and Method The satellite images were taken by the Landsat TM5 system on Thus, we obtained seven images, one for each band of the multispectral sensor. Each image supplied certain information about the land cover formations (tab. I). Band TM Table I. The basic characteristics of the TM5 sensor [after 13] Electromagnetic Resolution Utility spectrum Spectral range (µm) 1 0,45 0, ,525 0, ,63 0, ,75 0,90 5 1,55 1, ,40 12,50 7 2,09 2,35 visible blue-green (reflected) visible green (reflected) visible red (reflected) near infrared (NIR) (reflected) mid-infrared (MIR) (reflected) thermal infrared (emitted) middle infrared (reflected) 120 m penetrates water, sensible to suspensions (sediments, plankton) together with bands 1 and 3 creates true colour composite images together with bands 1 and 2 creates true colour composite images this form of radiation is sensitive to a high degree of leafy vegetation since it has a high albedo in this band this portion of the spectrum is sensitive to variations in to variations in water content in both leafy vegetation and soil and in Fe 2 O 3 content in rocks and soils this radiation is detected as heat energy; the sensor can distinguish a temperature difference of about 0,6 o C this radiation is sensitive to humidity, so it is suited to detecting certain hydrophilous minerals and the moisture contents in soil and vegetation, which vary in stress conditions Consequently, these images were processed with specific software (Idrisi for Windows 2.0 and Multispec 2.5). Composite images were created by replacing each colour channel (red, blue and green RGB) with different bands (TMs). These images were used to extract the reflectance (spectral signatures) of different areas, which were considered as 132

3 standards for certain land cover formations. The signatures were further used as criteria of reclassification, which used the minimum distance to mean algorithm. Results and Discussion The colour composite image 321 (Fig. 1) is very similar to an aerial photo. On this image one can see just the major formations as forests, grasslands etc. Fine discriminations of coniferous and deciduous forests and natural and altered grasslands are not possible on this image. Therefore the colour composite image 542 (Fig. 2) was created. This image involves the infrared radiation discrimination power, especially NIR. Chlorophyll absorbs visible radiation with wavelengths between 0,4 0,7 µm. The reflection of NIR is positively correlated with the foliage development [1, 7]. Thus the colour composite image 542 is useful for the fine discrimination of forest and grassland characteristics: the coniferous forests appear in dark green, the deciduous forests appears in green and the grasslands appear light green or in yellowish green [14]. Obviously, the 542 colour composite image would be difficult to interpret without taking into consideration the true colour (321) composite image. In the 542 colour composite image the limit between deciduous forests and some grasslands is hardly visible, because the green hues are quite similar for the human eye. Therefore, forests boundaries are obvious on the true colour composite image, while the 542 colour composite image allows the fine differentiation between the two forest types. In addition, the 542 colour composite image, unlike the 321 images, permits the recognition of eroded areas, buildings and rocks. This characteristic is caused by the sensitivity of 1,55 1,75 µm radiation (MIR) to water content and Fe 2 O 3 content in the surface formations. The reflectance of the surface formations in each spectral domain (MIR, NIR and visible green) is important because it clearly reveals the spectral signatures of each formation from the studied area. The waters reflect more in visible green, less in NIR and almost nothing in NIR (Fig. 3). The buildings have a constant decreasing reflectance from maximum in MIR, through medium in NIR, to minimum in visible green (Fig. 4). Partly eroded areas posses a spectral signature quite similar to the buildings one, with the difference that the reflectance in NIR is higher because of the scarce but present vegetation from these areas (Fig. 5). The areas covered with rocks and pebbles have a low reflectance in NIR and visible green (Fig. 6). Among the plant formations, the grasslands have the highest reflectance (Fig. 7); the highest value is in NIR, followed by the one in MIR and visible green, respectively. The 133

4 high reflectance in MIR is caused by the small height of the grass layer. The comparison between grasslands and partly eroded areas signatures reveals that in the first case the reflectance in NIR is greater than the one in MIR, while in the second case the reflectance in NIR in smaller than the one in MIR. The forests have a relatively low reflectance in MIR. The forests composed mainly of deciduous trees reflect more in NIR and visible green (Fig. 8) than the ones in which coniferous trees are dominant (Fig. 9). This fact renders their discrimination less difficult. The reflectance in each spectral band is represented by up to 256 hues. Hence, the 542 colour composite image contains hues, that is possible combinations. Not all the combinations represent a different type of formation, and many are hardly distinguishable for the human eye. Thus, grouping the pixels with similar colours into distinguishable classes can reduce this huge quantity of information. The processing of field observations and relevees allowed the selection of certain areas for obtaining the spectral signatures for water, rocks, buildings, partly eroded areas, grasslands, deciduous forests and coniferous forests. The standard areas were used as criteria for the reclassification of the pixels of the colour composite image 542 through the minimum distance to mean algorithm [6]. This process led to a new image (Fig. 10) that presents each class in a distinct colour. The reclassification result reveals the localisation of the analysed formations and the proportions of the classes in the studied area (Fig. 11). The proportions resulted from the multiplication of the number of pixels in the image and the value of 900m 2 (each pixel has an area of 30m x 30m which is the resolution of the satellite sensor). The studied area estimated by teledetection is 144,3726 km 2. This value is similar to the area estimation based on a topographic map. The water covers 16,83% of the studied area. Izvoru Muntelui-Bicaz Accumulation Lake gives the most of this value. The proportion of the buildings is 3, 24% of the studied area. The areas covered with rocks (severe eroded areas, mountain stream shores etc.) represent 4,67% of the total. Partly eroded areas hold a high percentage (17%). Altered grasslands represent most of them. The meadows in a good status occupy 20,16% of the studied area. Forests are the best represented 5,66% coniferous forests and 32,45% deciduous forests. Conclusions Deciduous forest is the best-represented formation of the studied area. The high percentage of the partly eroded areas indicates an important human impact, especially over grasslands. This fact would endorse at least a land usage monitoring. Indirectly, the location of the eroded areas in grasslands outlines the role of the forests in the land protection against erosion. On a long-term scale, this phenomenon may 134

5 have a harmful effect over life and energetic use of the reservoir through the increase of sediment content in water and, consequently, through the slow decrease. Teledetection allows the localisation of land formations and the assessment of their status and quality. BIBLIOGRAPHY 1. ABER J. S Landsat Remote Sensing /academic.emporia.edu/ aberjame/ remote/ landsat/ landsat.htm; 2. BURDUJA C., GAVRILESCU GH Studiul floristic şi fitocenologic al spaţiului din jurul lacului de acumulare Bicaz I. Cercetări floristice asupra versantului stâng, între Dealul Gicovanu şi Piciorul Malu (Hangu), Lucr. Staţ. de Cerc. Biol. Stejarul, Piatra Neamţ, 1970: BURDUJA C., GAVRILESCU GH Studiul floristic şi fitocenologic al spaţiului din jurul lacului de acumulare Bicaz II. Cercetări floristice asupra versantului stâng, între Piciorul Malu (Hangu) şi Gura Largu (Poiana Teiului), Lucr. Staţ. de Cerc. Biol. Stejarul, Piatra Neamţ, 1976: CHIFU T., MITITELU D., DASCALESCU D., Flora şi vegetaţia judeţului Neamţ, Mem. Sect. şt. Acad. Rom., X, Nr. 1 (1987), Bucureşti 5. CHIFU T., ZAMFIRESCU OANA O nouă contribuţie la sintaxonomia pădurilor din clasa Querco- Fagetea Br.-Bl. et Vlieger in Vlieger 1937 de pe teritoriul Moldovei (România), Bul. Grăd. Bot., Iaşi, 10: LORUP E. J. ***. Idrisi Tutorial on WWW 7. SHORT N. M. Sr. ***. Remote Sensing Tutorial 8. ZAMFIRESCU OANA O nouă staţiune pentru specia Menyanthes trifoliata L. (fam. Menyanthaceae). Bul. Grăd. Bot., Iaşi, 10: ZAMFIRESCU OANA Analiza bioformelor florei vasculare din ecosistemelor naturale de pe malul stâng al lacului de acumulare Izvorul Muntelui-Bicaz, Bul. Grăd. Bot. Iaşi, 11: ZAMFIRESCU OANA Analysis of vascular flora geoelements from the left side of Izvorul Muntelui- Bicaz acumulation lake, Anal. Şt. Univ. Al. I. Cuza, Iaşi, XLVIII, s II-a, Biol. Veget.: ZAMFIRESCU OANA, CHIFU T Contribution to the study of the forest vegetation from the left side of Izvorul Muntelui-Bicaz acumulation lake, Anal. Şt. Univ. Al. I. Cuza, Iaşi, XLVIII, s. II-a, Biol. Veget.: ZAMFIRESCU OANA, CHIFU T., SÂRBU I Studiul fitocenologic al pajiştilor naturale de pe versantul stâng al lacului de acumulare Izvorul Muntelui-Bicaz (under press) 13. *** Basic Introduction in Geospatial Data *** MultiSpec Tutorial 135

6 Fig. 1. True colour composite image of Izvoru Muntelui-Bicaz left side (R=TM3, G=TM2, B=TM1) 136

7 Fig. 2. False colour composite image of Izvoru Muntelui-Bicaz left side (R=TM5, G=TM4, B=TM2) 137

8 Fig. 3. Water reflectance Fig. 4. Buildings reflectance Fig. 5. Partly eroded area reflectance Fig. 6. Rocks reflectance Fig. 7. Grasslands reflectance Fig. 8. Deciduous forests reflectance 138

9 Fig. 9. Coniferous forests reflectance Fig. 10. The map of the land formations from the left side of Izvoru Muntelui-Bicaz Reservoir (542 image pixels reclassification through minimum distance to mean algorithm) 139

10 35% 32,45% 30% 25% 20% 16,83% 17,00% 20,16% 15% 10% 5% 3,24% 5,66% 4,67% 0% water buildings coniferous forests erosions deciduous forest grasslands rocks Fig. 11. The percentage of the areas of the formations from the Izvoru Muntelui-Bicaz reservoir left side (total teledetection estimated area = m 2 ) 140